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 Types.
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 Types
820 @cindex decimal floating types
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 As an extension, the GNU C compiler supports decimal floating types as
832 defined in the N1176 draft of ISO/IEC WDTR24732. Support for decimal
833 floating types in GCC will evolve as the draft technical report changes.
834 Calling conventions for any target might also change. Not all targets
835 support decimal floating types.
837 The decimal floating types are @code{_Decimal32}, @code{_Decimal64}, and
838 @code{_Decimal128}. They use a radix of ten, unlike the floating types
839 @code{float}, @code{double}, and @code{long double} whose radix is not
840 specified by the C standard but is usually two.
842 Support for decimal floating types includes the arithmetic operators
843 add, subtract, multiply, divide; unary arithmetic operators;
844 relational operators; equality operators; and conversions to and from
845 integer and other floating types. Use a suffix @samp{df} or
846 @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd}
847 or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for
850 GCC support of decimal float as specified by the draft technical report
855 Translation time data type (TTDT) is not supported.
858 Characteristics of decimal floating types are defined in header file
859 @file{decfloat.h} rather than @file{float.h}.
862 When the value of a decimal floating type cannot be represented in the
863 integer type to which it is being converted, the result is undefined
864 rather than the result value specified by the draft technical report.
867 Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128}
868 are supported by the DWARF2 debug information format.
874 ISO C99 supports floating-point numbers written not only in the usual
875 decimal notation, such as @code{1.55e1}, but also numbers such as
876 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
877 supports this in C89 mode (except in some cases when strictly
878 conforming) and in C++. In that format the
879 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
880 mandatory. The exponent is a decimal number that indicates the power of
881 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
888 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
889 is the same as @code{1.55e1}.
891 Unlike for floating-point numbers in the decimal notation the exponent
892 is always required in the hexadecimal notation. Otherwise the compiler
893 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
894 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
895 extension for floating-point constants of type @code{float}.
898 @section Arrays of Length Zero
899 @cindex arrays of length zero
900 @cindex zero-length arrays
901 @cindex length-zero arrays
902 @cindex flexible array members
904 Zero-length arrays are allowed in GNU C@. They are very useful as the
905 last element of a structure which is really a header for a variable-length
914 struct line *thisline = (struct line *)
915 malloc (sizeof (struct line) + this_length);
916 thisline->length = this_length;
919 In ISO C90, you would have to give @code{contents} a length of 1, which
920 means either you waste space or complicate the argument to @code{malloc}.
922 In ISO C99, you would use a @dfn{flexible array member}, which is
923 slightly different in syntax and semantics:
927 Flexible array members are written as @code{contents[]} without
931 Flexible array members have incomplete type, and so the @code{sizeof}
932 operator may not be applied. As a quirk of the original implementation
933 of zero-length arrays, @code{sizeof} evaluates to zero.
936 Flexible array members may only appear as the last member of a
937 @code{struct} that is otherwise non-empty.
940 A structure containing a flexible array member, or a union containing
941 such a structure (possibly recursively), may not be a member of a
942 structure or an element of an array. (However, these uses are
943 permitted by GCC as extensions.)
946 GCC versions before 3.0 allowed zero-length arrays to be statically
947 initialized, as if they were flexible arrays. In addition to those
948 cases that were useful, it also allowed initializations in situations
949 that would corrupt later data. Non-empty initialization of zero-length
950 arrays is now treated like any case where there are more initializer
951 elements than the array holds, in that a suitable warning about "excess
952 elements in array" is given, and the excess elements (all of them, in
953 this case) are ignored.
955 Instead GCC allows static initialization of flexible array members.
956 This is equivalent to defining a new structure containing the original
957 structure followed by an array of sufficient size to contain the data.
958 I.e.@: in the following, @code{f1} is constructed as if it were declared
964 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
967 struct f1 f1; int data[3];
968 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
972 The convenience of this extension is that @code{f1} has the desired
973 type, eliminating the need to consistently refer to @code{f2.f1}.
975 This has symmetry with normal static arrays, in that an array of
976 unknown size is also written with @code{[]}.
978 Of course, this extension only makes sense if the extra data comes at
979 the end of a top-level object, as otherwise we would be overwriting
980 data at subsequent offsets. To avoid undue complication and confusion
981 with initialization of deeply nested arrays, we simply disallow any
982 non-empty initialization except when the structure is the top-level
986 struct foo @{ int x; int y[]; @};
987 struct bar @{ struct foo z; @};
989 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
990 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
991 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
992 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
995 @node Empty Structures
996 @section Structures With No Members
997 @cindex empty structures
998 @cindex zero-size structures
1000 GCC permits a C structure to have no members:
1007 The structure will have size zero. In C++, empty structures are part
1008 of the language. G++ treats empty structures as if they had a single
1009 member of type @code{char}.
1011 @node Variable Length
1012 @section Arrays of Variable Length
1013 @cindex variable-length arrays
1014 @cindex arrays of variable length
1017 Variable-length automatic arrays are allowed in ISO C99, and as an
1018 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1019 implementation of variable-length arrays does not yet conform in detail
1020 to the ISO C99 standard.) These arrays are
1021 declared like any other automatic arrays, but with a length that is not
1022 a constant expression. The storage is allocated at the point of
1023 declaration and deallocated when the brace-level is exited. For
1028 concat_fopen (char *s1, char *s2, char *mode)
1030 char str[strlen (s1) + strlen (s2) + 1];
1033 return fopen (str, mode);
1037 @cindex scope of a variable length array
1038 @cindex variable-length array scope
1039 @cindex deallocating variable length arrays
1040 Jumping or breaking out of the scope of the array name deallocates the
1041 storage. Jumping into the scope is not allowed; you get an error
1044 @cindex @code{alloca} vs variable-length arrays
1045 You can use the function @code{alloca} to get an effect much like
1046 variable-length arrays. The function @code{alloca} is available in
1047 many other C implementations (but not in all). On the other hand,
1048 variable-length arrays are more elegant.
1050 There are other differences between these two methods. Space allocated
1051 with @code{alloca} exists until the containing @emph{function} returns.
1052 The space for a variable-length array is deallocated as soon as the array
1053 name's scope ends. (If you use both variable-length arrays and
1054 @code{alloca} in the same function, deallocation of a variable-length array
1055 will also deallocate anything more recently allocated with @code{alloca}.)
1057 You can also use variable-length arrays as arguments to functions:
1061 tester (int len, char data[len][len])
1067 The length of an array is computed once when the storage is allocated
1068 and is remembered for the scope of the array in case you access it with
1071 If you want to pass the array first and the length afterward, you can
1072 use a forward declaration in the parameter list---another GNU extension.
1076 tester (int len; char data[len][len], int len)
1082 @cindex parameter forward declaration
1083 The @samp{int len} before the semicolon is a @dfn{parameter forward
1084 declaration}, and it serves the purpose of making the name @code{len}
1085 known when the declaration of @code{data} is parsed.
1087 You can write any number of such parameter forward declarations in the
1088 parameter list. They can be separated by commas or semicolons, but the
1089 last one must end with a semicolon, which is followed by the ``real''
1090 parameter declarations. Each forward declaration must match a ``real''
1091 declaration in parameter name and data type. ISO C99 does not support
1092 parameter forward declarations.
1094 @node Variadic Macros
1095 @section Macros with a Variable Number of Arguments.
1096 @cindex variable number of arguments
1097 @cindex macro with variable arguments
1098 @cindex rest argument (in macro)
1099 @cindex variadic macros
1101 In the ISO C standard of 1999, a macro can be declared to accept a
1102 variable number of arguments much as a function can. The syntax for
1103 defining the macro is similar to that of a function. Here is an
1107 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1110 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1111 such a macro, it represents the zero or more tokens until the closing
1112 parenthesis that ends the invocation, including any commas. This set of
1113 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1114 wherever it appears. See the CPP manual for more information.
1116 GCC has long supported variadic macros, and used a different syntax that
1117 allowed you to give a name to the variable arguments just like any other
1118 argument. Here is an example:
1121 #define debug(format, args...) fprintf (stderr, format, args)
1124 This is in all ways equivalent to the ISO C example above, but arguably
1125 more readable and descriptive.
1127 GNU CPP has two further variadic macro extensions, and permits them to
1128 be used with either of the above forms of macro definition.
1130 In standard C, you are not allowed to leave the variable argument out
1131 entirely; but you are allowed to pass an empty argument. For example,
1132 this invocation is invalid in ISO C, because there is no comma after
1139 GNU CPP permits you to completely omit the variable arguments in this
1140 way. In the above examples, the compiler would complain, though since
1141 the expansion of the macro still has the extra comma after the format
1144 To help solve this problem, CPP behaves specially for variable arguments
1145 used with the token paste operator, @samp{##}. If instead you write
1148 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1151 and if the variable arguments are omitted or empty, the @samp{##}
1152 operator causes the preprocessor to remove the comma before it. If you
1153 do provide some variable arguments in your macro invocation, GNU CPP
1154 does not complain about the paste operation and instead places the
1155 variable arguments after the comma. Just like any other pasted macro
1156 argument, these arguments are not macro expanded.
1158 @node Escaped Newlines
1159 @section Slightly Looser Rules for Escaped Newlines
1160 @cindex escaped newlines
1161 @cindex newlines (escaped)
1163 Recently, the preprocessor has relaxed its treatment of escaped
1164 newlines. Previously, the newline had to immediately follow a
1165 backslash. The current implementation allows whitespace in the form
1166 of spaces, horizontal and vertical tabs, and form feeds between the
1167 backslash and the subsequent newline. The preprocessor issues a
1168 warning, but treats it as a valid escaped newline and combines the two
1169 lines to form a single logical line. This works within comments and
1170 tokens, as well as between tokens. Comments are @emph{not} treated as
1171 whitespace for the purposes of this relaxation, since they have not
1172 yet been replaced with spaces.
1175 @section Non-Lvalue Arrays May Have Subscripts
1176 @cindex subscripting
1177 @cindex arrays, non-lvalue
1179 @cindex subscripting and function values
1180 In ISO C99, arrays that are not lvalues still decay to pointers, and
1181 may be subscripted, although they may not be modified or used after
1182 the next sequence point and the unary @samp{&} operator may not be
1183 applied to them. As an extension, GCC allows such arrays to be
1184 subscripted in C89 mode, though otherwise they do not decay to
1185 pointers outside C99 mode. For example,
1186 this is valid in GNU C though not valid in C89:
1190 struct foo @{int a[4];@};
1196 return f().a[index];
1202 @section Arithmetic on @code{void}- and Function-Pointers
1203 @cindex void pointers, arithmetic
1204 @cindex void, size of pointer to
1205 @cindex function pointers, arithmetic
1206 @cindex function, size of pointer to
1208 In GNU C, addition and subtraction operations are supported on pointers to
1209 @code{void} and on pointers to functions. This is done by treating the
1210 size of a @code{void} or of a function as 1.
1212 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1213 and on function types, and returns 1.
1215 @opindex Wpointer-arith
1216 The option @option{-Wpointer-arith} requests a warning if these extensions
1220 @section Non-Constant Initializers
1221 @cindex initializers, non-constant
1222 @cindex non-constant initializers
1224 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1225 automatic variable are not required to be constant expressions in GNU C@.
1226 Here is an example of an initializer with run-time varying elements:
1229 foo (float f, float g)
1231 float beat_freqs[2] = @{ f-g, f+g @};
1236 @node Compound Literals
1237 @section Compound Literals
1238 @cindex constructor expressions
1239 @cindex initializations in expressions
1240 @cindex structures, constructor expression
1241 @cindex expressions, constructor
1242 @cindex compound literals
1243 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1245 ISO C99 supports compound literals. A compound literal looks like
1246 a cast containing an initializer. Its value is an object of the
1247 type specified in the cast, containing the elements specified in
1248 the initializer; it is an lvalue. As an extension, GCC supports
1249 compound literals in C89 mode and in C++.
1251 Usually, the specified type is a structure. Assume that
1252 @code{struct foo} and @code{structure} are declared as shown:
1255 struct foo @{int a; char b[2];@} structure;
1259 Here is an example of constructing a @code{struct foo} with a compound literal:
1262 structure = ((struct foo) @{x + y, 'a', 0@});
1266 This is equivalent to writing the following:
1270 struct foo temp = @{x + y, 'a', 0@};
1275 You can also construct an array. If all the elements of the compound literal
1276 are (made up of) simple constant expressions, suitable for use in
1277 initializers of objects of static storage duration, then the compound
1278 literal can be coerced to a pointer to its first element and used in
1279 such an initializer, as shown here:
1282 char **foo = (char *[]) @{ "x", "y", "z" @};
1285 Compound literals for scalar types and union types are is
1286 also allowed, but then the compound literal is equivalent
1289 As a GNU extension, GCC allows initialization of objects with static storage
1290 duration by compound literals (which is not possible in ISO C99, because
1291 the initializer is not a constant).
1292 It is handled as if the object was initialized only with the bracket
1293 enclosed list if compound literal's and object types match.
1294 The initializer list of the compound literal must be constant.
1295 If the object being initialized has array type of unknown size, the size is
1296 determined by compound literal size.
1299 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1300 static int y[] = (int []) @{1, 2, 3@};
1301 static int z[] = (int [3]) @{1@};
1305 The above lines are equivalent to the following:
1307 static struct foo x = @{1, 'a', 'b'@};
1308 static int y[] = @{1, 2, 3@};
1309 static int z[] = @{1, 0, 0@};
1312 @node Designated Inits
1313 @section Designated Initializers
1314 @cindex initializers with labeled elements
1315 @cindex labeled elements in initializers
1316 @cindex case labels in initializers
1317 @cindex designated initializers
1319 Standard C89 requires the elements of an initializer to appear in a fixed
1320 order, the same as the order of the elements in the array or structure
1323 In ISO C99 you can give the elements in any order, specifying the array
1324 indices or structure field names they apply to, and GNU C allows this as
1325 an extension in C89 mode as well. This extension is not
1326 implemented in GNU C++.
1328 To specify an array index, write
1329 @samp{[@var{index}] =} before the element value. For example,
1332 int a[6] = @{ [4] = 29, [2] = 15 @};
1339 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1343 The index values must be constant expressions, even if the array being
1344 initialized is automatic.
1346 An alternative syntax for this which has been obsolete since GCC 2.5 but
1347 GCC still accepts is to write @samp{[@var{index}]} before the element
1348 value, with no @samp{=}.
1350 To initialize a range of elements to the same value, write
1351 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1352 extension. For example,
1355 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1359 If the value in it has side-effects, the side-effects will happen only once,
1360 not for each initialized field by the range initializer.
1363 Note that the length of the array is the highest value specified
1366 In a structure initializer, specify the name of a field to initialize
1367 with @samp{.@var{fieldname} =} before the element value. For example,
1368 given the following structure,
1371 struct point @{ int x, y; @};
1375 the following initialization
1378 struct point p = @{ .y = yvalue, .x = xvalue @};
1385 struct point p = @{ xvalue, yvalue @};
1388 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1389 @samp{@var{fieldname}:}, as shown here:
1392 struct point p = @{ y: yvalue, x: xvalue @};
1396 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1397 @dfn{designator}. You can also use a designator (or the obsolete colon
1398 syntax) when initializing a union, to specify which element of the union
1399 should be used. For example,
1402 union foo @{ int i; double d; @};
1404 union foo f = @{ .d = 4 @};
1408 will convert 4 to a @code{double} to store it in the union using
1409 the second element. By contrast, casting 4 to type @code{union foo}
1410 would store it into the union as the integer @code{i}, since it is
1411 an integer. (@xref{Cast to Union}.)
1413 You can combine this technique of naming elements with ordinary C
1414 initialization of successive elements. Each initializer element that
1415 does not have a designator applies to the next consecutive element of the
1416 array or structure. For example,
1419 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1426 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1429 Labeling the elements of an array initializer is especially useful
1430 when the indices are characters or belong to an @code{enum} type.
1435 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1436 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1439 @cindex designator lists
1440 You can also write a series of @samp{.@var{fieldname}} and
1441 @samp{[@var{index}]} designators before an @samp{=} to specify a
1442 nested subobject to initialize; the list is taken relative to the
1443 subobject corresponding to the closest surrounding brace pair. For
1444 example, with the @samp{struct point} declaration above:
1447 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1451 If the same field is initialized multiple times, it will have value from
1452 the last initialization. If any such overridden initialization has
1453 side-effect, it is unspecified whether the side-effect happens or not.
1454 Currently, GCC will discard them and issue a warning.
1457 @section Case Ranges
1459 @cindex ranges in case statements
1461 You can specify a range of consecutive values in a single @code{case} label,
1465 case @var{low} ... @var{high}:
1469 This has the same effect as the proper number of individual @code{case}
1470 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1472 This feature is especially useful for ranges of ASCII character codes:
1478 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1479 it may be parsed wrong when you use it with integer values. For example,
1494 @section Cast to a Union Type
1495 @cindex cast to a union
1496 @cindex union, casting to a
1498 A cast to union type is similar to other casts, except that the type
1499 specified is a union type. You can specify the type either with
1500 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1501 a constructor though, not a cast, and hence does not yield an lvalue like
1502 normal casts. (@xref{Compound Literals}.)
1504 The types that may be cast to the union type are those of the members
1505 of the union. Thus, given the following union and variables:
1508 union foo @{ int i; double d; @};
1514 both @code{x} and @code{y} can be cast to type @code{union foo}.
1516 Using the cast as the right-hand side of an assignment to a variable of
1517 union type is equivalent to storing in a member of the union:
1522 u = (union foo) x @equiv{} u.i = x
1523 u = (union foo) y @equiv{} u.d = y
1526 You can also use the union cast as a function argument:
1529 void hack (union foo);
1531 hack ((union foo) x);
1534 @node Mixed Declarations
1535 @section Mixed Declarations and Code
1536 @cindex mixed declarations and code
1537 @cindex declarations, mixed with code
1538 @cindex code, mixed with declarations
1540 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1541 within compound statements. As an extension, GCC also allows this in
1542 C89 mode. For example, you could do:
1551 Each identifier is visible from where it is declared until the end of
1552 the enclosing block.
1554 @node Function Attributes
1555 @section Declaring Attributes of Functions
1556 @cindex function attributes
1557 @cindex declaring attributes of functions
1558 @cindex functions that never return
1559 @cindex functions that return more than once
1560 @cindex functions that have no side effects
1561 @cindex functions in arbitrary sections
1562 @cindex functions that behave like malloc
1563 @cindex @code{volatile} applied to function
1564 @cindex @code{const} applied to function
1565 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1566 @cindex functions with non-null pointer arguments
1567 @cindex functions that are passed arguments in registers on the 386
1568 @cindex functions that pop the argument stack on the 386
1569 @cindex functions that do not pop the argument stack on the 386
1571 In GNU C, you declare certain things about functions called in your program
1572 which help the compiler optimize function calls and check your code more
1575 The keyword @code{__attribute__} allows you to specify special
1576 attributes when making a declaration. This keyword is followed by an
1577 attribute specification inside double parentheses. The following
1578 attributes are currently defined for functions on all targets:
1579 @code{noreturn}, @code{returns_twice}, @code{noinline}, @code{always_inline},
1580 @code{flatten}, @code{pure}, @code{const}, @code{nothrow}, @code{sentinel},
1581 @code{format}, @code{format_arg}, @code{no_instrument_function},
1582 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1583 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
1584 @code{alias}, @code{warn_unused_result}, @code{nonnull},
1585 @code{gnu_inline} and @code{externally_visible}. Several other
1586 attributes are defined for functions on particular target systems. Other
1587 attributes, including @code{section} are supported for variables declarations
1588 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1590 You may also specify attributes with @samp{__} preceding and following
1591 each keyword. This allows you to use them in header files without
1592 being concerned about a possible macro of the same name. For example,
1593 you may use @code{__noreturn__} instead of @code{noreturn}.
1595 @xref{Attribute Syntax}, for details of the exact syntax for using
1599 @c Keep this table alphabetized by attribute name. Treat _ as space.
1601 @item alias ("@var{target}")
1602 @cindex @code{alias} attribute
1603 The @code{alias} attribute causes the declaration to be emitted as an
1604 alias for another symbol, which must be specified. For instance,
1607 void __f () @{ /* @r{Do something.} */; @}
1608 void f () __attribute__ ((weak, alias ("__f")));
1611 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
1612 mangled name for the target must be used. It is an error if @samp{__f}
1613 is not defined in the same translation unit.
1615 Not all target machines support this attribute.
1618 @cindex @code{always_inline} function attribute
1619 Generally, functions are not inlined unless optimization is specified.
1620 For functions declared inline, this attribute inlines the function even
1621 if no optimization level was specified.
1624 @cindex @code{gnu_inline} function attribute
1625 This attribute on an inline declaration results in the old GNU C89
1626 inline behavior even in the ISO C99 mode.
1628 @cindex @code{flatten} function attribute
1630 Generally, inlining into a function is limited. For a function marked with
1631 this attribute, every call inside this function will be inlined, if possible.
1632 Whether the function itself is considered for inlining depends on its size and
1633 the current inlining parameters. The @code{flatten} attribute only works
1634 reliably in unit-at-a-time mode.
1637 @cindex functions that do pop the argument stack on the 386
1639 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1640 assume that the calling function will pop off the stack space used to
1641 pass arguments. This is
1642 useful to override the effects of the @option{-mrtd} switch.
1645 @cindex @code{const} function attribute
1646 Many functions do not examine any values except their arguments, and
1647 have no effects except the return value. Basically this is just slightly
1648 more strict class than the @code{pure} attribute below, since function is not
1649 allowed to read global memory.
1651 @cindex pointer arguments
1652 Note that a function that has pointer arguments and examines the data
1653 pointed to must @emph{not} be declared @code{const}. Likewise, a
1654 function that calls a non-@code{const} function usually must not be
1655 @code{const}. It does not make sense for a @code{const} function to
1658 The attribute @code{const} is not implemented in GCC versions earlier
1659 than 2.5. An alternative way to declare that a function has no side
1660 effects, which works in the current version and in some older versions,
1664 typedef int intfn ();
1666 extern const intfn square;
1669 This approach does not work in GNU C++ from 2.6.0 on, since the language
1670 specifies that the @samp{const} must be attached to the return value.
1674 @cindex @code{constructor} function attribute
1675 @cindex @code{destructor} function attribute
1676 The @code{constructor} attribute causes the function to be called
1677 automatically before execution enters @code{main ()}. Similarly, the
1678 @code{destructor} attribute causes the function to be called
1679 automatically after @code{main ()} has completed or @code{exit ()} has
1680 been called. Functions with these attributes are useful for
1681 initializing data that will be used implicitly during the execution of
1684 These attributes are not currently implemented for Objective-C@.
1687 @cindex @code{deprecated} attribute.
1688 The @code{deprecated} attribute results in a warning if the function
1689 is used anywhere in the source file. This is useful when identifying
1690 functions that are expected to be removed in a future version of a
1691 program. The warning also includes the location of the declaration
1692 of the deprecated function, to enable users to easily find further
1693 information about why the function is deprecated, or what they should
1694 do instead. Note that the warnings only occurs for uses:
1697 int old_fn () __attribute__ ((deprecated));
1699 int (*fn_ptr)() = old_fn;
1702 results in a warning on line 3 but not line 2.
1704 The @code{deprecated} attribute can also be used for variables and
1705 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
1708 @cindex @code{__declspec(dllexport)}
1709 On Microsoft Windows targets and Symbian OS targets the
1710 @code{dllexport} attribute causes the compiler to provide a global
1711 pointer to a pointer in a DLL, so that it can be referenced with the
1712 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
1713 name is formed by combining @code{_imp__} and the function or variable
1716 You can use @code{__declspec(dllexport)} as a synonym for
1717 @code{__attribute__ ((dllexport))} for compatibility with other
1720 On systems that support the @code{visibility} attribute, this
1721 attribute also implies ``default'' visibility, unless a
1722 @code{visibility} attribute is explicitly specified. You should avoid
1723 the use of @code{dllexport} with ``hidden'' or ``internal''
1724 visibility; in the future GCC may issue an error for those cases.
1726 Currently, the @code{dllexport} attribute is ignored for inlined
1727 functions, unless the @option{-fkeep-inline-functions} flag has been
1728 used. The attribute is also ignored for undefined symbols.
1730 When applied to C++ classes, the attribute marks defined non-inlined
1731 member functions and static data members as exports. Static consts
1732 initialized in-class are not marked unless they are also defined
1735 For Microsoft Windows targets there are alternative methods for
1736 including the symbol in the DLL's export table such as using a
1737 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
1738 the @option{--export-all} linker flag.
1741 @cindex @code{__declspec(dllimport)}
1742 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
1743 attribute causes the compiler to reference a function or variable via
1744 a global pointer to a pointer that is set up by the DLL exporting the
1745 symbol. The attribute implies @code{extern} storage. On Microsoft
1746 Windows targets, the pointer name is formed by combining @code{_imp__}
1747 and the function or variable name.
1749 You can use @code{__declspec(dllimport)} as a synonym for
1750 @code{__attribute__ ((dllimport))} for compatibility with other
1753 Currently, the attribute is ignored for inlined functions. If the
1754 attribute is applied to a symbol @emph{definition}, an error is reported.
1755 If a symbol previously declared @code{dllimport} is later defined, the
1756 attribute is ignored in subsequent references, and a warning is emitted.
1757 The attribute is also overridden by a subsequent declaration as
1760 When applied to C++ classes, the attribute marks non-inlined
1761 member functions and static data members as imports. However, the
1762 attribute is ignored for virtual methods to allow creation of vtables
1765 On the SH Symbian OS target the @code{dllimport} attribute also has
1766 another affect---it can cause the vtable and run-time type information
1767 for a class to be exported. This happens when the class has a
1768 dllimport'ed constructor or a non-inline, non-pure virtual function
1769 and, for either of those two conditions, the class also has a inline
1770 constructor or destructor and has a key function that is defined in
1771 the current translation unit.
1773 For Microsoft Windows based targets the use of the @code{dllimport}
1774 attribute on functions is not necessary, but provides a small
1775 performance benefit by eliminating a thunk in the DLL@. The use of the
1776 @code{dllimport} attribute on imported variables was required on older
1777 versions of the GNU linker, but can now be avoided by passing the
1778 @option{--enable-auto-import} switch to the GNU linker. As with
1779 functions, using the attribute for a variable eliminates a thunk in
1782 One drawback to using this attribute is that a pointer to a function
1783 or variable marked as @code{dllimport} cannot be used as a constant
1784 address. On Microsoft Windows targets, the attribute can be disabled
1785 for functions by setting the @option{-mnop-fun-dllimport} flag.
1788 @cindex eight bit data on the H8/300, H8/300H, and H8S
1789 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1790 variable should be placed into the eight bit data section.
1791 The compiler will generate more efficient code for certain operations
1792 on data in the eight bit data area. Note the eight bit data area is limited to
1795 You must use GAS and GLD from GNU binutils version 2.7 or later for
1796 this attribute to work correctly.
1798 @item exception_handler
1799 @cindex exception handler functions on the Blackfin processor
1800 Use this attribute on the Blackfin to indicate that the specified function
1801 is an exception handler. The compiler will generate function entry and
1802 exit sequences suitable for use in an exception handler when this
1803 attribute is present.
1806 @cindex functions which handle memory bank switching
1807 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
1808 use a calling convention that takes care of switching memory banks when
1809 entering and leaving a function. This calling convention is also the
1810 default when using the @option{-mlong-calls} option.
1812 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
1813 to call and return from a function.
1815 On 68HC11 the compiler will generate a sequence of instructions
1816 to invoke a board-specific routine to switch the memory bank and call the
1817 real function. The board-specific routine simulates a @code{call}.
1818 At the end of a function, it will jump to a board-specific routine
1819 instead of using @code{rts}. The board-specific return routine simulates
1823 @cindex functions that pop the argument stack on the 386
1824 On the Intel 386, the @code{fastcall} attribute causes the compiler to
1825 pass the first argument (if of integral type) in the register ECX and
1826 the second argument (if of integral type) in the register EDX@. Subsequent
1827 and other typed arguments are passed on the stack. The called function will
1828 pop the arguments off the stack. If the number of arguments is variable all
1829 arguments are pushed on the stack.
1831 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
1832 @cindex @code{format} function attribute
1834 The @code{format} attribute specifies that a function takes @code{printf},
1835 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
1836 should be type-checked against a format string. For example, the
1841 my_printf (void *my_object, const char *my_format, ...)
1842 __attribute__ ((format (printf, 2, 3)));
1846 causes the compiler to check the arguments in calls to @code{my_printf}
1847 for consistency with the @code{printf} style format string argument
1850 The parameter @var{archetype} determines how the format string is
1851 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
1852 or @code{strfmon}. (You can also use @code{__printf__},
1853 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
1854 parameter @var{string-index} specifies which argument is the format
1855 string argument (starting from 1), while @var{first-to-check} is the
1856 number of the first argument to check against the format string. For
1857 functions where the arguments are not available to be checked (such as
1858 @code{vprintf}), specify the third parameter as zero. In this case the
1859 compiler only checks the format string for consistency. For
1860 @code{strftime} formats, the third parameter is required to be zero.
1861 Since non-static C++ methods have an implicit @code{this} argument, the
1862 arguments of such methods should be counted from two, not one, when
1863 giving values for @var{string-index} and @var{first-to-check}.
1865 In the example above, the format string (@code{my_format}) is the second
1866 argument of the function @code{my_print}, and the arguments to check
1867 start with the third argument, so the correct parameters for the format
1868 attribute are 2 and 3.
1870 @opindex ffreestanding
1871 @opindex fno-builtin
1872 The @code{format} attribute allows you to identify your own functions
1873 which take format strings as arguments, so that GCC can check the
1874 calls to these functions for errors. The compiler always (unless
1875 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
1876 for the standard library functions @code{printf}, @code{fprintf},
1877 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
1878 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
1879 warnings are requested (using @option{-Wformat}), so there is no need to
1880 modify the header file @file{stdio.h}. In C99 mode, the functions
1881 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
1882 @code{vsscanf} are also checked. Except in strictly conforming C
1883 standard modes, the X/Open function @code{strfmon} is also checked as
1884 are @code{printf_unlocked} and @code{fprintf_unlocked}.
1885 @xref{C Dialect Options,,Options Controlling C Dialect}.
1887 The target may provide additional types of format checks.
1888 @xref{Target Format Checks,,Format Checks Specific to Particular
1891 @item format_arg (@var{string-index})
1892 @cindex @code{format_arg} function attribute
1893 @opindex Wformat-nonliteral
1894 The @code{format_arg} attribute specifies that a function takes a format
1895 string for a @code{printf}, @code{scanf}, @code{strftime} or
1896 @code{strfmon} style function and modifies it (for example, to translate
1897 it into another language), so the result can be passed to a
1898 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
1899 function (with the remaining arguments to the format function the same
1900 as they would have been for the unmodified string). For example, the
1905 my_dgettext (char *my_domain, const char *my_format)
1906 __attribute__ ((format_arg (2)));
1910 causes the compiler to check the arguments in calls to a @code{printf},
1911 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
1912 format string argument is a call to the @code{my_dgettext} function, for
1913 consistency with the format string argument @code{my_format}. If the
1914 @code{format_arg} attribute had not been specified, all the compiler
1915 could tell in such calls to format functions would be that the format
1916 string argument is not constant; this would generate a warning when
1917 @option{-Wformat-nonliteral} is used, but the calls could not be checked
1918 without the attribute.
1920 The parameter @var{string-index} specifies which argument is the format
1921 string argument (starting from one). Since non-static C++ methods have
1922 an implicit @code{this} argument, the arguments of such methods should
1923 be counted from two.
1925 The @code{format-arg} attribute allows you to identify your own
1926 functions which modify format strings, so that GCC can check the
1927 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
1928 type function whose operands are a call to one of your own function.
1929 The compiler always treats @code{gettext}, @code{dgettext}, and
1930 @code{dcgettext} in this manner except when strict ISO C support is
1931 requested by @option{-ansi} or an appropriate @option{-std} option, or
1932 @option{-ffreestanding} or @option{-fno-builtin}
1933 is used. @xref{C Dialect Options,,Options
1934 Controlling C Dialect}.
1936 @item function_vector
1937 @cindex calling functions through the function vector on the H8/300 processors
1938 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1939 function should be called through the function vector. Calling a
1940 function through the function vector will reduce code size, however;
1941 the function vector has a limited size (maximum 128 entries on the H8/300
1942 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
1944 You must use GAS and GLD from GNU binutils version 2.7 or later for
1945 this attribute to work correctly.
1948 @cindex interrupt handler functions
1949 Use this attribute on the ARM, AVR, C4x, CRX, M32C, M32R/D, MS1, and Xstormy16
1950 ports to indicate that the specified function is an interrupt handler.
1951 The compiler will generate function entry and exit sequences suitable
1952 for use in an interrupt handler when this attribute is present.
1954 Note, interrupt handlers for the Blackfin, m68k, H8/300, H8/300H, H8S, and
1955 SH processors can be specified via the @code{interrupt_handler} attribute.
1957 Note, on the AVR, interrupts will be enabled inside the function.
1959 Note, for the ARM, you can specify the kind of interrupt to be handled by
1960 adding an optional parameter to the interrupt attribute like this:
1963 void f () __attribute__ ((interrupt ("IRQ")));
1966 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
1968 @item interrupt_handler
1969 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
1970 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
1971 indicate that the specified function is an interrupt handler. The compiler
1972 will generate function entry and exit sequences suitable for use in an
1973 interrupt handler when this attribute is present.
1976 @cindex User stack pointer in interrupts on the Blackfin
1977 When used together with @code{interrupt_handler}, @code{exception_handler}
1978 or @code{nmi_handler}, code will be generated to load the stack pointer
1979 from the USP register in the function prologue.
1981 @item long_call/short_call
1982 @cindex indirect calls on ARM
1983 This attribute specifies how a particular function is called on
1984 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
1985 command line switch and @code{#pragma long_calls} settings. The
1986 @code{long_call} attribute indicates that the function might be far
1987 away from the call site and require a different (more expensive)
1988 calling sequence. The @code{short_call} attribute always places
1989 the offset to the function from the call site into the @samp{BL}
1990 instruction directly.
1992 @item longcall/shortcall
1993 @cindex functions called via pointer on the RS/6000 and PowerPC
1994 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
1995 indicates that the function might be far away from the call site and
1996 require a different (more expensive) calling sequence. The
1997 @code{shortcall} attribute indicates that the function is always close
1998 enough for the shorter calling sequence to be used. These attributes
1999 override both the @option{-mlongcall} switch and, on the RS/6000 and
2000 PowerPC, the @code{#pragma longcall} setting.
2002 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2003 calls are necessary.
2006 @cindex indirect calls on MIPS
2007 This attribute specifies how a particular function is called on MIPS@.
2008 The attribute overrides the @option{-mlong-calls} (@pxref{MIPS Options})
2009 command line switch. This attribute causes the compiler to always call
2010 the function by first loading its address into a register, and then using
2011 the contents of that register.
2014 @cindex @code{malloc} attribute
2015 The @code{malloc} attribute is used to tell the compiler that a function
2016 may be treated as if any non-@code{NULL} pointer it returns cannot
2017 alias any other pointer valid when the function returns.
2018 This will often improve optimization.
2019 Standard functions with this property include @code{malloc} and
2020 @code{calloc}. @code{realloc}-like functions have this property as
2021 long as the old pointer is never referred to (including comparing it
2022 to the new pointer) after the function returns a non-@code{NULL}
2025 @item model (@var{model-name})
2026 @cindex function addressability on the M32R/D
2027 @cindex variable addressability on the IA-64
2029 On the M32R/D, use this attribute to set the addressability of an
2030 object, and of the code generated for a function. The identifier
2031 @var{model-name} is one of @code{small}, @code{medium}, or
2032 @code{large}, representing each of the code models.
2034 Small model objects live in the lower 16MB of memory (so that their
2035 addresses can be loaded with the @code{ld24} instruction), and are
2036 callable with the @code{bl} instruction.
2038 Medium model objects may live anywhere in the 32-bit address space (the
2039 compiler will generate @code{seth/add3} instructions to load their addresses),
2040 and are callable with the @code{bl} instruction.
2042 Large model objects may live anywhere in the 32-bit address space (the
2043 compiler will generate @code{seth/add3} instructions to load their addresses),
2044 and may not be reachable with the @code{bl} instruction (the compiler will
2045 generate the much slower @code{seth/add3/jl} instruction sequence).
2047 On IA-64, use this attribute to set the addressability of an object.
2048 At present, the only supported identifier for @var{model-name} is
2049 @code{small}, indicating addressability via ``small'' (22-bit)
2050 addresses (so that their addresses can be loaded with the @code{addl}
2051 instruction). Caveat: such addressing is by definition not position
2052 independent and hence this attribute must not be used for objects
2053 defined by shared libraries.
2056 @cindex function without a prologue/epilogue code
2057 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
2058 specified function does not need prologue/epilogue sequences generated by
2059 the compiler. It is up to the programmer to provide these sequences.
2062 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2063 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2064 use the normal calling convention based on @code{jsr} and @code{rts}.
2065 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2069 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2070 Use this attribute together with @code{interrupt_handler},
2071 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2072 entry code should enable nested interrupts or exceptions.
2075 @cindex NMI handler functions on the Blackfin processor
2076 Use this attribute on the Blackfin to indicate that the specified function
2077 is an NMI handler. The compiler will generate function entry and
2078 exit sequences suitable for use in an NMI handler when this
2079 attribute is present.
2081 @item no_instrument_function
2082 @cindex @code{no_instrument_function} function attribute
2083 @opindex finstrument-functions
2084 If @option{-finstrument-functions} is given, profiling function calls will
2085 be generated at entry and exit of most user-compiled functions.
2086 Functions with this attribute will not be so instrumented.
2089 @cindex @code{noinline} function attribute
2090 This function attribute prevents a function from being considered for
2093 @item nonnull (@var{arg-index}, @dots{})
2094 @cindex @code{nonnull} function attribute
2095 The @code{nonnull} attribute specifies that some function parameters should
2096 be non-null pointers. For instance, the declaration:
2100 my_memcpy (void *dest, const void *src, size_t len)
2101 __attribute__((nonnull (1, 2)));
2105 causes the compiler to check that, in calls to @code{my_memcpy},
2106 arguments @var{dest} and @var{src} are non-null. If the compiler
2107 determines that a null pointer is passed in an argument slot marked
2108 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2109 is issued. The compiler may also choose to make optimizations based
2110 on the knowledge that certain function arguments will not be null.
2112 If no argument index list is given to the @code{nonnull} attribute,
2113 all pointer arguments are marked as non-null. To illustrate, the
2114 following declaration is equivalent to the previous example:
2118 my_memcpy (void *dest, const void *src, size_t len)
2119 __attribute__((nonnull));
2123 @cindex @code{noreturn} function attribute
2124 A few standard library functions, such as @code{abort} and @code{exit},
2125 cannot return. GCC knows this automatically. Some programs define
2126 their own functions that never return. You can declare them
2127 @code{noreturn} to tell the compiler this fact. For example,
2131 void fatal () __attribute__ ((noreturn));
2134 fatal (/* @r{@dots{}} */)
2136 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2142 The @code{noreturn} keyword tells the compiler to assume that
2143 @code{fatal} cannot return. It can then optimize without regard to what
2144 would happen if @code{fatal} ever did return. This makes slightly
2145 better code. More importantly, it helps avoid spurious warnings of
2146 uninitialized variables.
2148 The @code{noreturn} keyword does not affect the exceptional path when that
2149 applies: a @code{noreturn}-marked function may still return to the caller
2150 by throwing an exception or calling @code{longjmp}.
2152 Do not assume that registers saved by the calling function are
2153 restored before calling the @code{noreturn} function.
2155 It does not make sense for a @code{noreturn} function to have a return
2156 type other than @code{void}.
2158 The attribute @code{noreturn} is not implemented in GCC versions
2159 earlier than 2.5. An alternative way to declare that a function does
2160 not return, which works in the current version and in some older
2161 versions, is as follows:
2164 typedef void voidfn ();
2166 volatile voidfn fatal;
2169 This approach does not work in GNU C++.
2172 @cindex @code{nothrow} function attribute
2173 The @code{nothrow} attribute is used to inform the compiler that a
2174 function cannot throw an exception. For example, most functions in
2175 the standard C library can be guaranteed not to throw an exception
2176 with the notable exceptions of @code{qsort} and @code{bsearch} that
2177 take function pointer arguments. The @code{nothrow} attribute is not
2178 implemented in GCC versions earlier than 3.3.
2181 @cindex @code{pure} function attribute
2182 Many functions have no effects except the return value and their
2183 return value depends only on the parameters and/or global variables.
2184 Such a function can be subject
2185 to common subexpression elimination and loop optimization just as an
2186 arithmetic operator would be. These functions should be declared
2187 with the attribute @code{pure}. For example,
2190 int square (int) __attribute__ ((pure));
2194 says that the hypothetical function @code{square} is safe to call
2195 fewer times than the program says.
2197 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2198 Interesting non-pure functions are functions with infinite loops or those
2199 depending on volatile memory or other system resource, that may change between
2200 two consecutive calls (such as @code{feof} in a multithreading environment).
2202 The attribute @code{pure} is not implemented in GCC versions earlier
2205 @item regparm (@var{number})
2206 @cindex @code{regparm} attribute
2207 @cindex functions that are passed arguments in registers on the 386
2208 On the Intel 386, the @code{regparm} attribute causes the compiler to
2209 pass arguments number one to @var{number} if they are of integral type
2210 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2211 take a variable number of arguments will continue to be passed all of their
2212 arguments on the stack.
2214 Beware that on some ELF systems this attribute is unsuitable for
2215 global functions in shared libraries with lazy binding (which is the
2216 default). Lazy binding will send the first call via resolving code in
2217 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2218 per the standard calling conventions. Solaris 8 is affected by this.
2219 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2220 safe since the loaders there save all registers. (Lazy binding can be
2221 disabled with the linker or the loader if desired, to avoid the
2225 @cindex @code{x87regparm} attribute
2226 On the Intel x86 with 80387 @code{x87regparm} attribute causes the
2227 compiler to pass up to 3 floating point arguments in 80387 registers
2228 instead of on the stack. Functions that take a variable number of
2229 arguments will continue to pass all of their floating point arguments
2233 @cindex @code{sseregparm} attribute
2234 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2235 causes the compiler to pass up to 3 floating point arguments in
2236 SSE registers instead of on the stack. Functions that take a
2237 variable number of arguments will continue to pass all of their
2238 floating point arguments on the stack.
2240 @item force_align_arg_pointer
2241 @cindex @code{force_align_arg_pointer} attribute
2242 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
2243 applied to individual function definitions, generating an alternate
2244 prologue and epilogue that realigns the runtime stack. This supports
2245 mixing legacy codes that run with a 4-byte aligned stack with modern
2246 codes that keep a 16-byte stack for SSE compatibility. The alternate
2247 prologue and epilogue are slower and bigger than the regular ones, and
2248 the alternate prologue requires a scratch register; this lowers the
2249 number of registers available if used in conjunction with the
2250 @code{regparm} attribute. The @code{force_align_arg_pointer}
2251 attribute is incompatible with nested functions; this is considered a
2255 @cindex @code{returns_twice} attribute
2256 The @code{returns_twice} attribute tells the compiler that a function may
2257 return more than one time. The compiler will ensure that all registers
2258 are dead before calling such a function and will emit a warning about
2259 the variables that may be clobbered after the second return from the
2260 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2261 The @code{longjmp}-like counterpart of such function, if any, might need
2262 to be marked with the @code{noreturn} attribute.
2265 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2266 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2267 all registers except the stack pointer should be saved in the prologue
2268 regardless of whether they are used or not.
2270 @item section ("@var{section-name}")
2271 @cindex @code{section} function attribute
2272 Normally, the compiler places the code it generates in the @code{text} section.
2273 Sometimes, however, you need additional sections, or you need certain
2274 particular functions to appear in special sections. The @code{section}
2275 attribute specifies that a function lives in a particular section.
2276 For example, the declaration:
2279 extern void foobar (void) __attribute__ ((section ("bar")));
2283 puts the function @code{foobar} in the @code{bar} section.
2285 Some file formats do not support arbitrary sections so the @code{section}
2286 attribute is not available on all platforms.
2287 If you need to map the entire contents of a module to a particular
2288 section, consider using the facilities of the linker instead.
2291 @cindex @code{sentinel} function attribute
2292 This function attribute ensures that a parameter in a function call is
2293 an explicit @code{NULL}. The attribute is only valid on variadic
2294 functions. By default, the sentinel is located at position zero, the
2295 last parameter of the function call. If an optional integer position
2296 argument P is supplied to the attribute, the sentinel must be located at
2297 position P counting backwards from the end of the argument list.
2300 __attribute__ ((sentinel))
2302 __attribute__ ((sentinel(0)))
2305 The attribute is automatically set with a position of 0 for the built-in
2306 functions @code{execl} and @code{execlp}. The built-in function
2307 @code{execle} has the attribute set with a position of 1.
2309 A valid @code{NULL} in this context is defined as zero with any pointer
2310 type. If your system defines the @code{NULL} macro with an integer type
2311 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2312 with a copy that redefines NULL appropriately.
2314 The warnings for missing or incorrect sentinels are enabled with
2318 See long_call/short_call.
2321 See longcall/shortcall.
2324 @cindex signal handler functions on the AVR processors
2325 Use this attribute on the AVR to indicate that the specified
2326 function is a signal handler. The compiler will generate function
2327 entry and exit sequences suitable for use in a signal handler when this
2328 attribute is present. Interrupts will be disabled inside the function.
2331 Use this attribute on the SH to indicate an @code{interrupt_handler}
2332 function should switch to an alternate stack. It expects a string
2333 argument that names a global variable holding the address of the
2338 void f () __attribute__ ((interrupt_handler,
2339 sp_switch ("alt_stack")));
2343 @cindex functions that pop the argument stack on the 386
2344 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2345 assume that the called function will pop off the stack space used to
2346 pass arguments, unless it takes a variable number of arguments.
2349 @cindex tiny data section on the H8/300H and H8S
2350 Use this attribute on the H8/300H and H8S to indicate that the specified
2351 variable should be placed into the tiny data section.
2352 The compiler will generate more efficient code for loads and stores
2353 on data in the tiny data section. Note the tiny data area is limited to
2354 slightly under 32kbytes of data.
2357 Use this attribute on the SH for an @code{interrupt_handler} to return using
2358 @code{trapa} instead of @code{rte}. This attribute expects an integer
2359 argument specifying the trap number to be used.
2362 @cindex @code{unused} attribute.
2363 This attribute, attached to a function, means that the function is meant
2364 to be possibly unused. GCC will not produce a warning for this
2368 @cindex @code{used} attribute.
2369 This attribute, attached to a function, means that code must be emitted
2370 for the function even if it appears that the function is not referenced.
2371 This is useful, for example, when the function is referenced only in
2374 @item visibility ("@var{visibility_type}")
2375 @cindex @code{visibility} attribute
2376 This attribute affects the linkage of the declaration to which it is attached.
2377 There are four supported @var{visibility_type} values: default,
2378 hidden, protected or internal visibility.
2381 void __attribute__ ((visibility ("protected")))
2382 f () @{ /* @r{Do something.} */; @}
2383 int i __attribute__ ((visibility ("hidden")));
2386 The possible values of @var{visibility_type} correspond to the
2387 visibility settings in the ELF gABI.
2390 @c keep this list of visibilities in alphabetical order.
2393 Default visibility is the normal case for the object file format.
2394 This value is available for the visibility attribute to override other
2395 options that may change the assumed visibility of entities.
2397 On ELF, default visibility means that the declaration is visible to other
2398 modules and, in shared libraries, means that the declared entity may be
2401 On Darwin, default visibility means that the declaration is visible to
2404 Default visibility corresponds to ``external linkage'' in the language.
2407 Hidden visibility indicates that the entity declared will have a new
2408 form of linkage, which we'll call ``hidden linkage''. Two
2409 declarations of an object with hidden linkage refer to the same object
2410 if they are in the same shared object.
2413 Internal visibility is like hidden visibility, but with additional
2414 processor specific semantics. Unless otherwise specified by the
2415 psABI, GCC defines internal visibility to mean that a function is
2416 @emph{never} called from another module. Compare this with hidden
2417 functions which, while they cannot be referenced directly by other
2418 modules, can be referenced indirectly via function pointers. By
2419 indicating that a function cannot be called from outside the module,
2420 GCC may for instance omit the load of a PIC register since it is known
2421 that the calling function loaded the correct value.
2424 Protected visibility is like default visibility except that it
2425 indicates that references within the defining module will bind to the
2426 definition in that module. That is, the declared entity cannot be
2427 overridden by another module.
2431 All visibilities are supported on many, but not all, ELF targets
2432 (supported when the assembler supports the @samp{.visibility}
2433 pseudo-op). Default visibility is supported everywhere. Hidden
2434 visibility is supported on Darwin targets.
2436 The visibility attribute should be applied only to declarations which
2437 would otherwise have external linkage. The attribute should be applied
2438 consistently, so that the same entity should not be declared with
2439 different settings of the attribute.
2441 In C++, the visibility attribute applies to types as well as functions
2442 and objects, because in C++ types have linkage. A class must not have
2443 greater visibility than its non-static data member types and bases,
2444 and class members default to the visibility of their class. Also, a
2445 declaration without explicit visibility is limited to the visibility
2448 In C++, you can mark member functions and static member variables of a
2449 class with the visibility attribute. This is useful if if you know a
2450 particular method or static member variable should only be used from
2451 one shared object; then you can mark it hidden while the rest of the
2452 class has default visibility. Care must be taken to avoid breaking
2453 the One Definition Rule; for example, it is usually not useful to mark
2454 an inline method as hidden without marking the whole class as hidden.
2456 A C++ namespace declaration can also have the visibility attribute.
2457 This attribute applies only to the particular namespace body, not to
2458 other definitions of the same namespace; it is equivalent to using
2459 @samp{#pragma GCC visibility} before and after the namespace
2460 definition (@pxref{Visibility Pragmas}).
2462 In C++, if a template argument has limited visibility, this
2463 restriction is implicitly propagated to the template instantiation.
2464 Otherwise, template instantiations and specializations default to the
2465 visibility of their template.
2467 If both the template and enclosing class have explicit visibility, the
2468 visibility from the template is used.
2470 @item warn_unused_result
2471 @cindex @code{warn_unused_result} attribute
2472 The @code{warn_unused_result} attribute causes a warning to be emitted
2473 if a caller of the function with this attribute does not use its
2474 return value. This is useful for functions where not checking
2475 the result is either a security problem or always a bug, such as
2479 int fn () __attribute__ ((warn_unused_result));
2482 if (fn () < 0) return -1;
2488 results in warning on line 5.
2491 @cindex @code{weak} attribute
2492 The @code{weak} attribute causes the declaration to be emitted as a weak
2493 symbol rather than a global. This is primarily useful in defining
2494 library functions which can be overridden in user code, though it can
2495 also be used with non-function declarations. Weak symbols are supported
2496 for ELF targets, and also for a.out targets when using the GNU assembler
2500 @itemx weakref ("@var{target}")
2501 @cindex @code{weakref} attribute
2502 The @code{weakref} attribute marks a declaration as a weak reference.
2503 Without arguments, it should be accompanied by an @code{alias} attribute
2504 naming the target symbol. Optionally, the @var{target} may be given as
2505 an argument to @code{weakref} itself. In either case, @code{weakref}
2506 implicitly marks the declaration as @code{weak}. Without a
2507 @var{target}, given as an argument to @code{weakref} or to @code{alias},
2508 @code{weakref} is equivalent to @code{weak}.
2511 static int x() __attribute__ ((weakref ("y")));
2512 /* is equivalent to... */
2513 static int x() __attribute__ ((weak, weakref, alias ("y")));
2515 static int x() __attribute__ ((weakref));
2516 static int x() __attribute__ ((alias ("y")));
2519 A weak reference is an alias that does not by itself require a
2520 definition to be given for the target symbol. If the target symbol is
2521 only referenced through weak references, then the becomes a @code{weak}
2522 undefined symbol. If it is directly referenced, however, then such
2523 strong references prevail, and a definition will be required for the
2524 symbol, not necessarily in the same translation unit.
2526 The effect is equivalent to moving all references to the alias to a
2527 separate translation unit, renaming the alias to the aliased symbol,
2528 declaring it as weak, compiling the two separate translation units and
2529 performing a reloadable link on them.
2531 At present, a declaration to which @code{weakref} is attached can
2532 only be @code{static}.
2534 @item externally_visible
2535 @cindex @code{externally_visible} attribute.
2536 This attribute, attached to a global variable or function nullify
2537 effect of @option{-fwhole-program} command line option, so the object
2538 remain visible outside the current compilation unit
2542 You can specify multiple attributes in a declaration by separating them
2543 by commas within the double parentheses or by immediately following an
2544 attribute declaration with another attribute declaration.
2546 @cindex @code{#pragma}, reason for not using
2547 @cindex pragma, reason for not using
2548 Some people object to the @code{__attribute__} feature, suggesting that
2549 ISO C's @code{#pragma} should be used instead. At the time
2550 @code{__attribute__} was designed, there were two reasons for not doing
2555 It is impossible to generate @code{#pragma} commands from a macro.
2558 There is no telling what the same @code{#pragma} might mean in another
2562 These two reasons applied to almost any application that might have been
2563 proposed for @code{#pragma}. It was basically a mistake to use
2564 @code{#pragma} for @emph{anything}.
2566 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2567 to be generated from macros. In addition, a @code{#pragma GCC}
2568 namespace is now in use for GCC-specific pragmas. However, it has been
2569 found convenient to use @code{__attribute__} to achieve a natural
2570 attachment of attributes to their corresponding declarations, whereas
2571 @code{#pragma GCC} is of use for constructs that do not naturally form
2572 part of the grammar. @xref{Other Directives,,Miscellaneous
2573 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2575 @node Attribute Syntax
2576 @section Attribute Syntax
2577 @cindex attribute syntax
2579 This section describes the syntax with which @code{__attribute__} may be
2580 used, and the constructs to which attribute specifiers bind, for the C
2581 language. Some details may vary for C++ and Objective-C@. Because of
2582 infelicities in the grammar for attributes, some forms described here
2583 may not be successfully parsed in all cases.
2585 There are some problems with the semantics of attributes in C++. For
2586 example, there are no manglings for attributes, although they may affect
2587 code generation, so problems may arise when attributed types are used in
2588 conjunction with templates or overloading. Similarly, @code{typeid}
2589 does not distinguish between types with different attributes. Support
2590 for attributes in C++ may be restricted in future to attributes on
2591 declarations only, but not on nested declarators.
2593 @xref{Function Attributes}, for details of the semantics of attributes
2594 applying to functions. @xref{Variable Attributes}, for details of the
2595 semantics of attributes applying to variables. @xref{Type Attributes},
2596 for details of the semantics of attributes applying to structure, union
2597 and enumerated types.
2599 An @dfn{attribute specifier} is of the form
2600 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2601 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2602 each attribute is one of the following:
2606 Empty. Empty attributes are ignored.
2609 A word (which may be an identifier such as @code{unused}, or a reserved
2610 word such as @code{const}).
2613 A word, followed by, in parentheses, parameters for the attribute.
2614 These parameters take one of the following forms:
2618 An identifier. For example, @code{mode} attributes use this form.
2621 An identifier followed by a comma and a non-empty comma-separated list
2622 of expressions. For example, @code{format} attributes use this form.
2625 A possibly empty comma-separated list of expressions. For example,
2626 @code{format_arg} attributes use this form with the list being a single
2627 integer constant expression, and @code{alias} attributes use this form
2628 with the list being a single string constant.
2632 An @dfn{attribute specifier list} is a sequence of one or more attribute
2633 specifiers, not separated by any other tokens.
2635 In GNU C, an attribute specifier list may appear after the colon following a
2636 label, other than a @code{case} or @code{default} label. The only
2637 attribute it makes sense to use after a label is @code{unused}. This
2638 feature is intended for code generated by programs which contains labels
2639 that may be unused but which is compiled with @option{-Wall}. It would
2640 not normally be appropriate to use in it human-written code, though it
2641 could be useful in cases where the code that jumps to the label is
2642 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2643 such placement of attribute lists, as it is permissible for a
2644 declaration, which could begin with an attribute list, to be labelled in
2645 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2646 does not arise there.
2648 An attribute specifier list may appear as part of a @code{struct},
2649 @code{union} or @code{enum} specifier. It may go either immediately
2650 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2651 the closing brace. The former syntax is preferred.
2652 Where attribute specifiers follow the closing brace, they are considered
2653 to relate to the structure, union or enumerated type defined, not to any
2654 enclosing declaration the type specifier appears in, and the type
2655 defined is not complete until after the attribute specifiers.
2656 @c Otherwise, there would be the following problems: a shift/reduce
2657 @c conflict between attributes binding the struct/union/enum and
2658 @c binding to the list of specifiers/qualifiers; and "aligned"
2659 @c attributes could use sizeof for the structure, but the size could be
2660 @c changed later by "packed" attributes.
2662 Otherwise, an attribute specifier appears as part of a declaration,
2663 counting declarations of unnamed parameters and type names, and relates
2664 to that declaration (which may be nested in another declaration, for
2665 example in the case of a parameter declaration), or to a particular declarator
2666 within a declaration. Where an
2667 attribute specifier is applied to a parameter declared as a function or
2668 an array, it should apply to the function or array rather than the
2669 pointer to which the parameter is implicitly converted, but this is not
2670 yet correctly implemented.
2672 Any list of specifiers and qualifiers at the start of a declaration may
2673 contain attribute specifiers, whether or not such a list may in that
2674 context contain storage class specifiers. (Some attributes, however,
2675 are essentially in the nature of storage class specifiers, and only make
2676 sense where storage class specifiers may be used; for example,
2677 @code{section}.) There is one necessary limitation to this syntax: the
2678 first old-style parameter declaration in a function definition cannot
2679 begin with an attribute specifier, because such an attribute applies to
2680 the function instead by syntax described below (which, however, is not
2681 yet implemented in this case). In some other cases, attribute
2682 specifiers are permitted by this grammar but not yet supported by the
2683 compiler. All attribute specifiers in this place relate to the
2684 declaration as a whole. In the obsolescent usage where a type of
2685 @code{int} is implied by the absence of type specifiers, such a list of
2686 specifiers and qualifiers may be an attribute specifier list with no
2687 other specifiers or qualifiers.
2689 At present, the first parameter in a function prototype must have some
2690 type specifier which is not an attribute specifier; this resolves an
2691 ambiguity in the interpretation of @code{void f(int
2692 (__attribute__((foo)) x))}, but is subject to change. At present, if
2693 the parentheses of a function declarator contain only attributes then
2694 those attributes are ignored, rather than yielding an error or warning
2695 or implying a single parameter of type int, but this is subject to
2698 An attribute specifier list may appear immediately before a declarator
2699 (other than the first) in a comma-separated list of declarators in a
2700 declaration of more than one identifier using a single list of
2701 specifiers and qualifiers. Such attribute specifiers apply
2702 only to the identifier before whose declarator they appear. For
2706 __attribute__((noreturn)) void d0 (void),
2707 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2712 the @code{noreturn} attribute applies to all the functions
2713 declared; the @code{format} attribute only applies to @code{d1}.
2715 An attribute specifier list may appear immediately before the comma,
2716 @code{=} or semicolon terminating the declaration of an identifier other
2717 than a function definition. At present, such attribute specifiers apply
2718 to the declared object or function, but in future they may attach to the
2719 outermost adjacent declarator. In simple cases there is no difference,
2720 but, for example, in
2723 void (****f)(void) __attribute__((noreturn));
2727 at present the @code{noreturn} attribute applies to @code{f}, which
2728 causes a warning since @code{f} is not a function, but in future it may
2729 apply to the function @code{****f}. The precise semantics of what
2730 attributes in such cases will apply to are not yet specified. Where an
2731 assembler name for an object or function is specified (@pxref{Asm
2732 Labels}), at present the attribute must follow the @code{asm}
2733 specification; in future, attributes before the @code{asm} specification
2734 may apply to the adjacent declarator, and those after it to the declared
2737 An attribute specifier list may, in future, be permitted to appear after
2738 the declarator in a function definition (before any old-style parameter
2739 declarations or the function body).
2741 Attribute specifiers may be mixed with type qualifiers appearing inside
2742 the @code{[]} of a parameter array declarator, in the C99 construct by
2743 which such qualifiers are applied to the pointer to which the array is
2744 implicitly converted. Such attribute specifiers apply to the pointer,
2745 not to the array, but at present this is not implemented and they are
2748 An attribute specifier list may appear at the start of a nested
2749 declarator. At present, there are some limitations in this usage: the
2750 attributes correctly apply to the declarator, but for most individual
2751 attributes the semantics this implies are not implemented.
2752 When attribute specifiers follow the @code{*} of a pointer
2753 declarator, they may be mixed with any type qualifiers present.
2754 The following describes the formal semantics of this syntax. It will make the
2755 most sense if you are familiar with the formal specification of
2756 declarators in the ISO C standard.
2758 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2759 D1}, where @code{T} contains declaration specifiers that specify a type
2760 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2761 contains an identifier @var{ident}. The type specified for @var{ident}
2762 for derived declarators whose type does not include an attribute
2763 specifier is as in the ISO C standard.
2765 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2766 and the declaration @code{T D} specifies the type
2767 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2768 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2769 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2771 If @code{D1} has the form @code{*
2772 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2773 declaration @code{T D} specifies the type
2774 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2775 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2776 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2782 void (__attribute__((noreturn)) ****f) (void);
2786 specifies the type ``pointer to pointer to pointer to pointer to
2787 non-returning function returning @code{void}''. As another example,
2790 char *__attribute__((aligned(8))) *f;
2794 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2795 Note again that this does not work with most attributes; for example,
2796 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2797 is not yet supported.
2799 For compatibility with existing code written for compiler versions that
2800 did not implement attributes on nested declarators, some laxity is
2801 allowed in the placing of attributes. If an attribute that only applies
2802 to types is applied to a declaration, it will be treated as applying to
2803 the type of that declaration. If an attribute that only applies to
2804 declarations is applied to the type of a declaration, it will be treated
2805 as applying to that declaration; and, for compatibility with code
2806 placing the attributes immediately before the identifier declared, such
2807 an attribute applied to a function return type will be treated as
2808 applying to the function type, and such an attribute applied to an array
2809 element type will be treated as applying to the array type. If an
2810 attribute that only applies to function types is applied to a
2811 pointer-to-function type, it will be treated as applying to the pointer
2812 target type; if such an attribute is applied to a function return type
2813 that is not a pointer-to-function type, it will be treated as applying
2814 to the function type.
2816 @node Function Prototypes
2817 @section Prototypes and Old-Style Function Definitions
2818 @cindex function prototype declarations
2819 @cindex old-style function definitions
2820 @cindex promotion of formal parameters
2822 GNU C extends ISO C to allow a function prototype to override a later
2823 old-style non-prototype definition. Consider the following example:
2826 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2833 /* @r{Prototype function declaration.} */
2834 int isroot P((uid_t));
2836 /* @r{Old-style function definition.} */
2838 isroot (x) /* @r{??? lossage here ???} */
2845 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2846 not allow this example, because subword arguments in old-style
2847 non-prototype definitions are promoted. Therefore in this example the
2848 function definition's argument is really an @code{int}, which does not
2849 match the prototype argument type of @code{short}.
2851 This restriction of ISO C makes it hard to write code that is portable
2852 to traditional C compilers, because the programmer does not know
2853 whether the @code{uid_t} type is @code{short}, @code{int}, or
2854 @code{long}. Therefore, in cases like these GNU C allows a prototype
2855 to override a later old-style definition. More precisely, in GNU C, a
2856 function prototype argument type overrides the argument type specified
2857 by a later old-style definition if the former type is the same as the
2858 latter type before promotion. Thus in GNU C the above example is
2859 equivalent to the following:
2872 GNU C++ does not support old-style function definitions, so this
2873 extension is irrelevant.
2876 @section C++ Style Comments
2878 @cindex C++ comments
2879 @cindex comments, C++ style
2881 In GNU C, you may use C++ style comments, which start with @samp{//} and
2882 continue until the end of the line. Many other C implementations allow
2883 such comments, and they are included in the 1999 C standard. However,
2884 C++ style comments are not recognized if you specify an @option{-std}
2885 option specifying a version of ISO C before C99, or @option{-ansi}
2886 (equivalent to @option{-std=c89}).
2889 @section Dollar Signs in Identifier Names
2891 @cindex dollar signs in identifier names
2892 @cindex identifier names, dollar signs in
2894 In GNU C, you may normally use dollar signs in identifier names.
2895 This is because many traditional C implementations allow such identifiers.
2896 However, dollar signs in identifiers are not supported on a few target
2897 machines, typically because the target assembler does not allow them.
2899 @node Character Escapes
2900 @section The Character @key{ESC} in Constants
2902 You can use the sequence @samp{\e} in a string or character constant to
2903 stand for the ASCII character @key{ESC}.
2906 @section Inquiring on Alignment of Types or Variables
2908 @cindex type alignment
2909 @cindex variable alignment
2911 The keyword @code{__alignof__} allows you to inquire about how an object
2912 is aligned, or the minimum alignment usually required by a type. Its
2913 syntax is just like @code{sizeof}.
2915 For example, if the target machine requires a @code{double} value to be
2916 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2917 This is true on many RISC machines. On more traditional machine
2918 designs, @code{__alignof__ (double)} is 4 or even 2.
2920 Some machines never actually require alignment; they allow reference to any
2921 data type even at an odd address. For these machines, @code{__alignof__}
2922 reports the @emph{recommended} alignment of a type.
2924 If the operand of @code{__alignof__} is an lvalue rather than a type,
2925 its value is the required alignment for its type, taking into account
2926 any minimum alignment specified with GCC's @code{__attribute__}
2927 extension (@pxref{Variable Attributes}). For example, after this
2931 struct foo @{ int x; char y; @} foo1;
2935 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2936 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2938 It is an error to ask for the alignment of an incomplete type.
2940 @node Variable Attributes
2941 @section Specifying Attributes of Variables
2942 @cindex attribute of variables
2943 @cindex variable attributes
2945 The keyword @code{__attribute__} allows you to specify special
2946 attributes of variables or structure fields. This keyword is followed
2947 by an attribute specification inside double parentheses. Some
2948 attributes are currently defined generically for variables.
2949 Other attributes are defined for variables on particular target
2950 systems. Other attributes are available for functions
2951 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
2952 Other front ends might define more attributes
2953 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2955 You may also specify attributes with @samp{__} preceding and following
2956 each keyword. This allows you to use them in header files without
2957 being concerned about a possible macro of the same name. For example,
2958 you may use @code{__aligned__} instead of @code{aligned}.
2960 @xref{Attribute Syntax}, for details of the exact syntax for using
2964 @cindex @code{aligned} attribute
2965 @item aligned (@var{alignment})
2966 This attribute specifies a minimum alignment for the variable or
2967 structure field, measured in bytes. For example, the declaration:
2970 int x __attribute__ ((aligned (16))) = 0;
2974 causes the compiler to allocate the global variable @code{x} on a
2975 16-byte boundary. On a 68040, this could be used in conjunction with
2976 an @code{asm} expression to access the @code{move16} instruction which
2977 requires 16-byte aligned operands.
2979 You can also specify the alignment of structure fields. For example, to
2980 create a double-word aligned @code{int} pair, you could write:
2983 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2987 This is an alternative to creating a union with a @code{double} member
2988 that forces the union to be double-word aligned.
2990 As in the preceding examples, you can explicitly specify the alignment
2991 (in bytes) that you wish the compiler to use for a given variable or
2992 structure field. Alternatively, you can leave out the alignment factor
2993 and just ask the compiler to align a variable or field to the maximum
2994 useful alignment for the target machine you are compiling for. For
2995 example, you could write:
2998 short array[3] __attribute__ ((aligned));
3001 Whenever you leave out the alignment factor in an @code{aligned} attribute
3002 specification, the compiler automatically sets the alignment for the declared
3003 variable or field to the largest alignment which is ever used for any data
3004 type on the target machine you are compiling for. Doing this can often make
3005 copy operations more efficient, because the compiler can use whatever
3006 instructions copy the biggest chunks of memory when performing copies to
3007 or from the variables or fields that you have aligned this way.
3009 The @code{aligned} attribute can only increase the alignment; but you
3010 can decrease it by specifying @code{packed} as well. See below.
3012 Note that the effectiveness of @code{aligned} attributes may be limited
3013 by inherent limitations in your linker. On many systems, the linker is
3014 only able to arrange for variables to be aligned up to a certain maximum
3015 alignment. (For some linkers, the maximum supported alignment may
3016 be very very small.) If your linker is only able to align variables
3017 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3018 in an @code{__attribute__} will still only provide you with 8 byte
3019 alignment. See your linker documentation for further information.
3021 @item cleanup (@var{cleanup_function})
3022 @cindex @code{cleanup} attribute
3023 The @code{cleanup} attribute runs a function when the variable goes
3024 out of scope. This attribute can only be applied to auto function
3025 scope variables; it may not be applied to parameters or variables
3026 with static storage duration. The function must take one parameter,
3027 a pointer to a type compatible with the variable. The return value
3028 of the function (if any) is ignored.
3030 If @option{-fexceptions} is enabled, then @var{cleanup_function}
3031 will be run during the stack unwinding that happens during the
3032 processing of the exception. Note that the @code{cleanup} attribute
3033 does not allow the exception to be caught, only to perform an action.
3034 It is undefined what happens if @var{cleanup_function} does not
3039 @cindex @code{common} attribute
3040 @cindex @code{nocommon} attribute
3043 The @code{common} attribute requests GCC to place a variable in
3044 ``common'' storage. The @code{nocommon} attribute requests the
3045 opposite---to allocate space for it directly.
3047 These attributes override the default chosen by the
3048 @option{-fno-common} and @option{-fcommon} flags respectively.
3051 @cindex @code{deprecated} attribute
3052 The @code{deprecated} attribute results in a warning if the variable
3053 is used anywhere in the source file. This is useful when identifying
3054 variables that are expected to be removed in a future version of a
3055 program. The warning also includes the location of the declaration
3056 of the deprecated variable, to enable users to easily find further
3057 information about why the variable is deprecated, or what they should
3058 do instead. Note that the warning only occurs for uses:
3061 extern int old_var __attribute__ ((deprecated));
3063 int new_fn () @{ return old_var; @}
3066 results in a warning on line 3 but not line 2.
3068 The @code{deprecated} attribute can also be used for functions and
3069 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3071 @item mode (@var{mode})
3072 @cindex @code{mode} attribute
3073 This attribute specifies the data type for the declaration---whichever
3074 type corresponds to the mode @var{mode}. This in effect lets you
3075 request an integer or floating point type according to its width.
3077 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3078 indicate the mode corresponding to a one-byte integer, @samp{word} or
3079 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3080 or @samp{__pointer__} for the mode used to represent pointers.
3083 @cindex @code{packed} attribute
3084 The @code{packed} attribute specifies that a variable or structure field
3085 should have the smallest possible alignment---one byte for a variable,
3086 and one bit for a field, unless you specify a larger value with the
3087 @code{aligned} attribute.
3089 Here is a structure in which the field @code{x} is packed, so that it
3090 immediately follows @code{a}:
3096 int x[2] __attribute__ ((packed));
3100 @item section ("@var{section-name}")
3101 @cindex @code{section} variable attribute
3102 Normally, the compiler places the objects it generates in sections like
3103 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3104 or you need certain particular variables to appear in special sections,
3105 for example to map to special hardware. The @code{section}
3106 attribute specifies that a variable (or function) lives in a particular
3107 section. For example, this small program uses several specific section names:
3110 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3111 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3112 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3113 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3117 /* @r{Initialize stack pointer} */
3118 init_sp (stack + sizeof (stack));
3120 /* @r{Initialize initialized data} */
3121 memcpy (&init_data, &data, &edata - &data);
3123 /* @r{Turn on the serial ports} */
3130 Use the @code{section} attribute with an @emph{initialized} definition
3131 of a @emph{global} variable, as shown in the example. GCC issues
3132 a warning and otherwise ignores the @code{section} attribute in
3133 uninitialized variable declarations.
3135 You may only use the @code{section} attribute with a fully initialized
3136 global definition because of the way linkers work. The linker requires
3137 each object be defined once, with the exception that uninitialized
3138 variables tentatively go in the @code{common} (or @code{bss}) section
3139 and can be multiply ``defined''. You can force a variable to be
3140 initialized with the @option{-fno-common} flag or the @code{nocommon}
3143 Some file formats do not support arbitrary sections so the @code{section}
3144 attribute is not available on all platforms.
3145 If you need to map the entire contents of a module to a particular
3146 section, consider using the facilities of the linker instead.
3149 @cindex @code{shared} variable attribute
3150 On Microsoft Windows, in addition to putting variable definitions in a named
3151 section, the section can also be shared among all running copies of an
3152 executable or DLL@. For example, this small program defines shared data
3153 by putting it in a named section @code{shared} and marking the section
3157 int foo __attribute__((section ("shared"), shared)) = 0;
3162 /* @r{Read and write foo. All running
3163 copies see the same value.} */
3169 You may only use the @code{shared} attribute along with @code{section}
3170 attribute with a fully initialized global definition because of the way
3171 linkers work. See @code{section} attribute for more information.
3173 The @code{shared} attribute is only available on Microsoft Windows@.
3175 @item tls_model ("@var{tls_model}")
3176 @cindex @code{tls_model} attribute
3177 The @code{tls_model} attribute sets thread-local storage model
3178 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3179 overriding @option{-ftls-model=} command line switch on a per-variable
3181 The @var{tls_model} argument should be one of @code{global-dynamic},
3182 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3184 Not all targets support this attribute.
3187 This attribute, attached to a variable, means that the variable is meant
3188 to be possibly unused. GCC will not produce a warning for this
3192 This attribute, attached to a variable, means that the variable must be
3193 emitted even if it appears that the variable is not referenced.
3195 @item vector_size (@var{bytes})
3196 This attribute specifies the vector size for the variable, measured in
3197 bytes. For example, the declaration:
3200 int foo __attribute__ ((vector_size (16)));
3204 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3205 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3206 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3208 This attribute is only applicable to integral and float scalars,
3209 although arrays, pointers, and function return values are allowed in
3210 conjunction with this construct.
3212 Aggregates with this attribute are invalid, even if they are of the same
3213 size as a corresponding scalar. For example, the declaration:
3216 struct S @{ int a; @};
3217 struct S __attribute__ ((vector_size (16))) foo;
3221 is invalid even if the size of the structure is the same as the size of
3225 The @code{selectany} attribute causes an initialized global variable to
3226 have link-once semantics. When multiple definitions of the variable are
3227 encountered by the linker, the first is selected and the remainder are
3228 discarded. Following usage by the Microsoft compiler, the linker is told
3229 @emph{not} to warn about size or content differences of the multiple
3232 Although the primary usage of this attribute is for POD types, the
3233 attribute can also be applied to global C++ objects that are initialized
3234 by a constructor. In this case, the static initialization and destruction
3235 code for the object is emitted in each translation defining the object,
3236 but the calls to the constructor and destructor are protected by a
3237 link-once guard variable.
3239 The @code{selectany} attribute is only available on Microsoft Windows
3240 targets. You can use @code{__declspec (selectany)} as a synonym for
3241 @code{__attribute__ ((selectany))} for compatibility with other
3245 The @code{weak} attribute is described in @xref{Function Attributes}.
3248 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3251 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3255 @subsection M32R/D Variable Attributes
3257 One attribute is currently defined for the M32R/D@.
3260 @item model (@var{model-name})
3261 @cindex variable addressability on the M32R/D
3262 Use this attribute on the M32R/D to set the addressability of an object.
3263 The identifier @var{model-name} is one of @code{small}, @code{medium},
3264 or @code{large}, representing each of the code models.
3266 Small model objects live in the lower 16MB of memory (so that their
3267 addresses can be loaded with the @code{ld24} instruction).
3269 Medium and large model objects may live anywhere in the 32-bit address space
3270 (the compiler will generate @code{seth/add3} instructions to load their
3274 @anchor{i386 Variable Attributes}
3275 @subsection i386 Variable Attributes
3277 Two attributes are currently defined for i386 configurations:
3278 @code{ms_struct} and @code{gcc_struct}
3283 @cindex @code{ms_struct} attribute
3284 @cindex @code{gcc_struct} attribute
3286 If @code{packed} is used on a structure, or if bit-fields are used
3287 it may be that the Microsoft ABI packs them differently
3288 than GCC would normally pack them. Particularly when moving packed
3289 data between functions compiled with GCC and the native Microsoft compiler
3290 (either via function call or as data in a file), it may be necessary to access
3293 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3294 compilers to match the native Microsoft compiler.
3296 The Microsoft structure layout algorithm is fairly simple with the exception
3297 of the bitfield packing:
3299 The padding and alignment of members of structures and whether a bit field
3300 can straddle a storage-unit boundary
3303 @item Structure members are stored sequentially in the order in which they are
3304 declared: the first member has the lowest memory address and the last member
3307 @item Every data object has an alignment-requirement. The alignment-requirement
3308 for all data except structures, unions, and arrays is either the size of the
3309 object or the current packing size (specified with either the aligned attribute
3310 or the pack pragma), whichever is less. For structures, unions, and arrays,
3311 the alignment-requirement is the largest alignment-requirement of its members.
3312 Every object is allocated an offset so that:
3314 offset % alignment-requirement == 0
3316 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
3317 unit if the integral types are the same size and if the next bit field fits
3318 into the current allocation unit without crossing the boundary imposed by the
3319 common alignment requirements of the bit fields.
3322 Handling of zero-length bitfields:
3324 MSVC interprets zero-length bitfields in the following ways:
3327 @item If a zero-length bitfield is inserted between two bitfields that would
3328 normally be coalesced, the bitfields will not be coalesced.
3335 unsigned long bf_1 : 12;
3337 unsigned long bf_2 : 12;
3341 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
3342 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
3344 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
3345 alignment of the zero-length bitfield is greater than the member that follows it,
3346 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
3366 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
3367 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
3368 bitfield will not affect the alignment of @code{bar} or, as a result, the size
3371 Taking this into account, it is important to note the following:
3374 @item If a zero-length bitfield follows a normal bitfield, the type of the
3375 zero-length bitfield may affect the alignment of the structure as whole. For
3376 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
3377 normal bitfield, and is of type short.
3379 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
3380 still affect the alignment of the structure:
3390 Here, @code{t4} will take up 4 bytes.
3393 @item Zero-length bitfields following non-bitfield members are ignored:
3404 Here, @code{t5} will take up 2 bytes.
3408 @subsection PowerPC Variable Attributes
3410 Three attributes currently are defined for PowerPC configurations:
3411 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
3413 For full documentation of the struct attributes please see the
3414 documentation in the @xref{i386 Variable Attributes}, section.
3416 For documentation of @code{altivec} attribute please see the
3417 documentation in the @xref{PowerPC Type Attributes}, section.
3419 @subsection Xstormy16 Variable Attributes
3421 One attribute is currently defined for xstormy16 configurations:
3426 @cindex @code{below100} attribute
3428 If a variable has the @code{below100} attribute (@code{BELOW100} is
3429 allowed also), GCC will place the variable in the first 0x100 bytes of
3430 memory and use special opcodes to access it. Such variables will be
3431 placed in either the @code{.bss_below100} section or the
3432 @code{.data_below100} section.
3436 @node Type Attributes
3437 @section Specifying Attributes of Types
3438 @cindex attribute of types
3439 @cindex type attributes
3441 The keyword @code{__attribute__} allows you to specify special
3442 attributes of @code{struct} and @code{union} types when you define
3443 such types. This keyword is followed by an attribute specification
3444 inside double parentheses. Seven attributes are currently defined for
3445 types: @code{aligned}, @code{packed}, @code{transparent_union},
3446 @code{unused}, @code{deprecated}, @code{visibility}, and
3447 @code{may_alias}. Other attributes are defined for functions
3448 (@pxref{Function Attributes}) and for variables (@pxref{Variable
3451 You may also specify any one of these attributes with @samp{__}
3452 preceding and following its keyword. This allows you to use these
3453 attributes in header files without being concerned about a possible
3454 macro of the same name. For example, you may use @code{__aligned__}
3455 instead of @code{aligned}.
3457 You may specify type attributes either in a @code{typedef} declaration
3458 or in an enum, struct or union type declaration or definition.
3460 For an enum, struct or union type, you may specify attributes either
3461 between the enum, struct or union tag and the name of the type, or
3462 just past the closing curly brace of the @emph{definition}. The
3463 former syntax is preferred.
3465 @xref{Attribute Syntax}, for details of the exact syntax for using
3469 @cindex @code{aligned} attribute
3470 @item aligned (@var{alignment})
3471 This attribute specifies a minimum alignment (in bytes) for variables
3472 of the specified type. For example, the declarations:
3475 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3476 typedef int more_aligned_int __attribute__ ((aligned (8)));
3480 force the compiler to insure (as far as it can) that each variable whose
3481 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3482 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3483 variables of type @code{struct S} aligned to 8-byte boundaries allows
3484 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3485 store) instructions when copying one variable of type @code{struct S} to
3486 another, thus improving run-time efficiency.
3488 Note that the alignment of any given @code{struct} or @code{union} type
3489 is required by the ISO C standard to be at least a perfect multiple of
3490 the lowest common multiple of the alignments of all of the members of
3491 the @code{struct} or @code{union} in question. This means that you @emph{can}
3492 effectively adjust the alignment of a @code{struct} or @code{union}
3493 type by attaching an @code{aligned} attribute to any one of the members
3494 of such a type, but the notation illustrated in the example above is a
3495 more obvious, intuitive, and readable way to request the compiler to
3496 adjust the alignment of an entire @code{struct} or @code{union} type.
3498 As in the preceding example, you can explicitly specify the alignment
3499 (in bytes) that you wish the compiler to use for a given @code{struct}
3500 or @code{union} type. Alternatively, you can leave out the alignment factor
3501 and just ask the compiler to align a type to the maximum
3502 useful alignment for the target machine you are compiling for. For
3503 example, you could write:
3506 struct S @{ short f[3]; @} __attribute__ ((aligned));
3509 Whenever you leave out the alignment factor in an @code{aligned}
3510 attribute specification, the compiler automatically sets the alignment
3511 for the type to the largest alignment which is ever used for any data
3512 type on the target machine you are compiling for. Doing this can often
3513 make copy operations more efficient, because the compiler can use
3514 whatever instructions copy the biggest chunks of memory when performing
3515 copies to or from the variables which have types that you have aligned
3518 In the example above, if the size of each @code{short} is 2 bytes, then
3519 the size of the entire @code{struct S} type is 6 bytes. The smallest
3520 power of two which is greater than or equal to that is 8, so the
3521 compiler sets the alignment for the entire @code{struct S} type to 8
3524 Note that although you can ask the compiler to select a time-efficient
3525 alignment for a given type and then declare only individual stand-alone
3526 objects of that type, the compiler's ability to select a time-efficient
3527 alignment is primarily useful only when you plan to create arrays of
3528 variables having the relevant (efficiently aligned) type. If you
3529 declare or use arrays of variables of an efficiently-aligned type, then
3530 it is likely that your program will also be doing pointer arithmetic (or
3531 subscripting, which amounts to the same thing) on pointers to the
3532 relevant type, and the code that the compiler generates for these
3533 pointer arithmetic operations will often be more efficient for
3534 efficiently-aligned types than for other types.
3536 The @code{aligned} attribute can only increase the alignment; but you
3537 can decrease it by specifying @code{packed} as well. See below.
3539 Note that the effectiveness of @code{aligned} attributes may be limited
3540 by inherent limitations in your linker. On many systems, the linker is
3541 only able to arrange for variables to be aligned up to a certain maximum
3542 alignment. (For some linkers, the maximum supported alignment may
3543 be very very small.) If your linker is only able to align variables
3544 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3545 in an @code{__attribute__} will still only provide you with 8 byte
3546 alignment. See your linker documentation for further information.
3549 This attribute, attached to @code{struct} or @code{union} type
3550 definition, specifies that each member (other than zero-width bitfields)
3551 of the structure or union is placed to minimize the memory required. When
3552 attached to an @code{enum} definition, it indicates that the smallest
3553 integral type should be used.
3555 @opindex fshort-enums
3556 Specifying this attribute for @code{struct} and @code{union} types is
3557 equivalent to specifying the @code{packed} attribute on each of the
3558 structure or union members. Specifying the @option{-fshort-enums}
3559 flag on the line is equivalent to specifying the @code{packed}
3560 attribute on all @code{enum} definitions.
3562 In the following example @code{struct my_packed_struct}'s members are
3563 packed closely together, but the internal layout of its @code{s} member
3564 is not packed---to do that, @code{struct my_unpacked_struct} would need to
3568 struct my_unpacked_struct
3574 struct __attribute__ ((__packed__)) my_packed_struct
3578 struct my_unpacked_struct s;
3582 You may only specify this attribute on the definition of a @code{enum},
3583 @code{struct} or @code{union}, not on a @code{typedef} which does not
3584 also define the enumerated type, structure or union.
3586 @item transparent_union
3587 This attribute, attached to a @code{union} type definition, indicates
3588 that any function parameter having that union type causes calls to that
3589 function to be treated in a special way.
3591 First, the argument corresponding to a transparent union type can be of
3592 any type in the union; no cast is required. Also, if the union contains
3593 a pointer type, the corresponding argument can be a null pointer
3594 constant or a void pointer expression; and if the union contains a void
3595 pointer type, the corresponding argument can be any pointer expression.
3596 If the union member type is a pointer, qualifiers like @code{const} on
3597 the referenced type must be respected, just as with normal pointer
3600 Second, the argument is passed to the function using the calling
3601 conventions of the first member of the transparent union, not the calling
3602 conventions of the union itself. All members of the union must have the
3603 same machine representation; this is necessary for this argument passing
3606 Transparent unions are designed for library functions that have multiple
3607 interfaces for compatibility reasons. For example, suppose the
3608 @code{wait} function must accept either a value of type @code{int *} to
3609 comply with Posix, or a value of type @code{union wait *} to comply with
3610 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3611 @code{wait} would accept both kinds of arguments, but it would also
3612 accept any other pointer type and this would make argument type checking
3613 less useful. Instead, @code{<sys/wait.h>} might define the interface
3621 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3623 pid_t wait (wait_status_ptr_t);
3626 This interface allows either @code{int *} or @code{union wait *}
3627 arguments to be passed, using the @code{int *} calling convention.
3628 The program can call @code{wait} with arguments of either type:
3631 int w1 () @{ int w; return wait (&w); @}
3632 int w2 () @{ union wait w; return wait (&w); @}
3635 With this interface, @code{wait}'s implementation might look like this:
3638 pid_t wait (wait_status_ptr_t p)
3640 return waitpid (-1, p.__ip, 0);
3645 When attached to a type (including a @code{union} or a @code{struct}),
3646 this attribute means that variables of that type are meant to appear
3647 possibly unused. GCC will not produce a warning for any variables of
3648 that type, even if the variable appears to do nothing. This is often
3649 the case with lock or thread classes, which are usually defined and then
3650 not referenced, but contain constructors and destructors that have
3651 nontrivial bookkeeping functions.
3654 The @code{deprecated} attribute results in a warning if the type
3655 is used anywhere in the source file. This is useful when identifying
3656 types that are expected to be removed in a future version of a program.
3657 If possible, the warning also includes the location of the declaration
3658 of the deprecated type, to enable users to easily find further
3659 information about why the type is deprecated, or what they should do
3660 instead. Note that the warnings only occur for uses and then only
3661 if the type is being applied to an identifier that itself is not being
3662 declared as deprecated.
3665 typedef int T1 __attribute__ ((deprecated));
3669 typedef T1 T3 __attribute__ ((deprecated));
3670 T3 z __attribute__ ((deprecated));
3673 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3674 warning is issued for line 4 because T2 is not explicitly
3675 deprecated. Line 5 has no warning because T3 is explicitly
3676 deprecated. Similarly for line 6.
3678 The @code{deprecated} attribute can also be used for functions and
3679 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3682 Accesses to objects with types with this attribute are not subjected to
3683 type-based alias analysis, but are instead assumed to be able to alias
3684 any other type of objects, just like the @code{char} type. See
3685 @option{-fstrict-aliasing} for more information on aliasing issues.
3690 typedef short __attribute__((__may_alias__)) short_a;
3696 short_a *b = (short_a *) &a;
3700 if (a == 0x12345678)
3707 If you replaced @code{short_a} with @code{short} in the variable
3708 declaration, the above program would abort when compiled with
3709 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3710 above in recent GCC versions.
3713 In C++, attribute visibility (@pxref{Function Attributes}) can also be
3714 applied to class, struct, union and enum types. Unlike other type
3715 attributes, the attribute must appear between the initial keyword and
3716 the name of the type; it cannot appear after the body of the type.
3718 Note that the type visibility is applied to vague linkage entities
3719 associated with the class (vtable, typeinfo node, etc.). In
3720 particular, if a class is thrown as an exception in one shared object
3721 and caught in another, the class must have default visibility.
3722 Otherwise the two shared objects will be unable to use the same
3723 typeinfo node and exception handling will break.
3725 @subsection ARM Type Attributes
3727 On those ARM targets that support @code{dllimport} (such as Symbian
3728 OS), you can use the @code{notshared} attribute to indicate that the
3729 virtual table and other similar data for a class should not be
3730 exported from a DLL@. For example:
3733 class __declspec(notshared) C @{
3735 __declspec(dllimport) C();
3739 __declspec(dllexport)
3743 In this code, @code{C::C} is exported from the current DLL, but the
3744 virtual table for @code{C} is not exported. (You can use
3745 @code{__attribute__} instead of @code{__declspec} if you prefer, but
3746 most Symbian OS code uses @code{__declspec}.)
3748 @anchor{i386 Type Attributes}
3749 @subsection i386 Type Attributes
3751 Two attributes are currently defined for i386 configurations:
3752 @code{ms_struct} and @code{gcc_struct}
3756 @cindex @code{ms_struct}
3757 @cindex @code{gcc_struct}
3759 If @code{packed} is used on a structure, or if bit-fields are used
3760 it may be that the Microsoft ABI packs them differently
3761 than GCC would normally pack them. Particularly when moving packed
3762 data between functions compiled with GCC and the native Microsoft compiler
3763 (either via function call or as data in a file), it may be necessary to access
3766 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3767 compilers to match the native Microsoft compiler.
3770 To specify multiple attributes, separate them by commas within the
3771 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3774 @anchor{PowerPC Type Attributes}
3775 @subsection PowerPC Type Attributes
3777 Three attributes currently are defined for PowerPC configurations:
3778 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
3780 For full documentation of the struct attributes please see the
3781 documentation in the @xref{i386 Type Attributes}, section.
3783 The @code{altivec} attribute allows one to declare AltiVec vector data
3784 types supported by the AltiVec Programming Interface Manual. The
3785 attribute requires an argument to specify one of three vector types:
3786 @code{vector__}, @code{pixel__} (always followed by unsigned short),
3787 and @code{bool__} (always followed by unsigned).
3790 __attribute__((altivec(vector__)))
3791 __attribute__((altivec(pixel__))) unsigned short
3792 __attribute__((altivec(bool__))) unsigned
3795 These attributes mainly are intended to support the @code{__vector},
3796 @code{__pixel}, and @code{__bool} AltiVec keywords.
3799 @section An Inline Function is As Fast As a Macro
3800 @cindex inline functions
3801 @cindex integrating function code
3803 @cindex macros, inline alternative
3805 By declaring a function inline, you can direct GCC to make
3806 calls to that function faster. One way GCC can achieve this is to
3807 integrate that function's code into the code for its callers. This
3808 makes execution faster by eliminating the function-call overhead; in
3809 addition, if any of the actual argument values are constant, their
3810 known values may permit simplifications at compile time so that not
3811 all of the inline function's code needs to be included. The effect on
3812 code size is less predictable; object code may be larger or smaller
3813 with function inlining, depending on the particular case. You can
3814 also direct GCC to try to integrate all ``simple enough'' functions
3815 into their callers with the option @option{-finline-functions}.
3817 GCC implements three different semantics of declaring a function
3818 inline. One is available with @option{-std=gnu89} or when @code{gnu_inline}
3819 attribute is present on all inline declarations, another when
3820 @option{-std=c99} or @option{-std=gnu99}, and the third is used when
3823 To declare a function inline, use the @code{inline} keyword in its
3824 declaration, like this:
3834 If you are writing a header file to be included in ISO C89 programs, write
3835 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
3837 The three types of inlining behave similarly in two important cases:
3838 when the @code{inline} keyword is used on a @code{static} function,
3839 like the example above, and when a function is first declared without
3840 using the @code{inline} keyword and then is defined with
3841 @code{inline}, like this:
3844 extern int inc (int *a);
3852 In both of these common cases, the program behaves the same as if you
3853 had not used the @code{inline} keyword, except for its speed.
3855 @cindex inline functions, omission of
3856 @opindex fkeep-inline-functions
3857 When a function is both inline and @code{static}, if all calls to the
3858 function are integrated into the caller, and the function's address is
3859 never used, then the function's own assembler code is never referenced.
3860 In this case, GCC does not actually output assembler code for the
3861 function, unless you specify the option @option{-fkeep-inline-functions}.
3862 Some calls cannot be integrated for various reasons (in particular,
3863 calls that precede the function's definition cannot be integrated, and
3864 neither can recursive calls within the definition). If there is a
3865 nonintegrated call, then the function is compiled to assembler code as
3866 usual. The function must also be compiled as usual if the program
3867 refers to its address, because that can't be inlined.
3870 Note that certain usages in a function definition can make it unsuitable
3871 for inline substitution. Among these usages are: use of varargs, use of
3872 alloca, use of variable sized data types (@pxref{Variable Length}),
3873 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3874 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3875 will warn when a function marked @code{inline} could not be substituted,
3876 and will give the reason for the failure.
3878 @cindex automatic @code{inline} for C++ member fns
3879 @cindex @code{inline} automatic for C++ member fns
3880 @cindex member fns, automatically @code{inline}
3881 @cindex C++ member fns, automatically @code{inline}
3882 @opindex fno-default-inline
3883 As required by ISO C++, GCC considers member functions defined within
3884 the body of a class to be marked inline even if they are
3885 not explicitly declared with the @code{inline} keyword. You can
3886 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
3887 Options,,Options Controlling C++ Dialect}.
3889 GCC does not inline any functions when not optimizing unless you specify
3890 the @samp{always_inline} attribute for the function, like this:
3893 /* @r{Prototype.} */
3894 inline void foo (const char) __attribute__((always_inline));
3897 The remainder of this section is specific to GNU C89 inlining.
3899 @cindex non-static inline function
3900 When an inline function is not @code{static}, then the compiler must assume
3901 that there may be calls from other source files; since a global symbol can
3902 be defined only once in any program, the function must not be defined in
3903 the other source files, so the calls therein cannot be integrated.
3904 Therefore, a non-@code{static} inline function is always compiled on its
3905 own in the usual fashion.
3907 If you specify both @code{inline} and @code{extern} in the function
3908 definition, then the definition is used only for inlining. In no case
3909 is the function compiled on its own, not even if you refer to its
3910 address explicitly. Such an address becomes an external reference, as
3911 if you had only declared the function, and had not defined it.
3913 This combination of @code{inline} and @code{extern} has almost the
3914 effect of a macro. The way to use it is to put a function definition in
3915 a header file with these keywords, and put another copy of the
3916 definition (lacking @code{inline} and @code{extern}) in a library file.
3917 The definition in the header file will cause most calls to the function
3918 to be inlined. If any uses of the function remain, they will refer to
3919 the single copy in the library.
3922 @section Assembler Instructions with C Expression Operands
3923 @cindex extended @code{asm}
3924 @cindex @code{asm} expressions
3925 @cindex assembler instructions
3928 In an assembler instruction using @code{asm}, you can specify the
3929 operands of the instruction using C expressions. This means you need not
3930 guess which registers or memory locations will contain the data you want
3933 You must specify an assembler instruction template much like what
3934 appears in a machine description, plus an operand constraint string for
3937 For example, here is how to use the 68881's @code{fsinx} instruction:
3940 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3944 Here @code{angle} is the C expression for the input operand while
3945 @code{result} is that of the output operand. Each has @samp{"f"} as its
3946 operand constraint, saying that a floating point register is required.
3947 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3948 output operands' constraints must use @samp{=}. The constraints use the
3949 same language used in the machine description (@pxref{Constraints}).
3951 Each operand is described by an operand-constraint string followed by
3952 the C expression in parentheses. A colon separates the assembler
3953 template from the first output operand and another separates the last
3954 output operand from the first input, if any. Commas separate the
3955 operands within each group. The total number of operands is currently
3956 limited to 30; this limitation may be lifted in some future version of
3959 If there are no output operands but there are input operands, you must
3960 place two consecutive colons surrounding the place where the output
3963 As of GCC version 3.1, it is also possible to specify input and output
3964 operands using symbolic names which can be referenced within the
3965 assembler code. These names are specified inside square brackets
3966 preceding the constraint string, and can be referenced inside the
3967 assembler code using @code{%[@var{name}]} instead of a percentage sign
3968 followed by the operand number. Using named operands the above example
3972 asm ("fsinx %[angle],%[output]"
3973 : [output] "=f" (result)
3974 : [angle] "f" (angle));
3978 Note that the symbolic operand names have no relation whatsoever to
3979 other C identifiers. You may use any name you like, even those of
3980 existing C symbols, but you must ensure that no two operands within the same
3981 assembler construct use the same symbolic name.
3983 Output operand expressions must be lvalues; the compiler can check this.
3984 The input operands need not be lvalues. The compiler cannot check
3985 whether the operands have data types that are reasonable for the
3986 instruction being executed. It does not parse the assembler instruction
3987 template and does not know what it means or even whether it is valid
3988 assembler input. The extended @code{asm} feature is most often used for
3989 machine instructions the compiler itself does not know exist. If
3990 the output expression cannot be directly addressed (for example, it is a
3991 bit-field), your constraint must allow a register. In that case, GCC
3992 will use the register as the output of the @code{asm}, and then store
3993 that register into the output.
3995 The ordinary output operands must be write-only; GCC will assume that
3996 the values in these operands before the instruction are dead and need
3997 not be generated. Extended asm supports input-output or read-write
3998 operands. Use the constraint character @samp{+} to indicate such an
3999 operand and list it with the output operands. You should only use
4000 read-write operands when the constraints for the operand (or the
4001 operand in which only some of the bits are to be changed) allow a
4004 You may, as an alternative, logically split its function into two
4005 separate operands, one input operand and one write-only output
4006 operand. The connection between them is expressed by constraints
4007 which say they need to be in the same location when the instruction
4008 executes. You can use the same C expression for both operands, or
4009 different expressions. For example, here we write the (fictitious)
4010 @samp{combine} instruction with @code{bar} as its read-only source
4011 operand and @code{foo} as its read-write destination:
4014 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
4018 The constraint @samp{"0"} for operand 1 says that it must occupy the
4019 same location as operand 0. A number in constraint is allowed only in
4020 an input operand and it must refer to an output operand.
4022 Only a number in the constraint can guarantee that one operand will be in
4023 the same place as another. The mere fact that @code{foo} is the value
4024 of both operands is not enough to guarantee that they will be in the
4025 same place in the generated assembler code. The following would not
4029 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
4032 Various optimizations or reloading could cause operands 0 and 1 to be in
4033 different registers; GCC knows no reason not to do so. For example, the
4034 compiler might find a copy of the value of @code{foo} in one register and
4035 use it for operand 1, but generate the output operand 0 in a different
4036 register (copying it afterward to @code{foo}'s own address). Of course,
4037 since the register for operand 1 is not even mentioned in the assembler
4038 code, the result will not work, but GCC can't tell that.
4040 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
4041 the operand number for a matching constraint. For example:
4044 asm ("cmoveq %1,%2,%[result]"
4045 : [result] "=r"(result)
4046 : "r" (test), "r"(new), "[result]"(old));
4049 Sometimes you need to make an @code{asm} operand be a specific register,
4050 but there's no matching constraint letter for that register @emph{by
4051 itself}. To force the operand into that register, use a local variable
4052 for the operand and specify the register in the variable declaration.
4053 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
4054 register constraint letter that matches the register:
4057 register int *p1 asm ("r0") = @dots{};
4058 register int *p2 asm ("r1") = @dots{};
4059 register int *result asm ("r0");
4060 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4063 @anchor{Example of asm with clobbered asm reg}
4064 In the above example, beware that a register that is call-clobbered by
4065 the target ABI will be overwritten by any function call in the
4066 assignment, including library calls for arithmetic operators.
4067 Assuming it is a call-clobbered register, this may happen to @code{r0}
4068 above by the assignment to @code{p2}. If you have to use such a
4069 register, use temporary variables for expressions between the register
4074 register int *p1 asm ("r0") = @dots{};
4075 register int *p2 asm ("r1") = t1;
4076 register int *result asm ("r0");
4077 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4080 Some instructions clobber specific hard registers. To describe this,
4081 write a third colon after the input operands, followed by the names of
4082 the clobbered hard registers (given as strings). Here is a realistic
4083 example for the VAX:
4086 asm volatile ("movc3 %0,%1,%2"
4087 : /* @r{no outputs} */
4088 : "g" (from), "g" (to), "g" (count)
4089 : "r0", "r1", "r2", "r3", "r4", "r5");
4092 You may not write a clobber description in a way that overlaps with an
4093 input or output operand. For example, you may not have an operand
4094 describing a register class with one member if you mention that register
4095 in the clobber list. Variables declared to live in specific registers
4096 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
4097 have no part mentioned in the clobber description.
4098 There is no way for you to specify that an input
4099 operand is modified without also specifying it as an output
4100 operand. Note that if all the output operands you specify are for this
4101 purpose (and hence unused), you will then also need to specify
4102 @code{volatile} for the @code{asm} construct, as described below, to
4103 prevent GCC from deleting the @code{asm} statement as unused.
4105 If you refer to a particular hardware register from the assembler code,
4106 you will probably have to list the register after the third colon to
4107 tell the compiler the register's value is modified. In some assemblers,
4108 the register names begin with @samp{%}; to produce one @samp{%} in the
4109 assembler code, you must write @samp{%%} in the input.
4111 If your assembler instruction can alter the condition code register, add
4112 @samp{cc} to the list of clobbered registers. GCC on some machines
4113 represents the condition codes as a specific hardware register;
4114 @samp{cc} serves to name this register. On other machines, the
4115 condition code is handled differently, and specifying @samp{cc} has no
4116 effect. But it is valid no matter what the machine.
4118 If your assembler instructions access memory in an unpredictable
4119 fashion, add @samp{memory} to the list of clobbered registers. This
4120 will cause GCC to not keep memory values cached in registers across the
4121 assembler instruction and not optimize stores or loads to that memory.
4122 You will also want to add the @code{volatile} keyword if the memory
4123 affected is not listed in the inputs or outputs of the @code{asm}, as
4124 the @samp{memory} clobber does not count as a side-effect of the
4125 @code{asm}. If you know how large the accessed memory is, you can add
4126 it as input or output but if this is not known, you should add
4127 @samp{memory}. As an example, if you access ten bytes of a string, you
4128 can use a memory input like:
4131 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
4134 Note that in the following example the memory input is necessary,
4135 otherwise GCC might optimize the store to @code{x} away:
4142 asm ("magic stuff accessing an 'int' pointed to by '%1'"
4143 "=&d" (r) : "a" (y), "m" (*y));
4148 You can put multiple assembler instructions together in a single
4149 @code{asm} template, separated by the characters normally used in assembly
4150 code for the system. A combination that works in most places is a newline
4151 to break the line, plus a tab character to move to the instruction field
4152 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
4153 assembler allows semicolons as a line-breaking character. Note that some
4154 assembler dialects use semicolons to start a comment.
4155 The input operands are guaranteed not to use any of the clobbered
4156 registers, and neither will the output operands' addresses, so you can
4157 read and write the clobbered registers as many times as you like. Here
4158 is an example of multiple instructions in a template; it assumes the
4159 subroutine @code{_foo} accepts arguments in registers 9 and 10:
4162 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
4164 : "g" (from), "g" (to)
4168 Unless an output operand has the @samp{&} constraint modifier, GCC
4169 may allocate it in the same register as an unrelated input operand, on
4170 the assumption the inputs are consumed before the outputs are produced.
4171 This assumption may be false if the assembler code actually consists of
4172 more than one instruction. In such a case, use @samp{&} for each output
4173 operand that may not overlap an input. @xref{Modifiers}.
4175 If you want to test the condition code produced by an assembler
4176 instruction, you must include a branch and a label in the @code{asm}
4177 construct, as follows:
4180 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
4186 This assumes your assembler supports local labels, as the GNU assembler
4187 and most Unix assemblers do.
4189 Speaking of labels, jumps from one @code{asm} to another are not
4190 supported. The compiler's optimizers do not know about these jumps, and
4191 therefore they cannot take account of them when deciding how to
4194 @cindex macros containing @code{asm}
4195 Usually the most convenient way to use these @code{asm} instructions is to
4196 encapsulate them in macros that look like functions. For example,
4200 (@{ double __value, __arg = (x); \
4201 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
4206 Here the variable @code{__arg} is used to make sure that the instruction
4207 operates on a proper @code{double} value, and to accept only those
4208 arguments @code{x} which can convert automatically to a @code{double}.
4210 Another way to make sure the instruction operates on the correct data
4211 type is to use a cast in the @code{asm}. This is different from using a
4212 variable @code{__arg} in that it converts more different types. For
4213 example, if the desired type were @code{int}, casting the argument to
4214 @code{int} would accept a pointer with no complaint, while assigning the
4215 argument to an @code{int} variable named @code{__arg} would warn about
4216 using a pointer unless the caller explicitly casts it.
4218 If an @code{asm} has output operands, GCC assumes for optimization
4219 purposes the instruction has no side effects except to change the output
4220 operands. This does not mean instructions with a side effect cannot be
4221 used, but you must be careful, because the compiler may eliminate them
4222 if the output operands aren't used, or move them out of loops, or
4223 replace two with one if they constitute a common subexpression. Also,
4224 if your instruction does have a side effect on a variable that otherwise
4225 appears not to change, the old value of the variable may be reused later
4226 if it happens to be found in a register.
4228 You can prevent an @code{asm} instruction from being deleted
4229 by writing the keyword @code{volatile} after
4230 the @code{asm}. For example:
4233 #define get_and_set_priority(new) \
4235 asm volatile ("get_and_set_priority %0, %1" \
4236 : "=g" (__old) : "g" (new)); \
4241 The @code{volatile} keyword indicates that the instruction has
4242 important side-effects. GCC will not delete a volatile @code{asm} if
4243 it is reachable. (The instruction can still be deleted if GCC can
4244 prove that control-flow will never reach the location of the
4245 instruction.) Note that even a volatile @code{asm} instruction
4246 can be moved relative to other code, including across jump
4247 instructions. For example, on many targets there is a system
4248 register which can be set to control the rounding mode of
4249 floating point operations. You might try
4250 setting it with a volatile @code{asm}, like this PowerPC example:
4253 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
4258 This will not work reliably, as the compiler may move the addition back
4259 before the volatile @code{asm}. To make it work you need to add an
4260 artificial dependency to the @code{asm} referencing a variable in the code
4261 you don't want moved, for example:
4264 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
4268 Similarly, you can't expect a
4269 sequence of volatile @code{asm} instructions to remain perfectly
4270 consecutive. If you want consecutive output, use a single @code{asm}.
4271 Also, GCC will perform some optimizations across a volatile @code{asm}
4272 instruction; GCC does not ``forget everything'' when it encounters
4273 a volatile @code{asm} instruction the way some other compilers do.
4275 An @code{asm} instruction without any output operands will be treated
4276 identically to a volatile @code{asm} instruction.
4278 It is a natural idea to look for a way to give access to the condition
4279 code left by the assembler instruction. However, when we attempted to
4280 implement this, we found no way to make it work reliably. The problem
4281 is that output operands might need reloading, which would result in
4282 additional following ``store'' instructions. On most machines, these
4283 instructions would alter the condition code before there was time to
4284 test it. This problem doesn't arise for ordinary ``test'' and
4285 ``compare'' instructions because they don't have any output operands.
4287 For reasons similar to those described above, it is not possible to give
4288 an assembler instruction access to the condition code left by previous
4291 If you are writing a header file that should be includable in ISO C
4292 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
4295 @subsection Size of an @code{asm}
4297 Some targets require that GCC track the size of each instruction used in
4298 order to generate correct code. Because the final length of an
4299 @code{asm} is only known by the assembler, GCC must make an estimate as
4300 to how big it will be. The estimate is formed by counting the number of
4301 statements in the pattern of the @code{asm} and multiplying that by the
4302 length of the longest instruction on that processor. Statements in the
4303 @code{asm} are identified by newline characters and whatever statement
4304 separator characters are supported by the assembler; on most processors
4305 this is the `@code{;}' character.
4307 Normally, GCC's estimate is perfectly adequate to ensure that correct
4308 code is generated, but it is possible to confuse the compiler if you use
4309 pseudo instructions or assembler macros that expand into multiple real
4310 instructions or if you use assembler directives that expand to more
4311 space in the object file than would be needed for a single instruction.
4312 If this happens then the assembler will produce a diagnostic saying that
4313 a label is unreachable.
4315 @subsection i386 floating point asm operands
4317 There are several rules on the usage of stack-like regs in
4318 asm_operands insns. These rules apply only to the operands that are
4323 Given a set of input regs that die in an asm_operands, it is
4324 necessary to know which are implicitly popped by the asm, and
4325 which must be explicitly popped by gcc.
4327 An input reg that is implicitly popped by the asm must be
4328 explicitly clobbered, unless it is constrained to match an
4332 For any input reg that is implicitly popped by an asm, it is
4333 necessary to know how to adjust the stack to compensate for the pop.
4334 If any non-popped input is closer to the top of the reg-stack than
4335 the implicitly popped reg, it would not be possible to know what the
4336 stack looked like---it's not clear how the rest of the stack ``slides
4339 All implicitly popped input regs must be closer to the top of
4340 the reg-stack than any input that is not implicitly popped.
4342 It is possible that if an input dies in an insn, reload might
4343 use the input reg for an output reload. Consider this example:
4346 asm ("foo" : "=t" (a) : "f" (b));
4349 This asm says that input B is not popped by the asm, and that
4350 the asm pushes a result onto the reg-stack, i.e., the stack is one
4351 deeper after the asm than it was before. But, it is possible that
4352 reload will think that it can use the same reg for both the input and
4353 the output, if input B dies in this insn.
4355 If any input operand uses the @code{f} constraint, all output reg
4356 constraints must use the @code{&} earlyclobber.
4358 The asm above would be written as
4361 asm ("foo" : "=&t" (a) : "f" (b));
4365 Some operands need to be in particular places on the stack. All
4366 output operands fall in this category---there is no other way to
4367 know which regs the outputs appear in unless the user indicates
4368 this in the constraints.
4370 Output operands must specifically indicate which reg an output
4371 appears in after an asm. @code{=f} is not allowed: the operand
4372 constraints must select a class with a single reg.
4375 Output operands may not be ``inserted'' between existing stack regs.
4376 Since no 387 opcode uses a read/write operand, all output operands
4377 are dead before the asm_operands, and are pushed by the asm_operands.
4378 It makes no sense to push anywhere but the top of the reg-stack.
4380 Output operands must start at the top of the reg-stack: output
4381 operands may not ``skip'' a reg.
4384 Some asm statements may need extra stack space for internal
4385 calculations. This can be guaranteed by clobbering stack registers
4386 unrelated to the inputs and outputs.
4390 Here are a couple of reasonable asms to want to write. This asm
4391 takes one input, which is internally popped, and produces two outputs.
4394 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4397 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4398 and replaces them with one output. The user must code the @code{st(1)}
4399 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4402 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4408 @section Controlling Names Used in Assembler Code
4409 @cindex assembler names for identifiers
4410 @cindex names used in assembler code
4411 @cindex identifiers, names in assembler code
4413 You can specify the name to be used in the assembler code for a C
4414 function or variable by writing the @code{asm} (or @code{__asm__})
4415 keyword after the declarator as follows:
4418 int foo asm ("myfoo") = 2;
4422 This specifies that the name to be used for the variable @code{foo} in
4423 the assembler code should be @samp{myfoo} rather than the usual
4426 On systems where an underscore is normally prepended to the name of a C
4427 function or variable, this feature allows you to define names for the
4428 linker that do not start with an underscore.
4430 It does not make sense to use this feature with a non-static local
4431 variable since such variables do not have assembler names. If you are
4432 trying to put the variable in a particular register, see @ref{Explicit
4433 Reg Vars}. GCC presently accepts such code with a warning, but will
4434 probably be changed to issue an error, rather than a warning, in the
4437 You cannot use @code{asm} in this way in a function @emph{definition}; but
4438 you can get the same effect by writing a declaration for the function
4439 before its definition and putting @code{asm} there, like this:
4442 extern func () asm ("FUNC");
4449 It is up to you to make sure that the assembler names you choose do not
4450 conflict with any other assembler symbols. Also, you must not use a
4451 register name; that would produce completely invalid assembler code. GCC
4452 does not as yet have the ability to store static variables in registers.
4453 Perhaps that will be added.
4455 @node Explicit Reg Vars
4456 @section Variables in Specified Registers
4457 @cindex explicit register variables
4458 @cindex variables in specified registers
4459 @cindex specified registers
4460 @cindex registers, global allocation
4462 GNU C allows you to put a few global variables into specified hardware
4463 registers. You can also specify the register in which an ordinary
4464 register variable should be allocated.
4468 Global register variables reserve registers throughout the program.
4469 This may be useful in programs such as programming language
4470 interpreters which have a couple of global variables that are accessed
4474 Local register variables in specific registers do not reserve the
4475 registers, except at the point where they are used as input or output
4476 operands in an @code{asm} statement and the @code{asm} statement itself is
4477 not deleted. The compiler's data flow analysis is capable of determining
4478 where the specified registers contain live values, and where they are
4479 available for other uses. Stores into local register variables may be deleted
4480 when they appear to be dead according to dataflow analysis. References
4481 to local register variables may be deleted or moved or simplified.
4483 These local variables are sometimes convenient for use with the extended
4484 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4485 output of the assembler instruction directly into a particular register.
4486 (This will work provided the register you specify fits the constraints
4487 specified for that operand in the @code{asm}.)
4495 @node Global Reg Vars
4496 @subsection Defining Global Register Variables
4497 @cindex global register variables
4498 @cindex registers, global variables in
4500 You can define a global register variable in GNU C like this:
4503 register int *foo asm ("a5");
4507 Here @code{a5} is the name of the register which should be used. Choose a
4508 register which is normally saved and restored by function calls on your
4509 machine, so that library routines will not clobber it.
4511 Naturally the register name is cpu-dependent, so you would need to
4512 conditionalize your program according to cpu type. The register
4513 @code{a5} would be a good choice on a 68000 for a variable of pointer
4514 type. On machines with register windows, be sure to choose a ``global''
4515 register that is not affected magically by the function call mechanism.
4517 In addition, operating systems on one type of cpu may differ in how they
4518 name the registers; then you would need additional conditionals. For
4519 example, some 68000 operating systems call this register @code{%a5}.
4521 Eventually there may be a way of asking the compiler to choose a register
4522 automatically, but first we need to figure out how it should choose and
4523 how to enable you to guide the choice. No solution is evident.
4525 Defining a global register variable in a certain register reserves that
4526 register entirely for this use, at least within the current compilation.
4527 The register will not be allocated for any other purpose in the functions
4528 in the current compilation. The register will not be saved and restored by
4529 these functions. Stores into this register are never deleted even if they
4530 would appear to be dead, but references may be deleted or moved or
4533 It is not safe to access the global register variables from signal
4534 handlers, or from more than one thread of control, because the system
4535 library routines may temporarily use the register for other things (unless
4536 you recompile them specially for the task at hand).
4538 @cindex @code{qsort}, and global register variables
4539 It is not safe for one function that uses a global register variable to
4540 call another such function @code{foo} by way of a third function
4541 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4542 different source file in which the variable wasn't declared). This is
4543 because @code{lose} might save the register and put some other value there.
4544 For example, you can't expect a global register variable to be available in
4545 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4546 might have put something else in that register. (If you are prepared to
4547 recompile @code{qsort} with the same global register variable, you can
4548 solve this problem.)
4550 If you want to recompile @code{qsort} or other source files which do not
4551 actually use your global register variable, so that they will not use that
4552 register for any other purpose, then it suffices to specify the compiler
4553 option @option{-ffixed-@var{reg}}. You need not actually add a global
4554 register declaration to their source code.
4556 A function which can alter the value of a global register variable cannot
4557 safely be called from a function compiled without this variable, because it
4558 could clobber the value the caller expects to find there on return.
4559 Therefore, the function which is the entry point into the part of the
4560 program that uses the global register variable must explicitly save and
4561 restore the value which belongs to its caller.
4563 @cindex register variable after @code{longjmp}
4564 @cindex global register after @code{longjmp}
4565 @cindex value after @code{longjmp}
4568 On most machines, @code{longjmp} will restore to each global register
4569 variable the value it had at the time of the @code{setjmp}. On some
4570 machines, however, @code{longjmp} will not change the value of global
4571 register variables. To be portable, the function that called @code{setjmp}
4572 should make other arrangements to save the values of the global register
4573 variables, and to restore them in a @code{longjmp}. This way, the same
4574 thing will happen regardless of what @code{longjmp} does.
4576 All global register variable declarations must precede all function
4577 definitions. If such a declaration could appear after function
4578 definitions, the declaration would be too late to prevent the register from
4579 being used for other purposes in the preceding functions.
4581 Global register variables may not have initial values, because an
4582 executable file has no means to supply initial contents for a register.
4584 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4585 registers, but certain library functions, such as @code{getwd}, as well
4586 as the subroutines for division and remainder, modify g3 and g4. g1 and
4587 g2 are local temporaries.
4589 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4590 Of course, it will not do to use more than a few of those.
4592 @node Local Reg Vars
4593 @subsection Specifying Registers for Local Variables
4594 @cindex local variables, specifying registers
4595 @cindex specifying registers for local variables
4596 @cindex registers for local variables
4598 You can define a local register variable with a specified register
4602 register int *foo asm ("a5");
4606 Here @code{a5} is the name of the register which should be used. Note
4607 that this is the same syntax used for defining global register
4608 variables, but for a local variable it would appear within a function.
4610 Naturally the register name is cpu-dependent, but this is not a
4611 problem, since specific registers are most often useful with explicit
4612 assembler instructions (@pxref{Extended Asm}). Both of these things
4613 generally require that you conditionalize your program according to
4616 In addition, operating systems on one type of cpu may differ in how they
4617 name the registers; then you would need additional conditionals. For
4618 example, some 68000 operating systems call this register @code{%a5}.
4620 Defining such a register variable does not reserve the register; it
4621 remains available for other uses in places where flow control determines
4622 the variable's value is not live.
4624 This option does not guarantee that GCC will generate code that has
4625 this variable in the register you specify at all times. You may not
4626 code an explicit reference to this register in the @emph{assembler
4627 instruction template} part of an @code{asm} statement and assume it will
4628 always refer to this variable. However, using the variable as an
4629 @code{asm} @emph{operand} guarantees that the specified register is used
4632 Stores into local register variables may be deleted when they appear to be dead
4633 according to dataflow analysis. References to local register variables may
4634 be deleted or moved or simplified.
4636 As for global register variables, it's recommended that you choose a
4637 register which is normally saved and restored by function calls on
4638 your machine, so that library routines will not clobber it. A common
4639 pitfall is to initialize multiple call-clobbered registers with
4640 arbitrary expressions, where a function call or library call for an
4641 arithmetic operator will overwrite a register value from a previous
4642 assignment, for example @code{r0} below:
4644 register int *p1 asm ("r0") = @dots{};
4645 register int *p2 asm ("r1") = @dots{};
4647 In those cases, a solution is to use a temporary variable for
4648 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
4650 @node Alternate Keywords
4651 @section Alternate Keywords
4652 @cindex alternate keywords
4653 @cindex keywords, alternate
4655 @option{-ansi} and the various @option{-std} options disable certain
4656 keywords. This causes trouble when you want to use GNU C extensions, or
4657 a general-purpose header file that should be usable by all programs,
4658 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4659 @code{inline} are not available in programs compiled with
4660 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4661 program compiled with @option{-std=c99}). The ISO C99 keyword
4662 @code{restrict} is only available when @option{-std=gnu99} (which will
4663 eventually be the default) or @option{-std=c99} (or the equivalent
4664 @option{-std=iso9899:1999}) is used.
4666 The way to solve these problems is to put @samp{__} at the beginning and
4667 end of each problematical keyword. For example, use @code{__asm__}
4668 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4670 Other C compilers won't accept these alternative keywords; if you want to
4671 compile with another compiler, you can define the alternate keywords as
4672 macros to replace them with the customary keywords. It looks like this:
4680 @findex __extension__
4682 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4684 prevent such warnings within one expression by writing
4685 @code{__extension__} before the expression. @code{__extension__} has no
4686 effect aside from this.
4688 @node Incomplete Enums
4689 @section Incomplete @code{enum} Types
4691 You can define an @code{enum} tag without specifying its possible values.
4692 This results in an incomplete type, much like what you get if you write
4693 @code{struct foo} without describing the elements. A later declaration
4694 which does specify the possible values completes the type.
4696 You can't allocate variables or storage using the type while it is
4697 incomplete. However, you can work with pointers to that type.
4699 This extension may not be very useful, but it makes the handling of
4700 @code{enum} more consistent with the way @code{struct} and @code{union}
4703 This extension is not supported by GNU C++.
4705 @node Function Names
4706 @section Function Names as Strings
4707 @cindex @code{__func__} identifier
4708 @cindex @code{__FUNCTION__} identifier
4709 @cindex @code{__PRETTY_FUNCTION__} identifier
4711 GCC provides three magic variables which hold the name of the current
4712 function, as a string. The first of these is @code{__func__}, which
4713 is part of the C99 standard:
4716 The identifier @code{__func__} is implicitly declared by the translator
4717 as if, immediately following the opening brace of each function
4718 definition, the declaration
4721 static const char __func__[] = "function-name";
4724 appeared, where function-name is the name of the lexically-enclosing
4725 function. This name is the unadorned name of the function.
4728 @code{__FUNCTION__} is another name for @code{__func__}. Older
4729 versions of GCC recognize only this name. However, it is not
4730 standardized. For maximum portability, we recommend you use
4731 @code{__func__}, but provide a fallback definition with the
4735 #if __STDC_VERSION__ < 199901L
4737 # define __func__ __FUNCTION__
4739 # define __func__ "<unknown>"
4744 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4745 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4746 the type signature of the function as well as its bare name. For
4747 example, this program:
4751 extern int printf (char *, ...);
4758 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4759 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4777 __PRETTY_FUNCTION__ = void a::sub(int)
4780 These identifiers are not preprocessor macros. In GCC 3.3 and
4781 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4782 were treated as string literals; they could be used to initialize
4783 @code{char} arrays, and they could be concatenated with other string
4784 literals. GCC 3.4 and later treat them as variables, like
4785 @code{__func__}. In C++, @code{__FUNCTION__} and
4786 @code{__PRETTY_FUNCTION__} have always been variables.
4788 @node Return Address
4789 @section Getting the Return or Frame Address of a Function
4791 These functions may be used to get information about the callers of a
4794 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4795 This function returns the return address of the current function, or of
4796 one of its callers. The @var{level} argument is number of frames to
4797 scan up the call stack. A value of @code{0} yields the return address
4798 of the current function, a value of @code{1} yields the return address
4799 of the caller of the current function, and so forth. When inlining
4800 the expected behavior is that the function will return the address of
4801 the function that will be returned to. To work around this behavior use
4802 the @code{noinline} function attribute.
4804 The @var{level} argument must be a constant integer.
4806 On some machines it may be impossible to determine the return address of
4807 any function other than the current one; in such cases, or when the top
4808 of the stack has been reached, this function will return @code{0} or a
4809 random value. In addition, @code{__builtin_frame_address} may be used
4810 to determine if the top of the stack has been reached.
4812 This function should only be used with a nonzero argument for debugging
4816 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4817 This function is similar to @code{__builtin_return_address}, but it
4818 returns the address of the function frame rather than the return address
4819 of the function. Calling @code{__builtin_frame_address} with a value of
4820 @code{0} yields the frame address of the current function, a value of
4821 @code{1} yields the frame address of the caller of the current function,
4824 The frame is the area on the stack which holds local variables and saved
4825 registers. The frame address is normally the address of the first word
4826 pushed on to the stack by the function. However, the exact definition
4827 depends upon the processor and the calling convention. If the processor
4828 has a dedicated frame pointer register, and the function has a frame,
4829 then @code{__builtin_frame_address} will return the value of the frame
4832 On some machines it may be impossible to determine the frame address of
4833 any function other than the current one; in such cases, or when the top
4834 of the stack has been reached, this function will return @code{0} if
4835 the first frame pointer is properly initialized by the startup code.
4837 This function should only be used with a nonzero argument for debugging
4841 @node Vector Extensions
4842 @section Using vector instructions through built-in functions
4844 On some targets, the instruction set contains SIMD vector instructions that
4845 operate on multiple values contained in one large register at the same time.
4846 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4849 The first step in using these extensions is to provide the necessary data
4850 types. This should be done using an appropriate @code{typedef}:
4853 typedef int v4si __attribute__ ((vector_size (16)));
4856 The @code{int} type specifies the base type, while the attribute specifies
4857 the vector size for the variable, measured in bytes. For example, the
4858 declaration above causes the compiler to set the mode for the @code{v4si}
4859 type to be 16 bytes wide and divided into @code{int} sized units. For
4860 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4861 corresponding mode of @code{foo} will be @acronym{V4SI}.
4863 The @code{vector_size} attribute is only applicable to integral and
4864 float scalars, although arrays, pointers, and function return values
4865 are allowed in conjunction with this construct.
4867 All the basic integer types can be used as base types, both as signed
4868 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4869 @code{long long}. In addition, @code{float} and @code{double} can be
4870 used to build floating-point vector types.
4872 Specifying a combination that is not valid for the current architecture
4873 will cause GCC to synthesize the instructions using a narrower mode.
4874 For example, if you specify a variable of type @code{V4SI} and your
4875 architecture does not allow for this specific SIMD type, GCC will
4876 produce code that uses 4 @code{SIs}.
4878 The types defined in this manner can be used with a subset of normal C
4879 operations. Currently, GCC will allow using the following operators
4880 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4882 The operations behave like C++ @code{valarrays}. Addition is defined as
4883 the addition of the corresponding elements of the operands. For
4884 example, in the code below, each of the 4 elements in @var{a} will be
4885 added to the corresponding 4 elements in @var{b} and the resulting
4886 vector will be stored in @var{c}.
4889 typedef int v4si __attribute__ ((vector_size (16)));
4896 Subtraction, multiplication, division, and the logical operations
4897 operate in a similar manner. Likewise, the result of using the unary
4898 minus or complement operators on a vector type is a vector whose
4899 elements are the negative or complemented values of the corresponding
4900 elements in the operand.
4902 You can declare variables and use them in function calls and returns, as
4903 well as in assignments and some casts. You can specify a vector type as
4904 a return type for a function. Vector types can also be used as function
4905 arguments. It is possible to cast from one vector type to another,
4906 provided they are of the same size (in fact, you can also cast vectors
4907 to and from other datatypes of the same size).
4909 You cannot operate between vectors of different lengths or different
4910 signedness without a cast.
4912 A port that supports hardware vector operations, usually provides a set
4913 of built-in functions that can be used to operate on vectors. For
4914 example, a function to add two vectors and multiply the result by a
4915 third could look like this:
4918 v4si f (v4si a, v4si b, v4si c)
4920 v4si tmp = __builtin_addv4si (a, b);
4921 return __builtin_mulv4si (tmp, c);
4928 @findex __builtin_offsetof
4930 GCC implements for both C and C++ a syntactic extension to implement
4931 the @code{offsetof} macro.
4935 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
4937 offsetof_member_designator:
4939 | offsetof_member_designator "." @code{identifier}
4940 | offsetof_member_designator "[" @code{expr} "]"
4943 This extension is sufficient such that
4946 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
4949 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
4950 may be dependent. In either case, @var{member} may consist of a single
4951 identifier, or a sequence of member accesses and array references.
4953 @node Atomic Builtins
4954 @section Built-in functions for atomic memory access
4956 The following builtins are intended to be compatible with those described
4957 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
4958 section 7.4. As such, they depart from the normal GCC practice of using
4959 the ``__builtin_'' prefix, and further that they are overloaded such that
4960 they work on multiple types.
4962 The definition given in the Intel documentation allows only for the use of
4963 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
4964 counterparts. GCC will allow any integral scalar or pointer type that is
4965 1, 2, 4 or 8 bytes in length.
4967 Not all operations are supported by all target processors. If a particular
4968 operation cannot be implemented on the target processor, a warning will be
4969 generated and a call an external function will be generated. The external
4970 function will carry the same name as the builtin, with an additional suffix
4971 @samp{_@var{n}} where @var{n} is the size of the data type.
4973 @c ??? Should we have a mechanism to suppress this warning? This is almost
4974 @c useful for implementing the operation under the control of an external
4977 In most cases, these builtins are considered a @dfn{full barrier}. That is,
4978 no memory operand will be moved across the operation, either forward or
4979 backward. Further, instructions will be issued as necessary to prevent the
4980 processor from speculating loads across the operation and from queuing stores
4981 after the operation.
4983 All of the routines are are described in the Intel documentation to take
4984 ``an optional list of variables protected by the memory barrier''. It's
4985 not clear what is meant by that; it could mean that @emph{only} the
4986 following variables are protected, or it could mean that these variables
4987 should in addition be protected. At present GCC ignores this list and
4988 protects all variables which are globally accessible. If in the future
4989 we make some use of this list, an empty list will continue to mean all
4990 globally accessible variables.
4993 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
4994 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
4995 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
4996 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
4997 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
4998 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
4999 @findex __sync_fetch_and_add
5000 @findex __sync_fetch_and_sub
5001 @findex __sync_fetch_and_or
5002 @findex __sync_fetch_and_and
5003 @findex __sync_fetch_and_xor
5004 @findex __sync_fetch_and_nand
5005 These builtins perform the operation suggested by the name, and
5006 returns the value that had previously been in memory. That is,
5009 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
5010 @{ tmp = *ptr; *ptr = ~tmp & value; return tmp; @} // nand
5013 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
5014 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
5015 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
5016 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
5017 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
5018 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
5019 @findex __sync_add_and_fetch
5020 @findex __sync_sub_and_fetch
5021 @findex __sync_or_and_fetch
5022 @findex __sync_and_and_fetch
5023 @findex __sync_xor_and_fetch
5024 @findex __sync_nand_and_fetch
5025 These builtins perform the operation suggested by the name, and
5026 return the new value. That is,
5029 @{ *ptr @var{op}= value; return *ptr; @}
5030 @{ *ptr = ~*ptr & value; return *ptr; @} // nand
5033 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5034 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5035 @findex __sync_bool_compare_and_swap
5036 @findex __sync_val_compare_and_swap
5037 These builtins perform an atomic compare and swap. That is, if the current
5038 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
5041 The ``bool'' version returns true if the comparison is successful and
5042 @var{newval} was written. The ``val'' version returns the contents
5043 of @code{*@var{ptr}} before the operation.
5045 @item __sync_synchronize (...)
5046 @findex __sync_synchronize
5047 This builtin issues a full memory barrier.
5049 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
5050 @findex __sync_lock_test_and_set
5051 This builtin, as described by Intel, is not a traditional test-and-set
5052 operation, but rather an atomic exchange operation. It writes @var{value}
5053 into @code{*@var{ptr}}, and returns the previous contents of
5056 Many targets have only minimal support for such locks, and do not support
5057 a full exchange operation. In this case, a target may support reduced
5058 functionality here by which the @emph{only} valid value to store is the
5059 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
5060 is implementation defined.
5062 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
5063 This means that references after the builtin cannot move to (or be
5064 speculated to) before the builtin, but previous memory stores may not
5065 be globally visible yet, and previous memory loads may not yet be
5068 @item void __sync_lock_release (@var{type} *ptr, ...)
5069 @findex __sync_lock_release
5070 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
5071 Normally this means writing the constant 0 to @code{*@var{ptr}}.
5073 This builtin is not a full barrier, but rather a @dfn{release barrier}.
5074 This means that all previous memory stores are globally visible, and all
5075 previous memory loads have been satisfied, but following memory reads
5076 are not prevented from being speculated to before the barrier.
5079 @node Object Size Checking
5080 @section Object Size Checking Builtins
5081 @findex __builtin_object_size
5082 @findex __builtin___memcpy_chk
5083 @findex __builtin___mempcpy_chk
5084 @findex __builtin___memmove_chk
5085 @findex __builtin___memset_chk
5086 @findex __builtin___strcpy_chk
5087 @findex __builtin___stpcpy_chk
5088 @findex __builtin___strncpy_chk
5089 @findex __builtin___strcat_chk
5090 @findex __builtin___strncat_chk
5091 @findex __builtin___sprintf_chk
5092 @findex __builtin___snprintf_chk
5093 @findex __builtin___vsprintf_chk
5094 @findex __builtin___vsnprintf_chk
5095 @findex __builtin___printf_chk
5096 @findex __builtin___vprintf_chk
5097 @findex __builtin___fprintf_chk
5098 @findex __builtin___vfprintf_chk
5100 GCC implements a limited buffer overflow protection mechanism
5101 that can prevent some buffer overflow attacks.
5103 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
5104 is a built-in construct that returns a constant number of bytes from
5105 @var{ptr} to the end of the object @var{ptr} pointer points to
5106 (if known at compile time). @code{__builtin_object_size} never evaluates
5107 its arguments for side-effects. If there are any side-effects in them, it
5108 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5109 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
5110 point to and all of them are known at compile time, the returned number
5111 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
5112 0 and minimum if nonzero. If it is not possible to determine which objects
5113 @var{ptr} points to at compile time, @code{__builtin_object_size} should
5114 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5115 for @var{type} 2 or 3.
5117 @var{type} is an integer constant from 0 to 3. If the least significant
5118 bit is clear, objects are whole variables, if it is set, a closest
5119 surrounding subobject is considered the object a pointer points to.
5120 The second bit determines if maximum or minimum of remaining bytes
5124 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
5125 char *p = &var.buf1[1], *q = &var.b;
5127 /* Here the object p points to is var. */
5128 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
5129 /* The subobject p points to is var.buf1. */
5130 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
5131 /* The object q points to is var. */
5132 assert (__builtin_object_size (q, 0)
5133 == (char *) (&var + 1) - (char *) &var.b);
5134 /* The subobject q points to is var.b. */
5135 assert (__builtin_object_size (q, 1) == sizeof (var.b));
5139 There are built-in functions added for many common string operation
5140 functions, e.g. for @code{memcpy} @code{__builtin___memcpy_chk}
5141 built-in is provided. This built-in has an additional last argument,
5142 which is the number of bytes remaining in object the @var{dest}
5143 argument points to or @code{(size_t) -1} if the size is not known.
5145 The built-in functions are optimized into the normal string functions
5146 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
5147 it is known at compile time that the destination object will not
5148 be overflown. If the compiler can determine at compile time the
5149 object will be always overflown, it issues a warning.
5151 The intended use can be e.g.
5155 #define bos0(dest) __builtin_object_size (dest, 0)
5156 #define memcpy(dest, src, n) \
5157 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
5161 /* It is unknown what object p points to, so this is optimized
5162 into plain memcpy - no checking is possible. */
5163 memcpy (p, "abcde", n);
5164 /* Destination is known and length too. It is known at compile
5165 time there will be no overflow. */
5166 memcpy (&buf[5], "abcde", 5);
5167 /* Destination is known, but the length is not known at compile time.
5168 This will result in __memcpy_chk call that can check for overflow
5170 memcpy (&buf[5], "abcde", n);
5171 /* Destination is known and it is known at compile time there will
5172 be overflow. There will be a warning and __memcpy_chk call that
5173 will abort the program at runtime. */
5174 memcpy (&buf[6], "abcde", 5);
5177 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
5178 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
5179 @code{strcat} and @code{strncat}.
5181 There are also checking built-in functions for formatted output functions.
5183 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
5184 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5185 const char *fmt, ...);
5186 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
5188 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5189 const char *fmt, va_list ap);
5192 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
5193 etc. functions and can contain implementation specific flags on what
5194 additional security measures the checking function might take, such as
5195 handling @code{%n} differently.
5197 The @var{os} argument is the object size @var{s} points to, like in the
5198 other built-in functions. There is a small difference in the behavior
5199 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
5200 optimized into the non-checking functions only if @var{flag} is 0, otherwise
5201 the checking function is called with @var{os} argument set to
5204 In addition to this, there are checking built-in functions
5205 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
5206 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
5207 These have just one additional argument, @var{flag}, right before
5208 format string @var{fmt}. If the compiler is able to optimize them to
5209 @code{fputc} etc. functions, it will, otherwise the checking function
5210 should be called and the @var{flag} argument passed to it.
5212 @node Other Builtins
5213 @section Other built-in functions provided by GCC
5214 @cindex built-in functions
5215 @findex __builtin_isgreater
5216 @findex __builtin_isgreaterequal
5217 @findex __builtin_isless
5218 @findex __builtin_islessequal
5219 @findex __builtin_islessgreater
5220 @findex __builtin_isunordered
5221 @findex __builtin_powi
5222 @findex __builtin_powif
5223 @findex __builtin_powil
5381 @findex fprintf_unlocked
5383 @findex fputs_unlocked
5493 @findex printf_unlocked
5522 @findex significandf
5523 @findex significandl
5594 GCC provides a large number of built-in functions other than the ones
5595 mentioned above. Some of these are for internal use in the processing
5596 of exceptions or variable-length argument lists and will not be
5597 documented here because they may change from time to time; we do not
5598 recommend general use of these functions.
5600 The remaining functions are provided for optimization purposes.
5602 @opindex fno-builtin
5603 GCC includes built-in versions of many of the functions in the standard
5604 C library. The versions prefixed with @code{__builtin_} will always be
5605 treated as having the same meaning as the C library function even if you
5606 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
5607 Many of these functions are only optimized in certain cases; if they are
5608 not optimized in a particular case, a call to the library function will
5613 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
5614 @option{-std=c99}), the functions
5615 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
5616 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
5617 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
5618 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
5619 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
5620 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
5621 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
5622 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
5623 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
5624 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
5625 @code{significandf}, @code{significandl}, @code{significand},
5626 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
5627 @code{stpncpy}, @code{strcasecmp}, @code{strdup}, @code{strfmon},
5628 @code{strncasecmp}, @code{strndup}, @code{toascii}, @code{y0f},
5629 @code{y0l}, @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf},
5630 @code{ynl} and @code{yn}
5631 may be handled as built-in functions.
5632 All these functions have corresponding versions
5633 prefixed with @code{__builtin_}, which may be used even in strict C89
5636 The ISO C99 functions
5637 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
5638 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
5639 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
5640 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
5641 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
5642 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
5643 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
5644 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
5645 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
5646 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
5647 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
5648 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
5649 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
5650 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
5651 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
5652 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
5653 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
5654 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
5655 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
5656 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
5657 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
5658 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
5659 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
5660 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
5661 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
5662 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
5663 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
5664 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
5665 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
5666 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
5667 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
5668 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
5669 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
5670 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
5671 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
5672 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
5673 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
5674 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
5675 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
5676 are handled as built-in functions
5677 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5679 There are also built-in versions of the ISO C99 functions
5680 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5681 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5682 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5683 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5684 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5685 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5686 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5687 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5688 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5689 that are recognized in any mode since ISO C90 reserves these names for
5690 the purpose to which ISO C99 puts them. All these functions have
5691 corresponding versions prefixed with @code{__builtin_}.
5693 The ISO C94 functions
5694 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
5695 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
5696 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
5698 are handled as built-in functions
5699 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5701 The ISO C90 functions
5702 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5703 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5704 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5705 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
5706 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
5707 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
5708 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
5709 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
5710 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
5711 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
5712 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
5713 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
5714 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
5715 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
5716 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
5717 @code{vprintf} and @code{vsprintf}
5718 are all recognized as built-in functions unless
5719 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5720 is specified for an individual function). All of these functions have
5721 corresponding versions prefixed with @code{__builtin_}.
5723 GCC provides built-in versions of the ISO C99 floating point comparison
5724 macros that avoid raising exceptions for unordered operands. They have
5725 the same names as the standard macros ( @code{isgreater},
5726 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5727 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5728 prefixed. We intend for a library implementor to be able to simply
5729 @code{#define} each standard macro to its built-in equivalent.
5731 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
5733 You can use the built-in function @code{__builtin_types_compatible_p} to
5734 determine whether two types are the same.
5736 This built-in function returns 1 if the unqualified versions of the
5737 types @var{type1} and @var{type2} (which are types, not expressions) are
5738 compatible, 0 otherwise. The result of this built-in function can be
5739 used in integer constant expressions.
5741 This built-in function ignores top level qualifiers (e.g., @code{const},
5742 @code{volatile}). For example, @code{int} is equivalent to @code{const
5745 The type @code{int[]} and @code{int[5]} are compatible. On the other
5746 hand, @code{int} and @code{char *} are not compatible, even if the size
5747 of their types, on the particular architecture are the same. Also, the
5748 amount of pointer indirection is taken into account when determining
5749 similarity. Consequently, @code{short *} is not similar to
5750 @code{short **}. Furthermore, two types that are typedefed are
5751 considered compatible if their underlying types are compatible.
5753 An @code{enum} type is not considered to be compatible with another
5754 @code{enum} type even if both are compatible with the same integer
5755 type; this is what the C standard specifies.
5756 For example, @code{enum @{foo, bar@}} is not similar to
5757 @code{enum @{hot, dog@}}.
5759 You would typically use this function in code whose execution varies
5760 depending on the arguments' types. For example:
5765 typeof (x) tmp = (x); \
5766 if (__builtin_types_compatible_p (typeof (x), long double)) \
5767 tmp = foo_long_double (tmp); \
5768 else if (__builtin_types_compatible_p (typeof (x), double)) \
5769 tmp = foo_double (tmp); \
5770 else if (__builtin_types_compatible_p (typeof (x), float)) \
5771 tmp = foo_float (tmp); \
5778 @emph{Note:} This construct is only available for C@.
5782 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
5784 You can use the built-in function @code{__builtin_choose_expr} to
5785 evaluate code depending on the value of a constant expression. This
5786 built-in function returns @var{exp1} if @var{const_exp}, which is a
5787 constant expression that must be able to be determined at compile time,
5788 is nonzero. Otherwise it returns 0.
5790 This built-in function is analogous to the @samp{? :} operator in C,
5791 except that the expression returned has its type unaltered by promotion
5792 rules. Also, the built-in function does not evaluate the expression
5793 that was not chosen. For example, if @var{const_exp} evaluates to true,
5794 @var{exp2} is not evaluated even if it has side-effects.
5796 This built-in function can return an lvalue if the chosen argument is an
5799 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5800 type. Similarly, if @var{exp2} is returned, its return type is the same
5807 __builtin_choose_expr ( \
5808 __builtin_types_compatible_p (typeof (x), double), \
5810 __builtin_choose_expr ( \
5811 __builtin_types_compatible_p (typeof (x), float), \
5813 /* @r{The void expression results in a compile-time error} \
5814 @r{when assigning the result to something.} */ \
5818 @emph{Note:} This construct is only available for C@. Furthermore, the
5819 unused expression (@var{exp1} or @var{exp2} depending on the value of
5820 @var{const_exp}) may still generate syntax errors. This may change in
5825 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5826 You can use the built-in function @code{__builtin_constant_p} to
5827 determine if a value is known to be constant at compile-time and hence
5828 that GCC can perform constant-folding on expressions involving that
5829 value. The argument of the function is the value to test. The function
5830 returns the integer 1 if the argument is known to be a compile-time
5831 constant and 0 if it is not known to be a compile-time constant. A
5832 return of 0 does not indicate that the value is @emph{not} a constant,
5833 but merely that GCC cannot prove it is a constant with the specified
5834 value of the @option{-O} option.
5836 You would typically use this function in an embedded application where
5837 memory was a critical resource. If you have some complex calculation,
5838 you may want it to be folded if it involves constants, but need to call
5839 a function if it does not. For example:
5842 #define Scale_Value(X) \
5843 (__builtin_constant_p (X) \
5844 ? ((X) * SCALE + OFFSET) : Scale (X))
5847 You may use this built-in function in either a macro or an inline
5848 function. However, if you use it in an inlined function and pass an
5849 argument of the function as the argument to the built-in, GCC will
5850 never return 1 when you call the inline function with a string constant
5851 or compound literal (@pxref{Compound Literals}) and will not return 1
5852 when you pass a constant numeric value to the inline function unless you
5853 specify the @option{-O} option.
5855 You may also use @code{__builtin_constant_p} in initializers for static
5856 data. For instance, you can write
5859 static const int table[] = @{
5860 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5866 This is an acceptable initializer even if @var{EXPRESSION} is not a
5867 constant expression. GCC must be more conservative about evaluating the
5868 built-in in this case, because it has no opportunity to perform
5871 Previous versions of GCC did not accept this built-in in data
5872 initializers. The earliest version where it is completely safe is
5876 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5877 @opindex fprofile-arcs
5878 You may use @code{__builtin_expect} to provide the compiler with
5879 branch prediction information. In general, you should prefer to
5880 use actual profile feedback for this (@option{-fprofile-arcs}), as
5881 programmers are notoriously bad at predicting how their programs
5882 actually perform. However, there are applications in which this
5883 data is hard to collect.
5885 The return value is the value of @var{exp}, which should be an integral
5886 expression. The semantics of the built-in are that it is expected that
5887 @var{exp} == @var{c}. For example:
5890 if (__builtin_expect (x, 0))
5895 would indicate that we do not expect to call @code{foo}, since
5896 we expect @code{x} to be zero. Since you are limited to integral
5897 expressions for @var{exp}, you should use constructions such as
5900 if (__builtin_expect (ptr != NULL, 1))
5905 when testing pointer or floating-point values.
5908 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5909 This function is used to minimize cache-miss latency by moving data into
5910 a cache before it is accessed.
5911 You can insert calls to @code{__builtin_prefetch} into code for which
5912 you know addresses of data in memory that is likely to be accessed soon.
5913 If the target supports them, data prefetch instructions will be generated.
5914 If the prefetch is done early enough before the access then the data will
5915 be in the cache by the time it is accessed.
5917 The value of @var{addr} is the address of the memory to prefetch.
5918 There are two optional arguments, @var{rw} and @var{locality}.
5919 The value of @var{rw} is a compile-time constant one or zero; one
5920 means that the prefetch is preparing for a write to the memory address
5921 and zero, the default, means that the prefetch is preparing for a read.
5922 The value @var{locality} must be a compile-time constant integer between
5923 zero and three. A value of zero means that the data has no temporal
5924 locality, so it need not be left in the cache after the access. A value
5925 of three means that the data has a high degree of temporal locality and
5926 should be left in all levels of cache possible. Values of one and two
5927 mean, respectively, a low or moderate degree of temporal locality. The
5931 for (i = 0; i < n; i++)
5934 __builtin_prefetch (&a[i+j], 1, 1);
5935 __builtin_prefetch (&b[i+j], 0, 1);
5940 Data prefetch does not generate faults if @var{addr} is invalid, but
5941 the address expression itself must be valid. For example, a prefetch
5942 of @code{p->next} will not fault if @code{p->next} is not a valid
5943 address, but evaluation will fault if @code{p} is not a valid address.
5945 If the target does not support data prefetch, the address expression
5946 is evaluated if it includes side effects but no other code is generated
5947 and GCC does not issue a warning.
5950 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5951 Returns a positive infinity, if supported by the floating-point format,
5952 else @code{DBL_MAX}. This function is suitable for implementing the
5953 ISO C macro @code{HUGE_VAL}.
5956 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5957 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5960 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5961 Similar to @code{__builtin_huge_val}, except the return
5962 type is @code{long double}.
5965 @deftypefn {Built-in Function} double __builtin_inf (void)
5966 Similar to @code{__builtin_huge_val}, except a warning is generated
5967 if the target floating-point format does not support infinities.
5970 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
5971 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
5974 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
5975 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
5978 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
5979 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
5982 @deftypefn {Built-in Function} float __builtin_inff (void)
5983 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5984 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5987 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5988 Similar to @code{__builtin_inf}, except the return
5989 type is @code{long double}.
5992 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5993 This is an implementation of the ISO C99 function @code{nan}.
5995 Since ISO C99 defines this function in terms of @code{strtod}, which we
5996 do not implement, a description of the parsing is in order. The string
5997 is parsed as by @code{strtol}; that is, the base is recognized by
5998 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5999 in the significand such that the least significant bit of the number
6000 is at the least significant bit of the significand. The number is
6001 truncated to fit the significand field provided. The significand is
6002 forced to be a quiet NaN@.
6004 This function, if given a string literal all of which would have been
6005 consumed by strtol, is evaluated early enough that it is considered a
6006 compile-time constant.
6009 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
6010 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
6013 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
6014 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
6017 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
6018 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
6021 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
6022 Similar to @code{__builtin_nan}, except the return type is @code{float}.
6025 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
6026 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
6029 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
6030 Similar to @code{__builtin_nan}, except the significand is forced
6031 to be a signaling NaN@. The @code{nans} function is proposed by
6032 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
6035 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
6036 Similar to @code{__builtin_nans}, except the return type is @code{float}.
6039 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
6040 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
6043 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
6044 Returns one plus the index of the least significant 1-bit of @var{x}, or
6045 if @var{x} is zero, returns zero.
6048 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
6049 Returns the number of leading 0-bits in @var{x}, starting at the most
6050 significant bit position. If @var{x} is 0, the result is undefined.
6053 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
6054 Returns the number of trailing 0-bits in @var{x}, starting at the least
6055 significant bit position. If @var{x} is 0, the result is undefined.
6058 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
6059 Returns the number of 1-bits in @var{x}.
6062 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
6063 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
6067 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
6068 Similar to @code{__builtin_ffs}, except the argument type is
6069 @code{unsigned long}.
6072 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
6073 Similar to @code{__builtin_clz}, except the argument type is
6074 @code{unsigned long}.
6077 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
6078 Similar to @code{__builtin_ctz}, except the argument type is
6079 @code{unsigned long}.
6082 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
6083 Similar to @code{__builtin_popcount}, except the argument type is
6084 @code{unsigned long}.
6087 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
6088 Similar to @code{__builtin_parity}, except the argument type is
6089 @code{unsigned long}.
6092 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
6093 Similar to @code{__builtin_ffs}, except the argument type is
6094 @code{unsigned long long}.
6097 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
6098 Similar to @code{__builtin_clz}, except the argument type is
6099 @code{unsigned long long}.
6102 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
6103 Similar to @code{__builtin_ctz}, except the argument type is
6104 @code{unsigned long long}.
6107 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
6108 Similar to @code{__builtin_popcount}, except the argument type is
6109 @code{unsigned long long}.
6112 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
6113 Similar to @code{__builtin_parity}, except the argument type is
6114 @code{unsigned long long}.
6117 @deftypefn {Built-in Function} double __builtin_powi (double, int)
6118 Returns the first argument raised to the power of the second. Unlike the
6119 @code{pow} function no guarantees about precision and rounding are made.
6122 @deftypefn {Built-in Function} float __builtin_powif (float, int)
6123 Similar to @code{__builtin_powi}, except the argument and return types
6127 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
6128 Similar to @code{__builtin_powi}, except the argument and return types
6129 are @code{long double}.
6132 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
6133 Returns @var{x} with the order of the bytes reversed; for example,
6134 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
6138 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
6139 Similar to @code{__builtin_bswap32}, except the argument and return types
6143 @node Target Builtins
6144 @section Built-in Functions Specific to Particular Target Machines
6146 On some target machines, GCC supports many built-in functions specific
6147 to those machines. Generally these generate calls to specific machine
6148 instructions, but allow the compiler to schedule those calls.
6151 * Alpha Built-in Functions::
6152 * ARM Built-in Functions::
6153 * Blackfin Built-in Functions::
6154 * FR-V Built-in Functions::
6155 * X86 Built-in Functions::
6156 * MIPS DSP Built-in Functions::
6157 * MIPS Paired-Single Support::
6158 * PowerPC AltiVec Built-in Functions::
6159 * SPARC VIS Built-in Functions::
6162 @node Alpha Built-in Functions
6163 @subsection Alpha Built-in Functions
6165 These built-in functions are available for the Alpha family of
6166 processors, depending on the command-line switches used.
6168 The following built-in functions are always available. They
6169 all generate the machine instruction that is part of the name.
6172 long __builtin_alpha_implver (void)
6173 long __builtin_alpha_rpcc (void)
6174 long __builtin_alpha_amask (long)
6175 long __builtin_alpha_cmpbge (long, long)
6176 long __builtin_alpha_extbl (long, long)
6177 long __builtin_alpha_extwl (long, long)
6178 long __builtin_alpha_extll (long, long)
6179 long __builtin_alpha_extql (long, long)
6180 long __builtin_alpha_extwh (long, long)
6181 long __builtin_alpha_extlh (long, long)
6182 long __builtin_alpha_extqh (long, long)
6183 long __builtin_alpha_insbl (long, long)
6184 long __builtin_alpha_inswl (long, long)
6185 long __builtin_alpha_insll (long, long)
6186 long __builtin_alpha_insql (long, long)
6187 long __builtin_alpha_inswh (long, long)
6188 long __builtin_alpha_inslh (long, long)
6189 long __builtin_alpha_insqh (long, long)
6190 long __builtin_alpha_mskbl (long, long)
6191 long __builtin_alpha_mskwl (long, long)
6192 long __builtin_alpha_mskll (long, long)
6193 long __builtin_alpha_mskql (long, long)
6194 long __builtin_alpha_mskwh (long, long)
6195 long __builtin_alpha_msklh (long, long)
6196 long __builtin_alpha_mskqh (long, long)
6197 long __builtin_alpha_umulh (long, long)
6198 long __builtin_alpha_zap (long, long)
6199 long __builtin_alpha_zapnot (long, long)
6202 The following built-in functions are always with @option{-mmax}
6203 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
6204 later. They all generate the machine instruction that is part
6208 long __builtin_alpha_pklb (long)
6209 long __builtin_alpha_pkwb (long)
6210 long __builtin_alpha_unpkbl (long)
6211 long __builtin_alpha_unpkbw (long)
6212 long __builtin_alpha_minub8 (long, long)
6213 long __builtin_alpha_minsb8 (long, long)
6214 long __builtin_alpha_minuw4 (long, long)
6215 long __builtin_alpha_minsw4 (long, long)
6216 long __builtin_alpha_maxub8 (long, long)
6217 long __builtin_alpha_maxsb8 (long, long)
6218 long __builtin_alpha_maxuw4 (long, long)
6219 long __builtin_alpha_maxsw4 (long, long)
6220 long __builtin_alpha_perr (long, long)
6223 The following built-in functions are always with @option{-mcix}
6224 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
6225 later. They all generate the machine instruction that is part
6229 long __builtin_alpha_cttz (long)
6230 long __builtin_alpha_ctlz (long)
6231 long __builtin_alpha_ctpop (long)
6234 The following builtins are available on systems that use the OSF/1
6235 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
6236 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
6237 @code{rdval} and @code{wrval}.
6240 void *__builtin_thread_pointer (void)
6241 void __builtin_set_thread_pointer (void *)
6244 @node ARM Built-in Functions
6245 @subsection ARM Built-in Functions
6247 These built-in functions are available for the ARM family of
6248 processors, when the @option{-mcpu=iwmmxt} switch is used:
6251 typedef int v2si __attribute__ ((vector_size (8)));
6252 typedef short v4hi __attribute__ ((vector_size (8)));
6253 typedef char v8qi __attribute__ ((vector_size (8)));
6255 int __builtin_arm_getwcx (int)
6256 void __builtin_arm_setwcx (int, int)
6257 int __builtin_arm_textrmsb (v8qi, int)
6258 int __builtin_arm_textrmsh (v4hi, int)
6259 int __builtin_arm_textrmsw (v2si, int)
6260 int __builtin_arm_textrmub (v8qi, int)
6261 int __builtin_arm_textrmuh (v4hi, int)
6262 int __builtin_arm_textrmuw (v2si, int)
6263 v8qi __builtin_arm_tinsrb (v8qi, int)
6264 v4hi __builtin_arm_tinsrh (v4hi, int)
6265 v2si __builtin_arm_tinsrw (v2si, int)
6266 long long __builtin_arm_tmia (long long, int, int)
6267 long long __builtin_arm_tmiabb (long long, int, int)
6268 long long __builtin_arm_tmiabt (long long, int, int)
6269 long long __builtin_arm_tmiaph (long long, int, int)
6270 long long __builtin_arm_tmiatb (long long, int, int)
6271 long long __builtin_arm_tmiatt (long long, int, int)
6272 int __builtin_arm_tmovmskb (v8qi)
6273 int __builtin_arm_tmovmskh (v4hi)
6274 int __builtin_arm_tmovmskw (v2si)
6275 long long __builtin_arm_waccb (v8qi)
6276 long long __builtin_arm_wacch (v4hi)
6277 long long __builtin_arm_waccw (v2si)
6278 v8qi __builtin_arm_waddb (v8qi, v8qi)
6279 v8qi __builtin_arm_waddbss (v8qi, v8qi)
6280 v8qi __builtin_arm_waddbus (v8qi, v8qi)
6281 v4hi __builtin_arm_waddh (v4hi, v4hi)
6282 v4hi __builtin_arm_waddhss (v4hi, v4hi)
6283 v4hi __builtin_arm_waddhus (v4hi, v4hi)
6284 v2si __builtin_arm_waddw (v2si, v2si)
6285 v2si __builtin_arm_waddwss (v2si, v2si)
6286 v2si __builtin_arm_waddwus (v2si, v2si)
6287 v8qi __builtin_arm_walign (v8qi, v8qi, int)
6288 long long __builtin_arm_wand(long long, long long)
6289 long long __builtin_arm_wandn (long long, long long)
6290 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
6291 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
6292 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
6293 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
6294 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
6295 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
6296 v2si __builtin_arm_wcmpeqw (v2si, v2si)
6297 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
6298 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
6299 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
6300 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
6301 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
6302 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
6303 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
6304 long long __builtin_arm_wmacsz (v4hi, v4hi)
6305 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
6306 long long __builtin_arm_wmacuz (v4hi, v4hi)
6307 v4hi __builtin_arm_wmadds (v4hi, v4hi)
6308 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
6309 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
6310 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
6311 v2si __builtin_arm_wmaxsw (v2si, v2si)
6312 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
6313 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
6314 v2si __builtin_arm_wmaxuw (v2si, v2si)
6315 v8qi __builtin_arm_wminsb (v8qi, v8qi)
6316 v4hi __builtin_arm_wminsh (v4hi, v4hi)
6317 v2si __builtin_arm_wminsw (v2si, v2si)
6318 v8qi __builtin_arm_wminub (v8qi, v8qi)
6319 v4hi __builtin_arm_wminuh (v4hi, v4hi)
6320 v2si __builtin_arm_wminuw (v2si, v2si)
6321 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
6322 v4hi __builtin_arm_wmulul (v4hi, v4hi)
6323 v4hi __builtin_arm_wmulum (v4hi, v4hi)
6324 long long __builtin_arm_wor (long long, long long)
6325 v2si __builtin_arm_wpackdss (long long, long long)
6326 v2si __builtin_arm_wpackdus (long long, long long)
6327 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
6328 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
6329 v4hi __builtin_arm_wpackwss (v2si, v2si)
6330 v4hi __builtin_arm_wpackwus (v2si, v2si)
6331 long long __builtin_arm_wrord (long long, long long)
6332 long long __builtin_arm_wrordi (long long, int)
6333 v4hi __builtin_arm_wrorh (v4hi, long long)
6334 v4hi __builtin_arm_wrorhi (v4hi, int)
6335 v2si __builtin_arm_wrorw (v2si, long long)
6336 v2si __builtin_arm_wrorwi (v2si, int)
6337 v2si __builtin_arm_wsadb (v8qi, v8qi)
6338 v2si __builtin_arm_wsadbz (v8qi, v8qi)
6339 v2si __builtin_arm_wsadh (v4hi, v4hi)
6340 v2si __builtin_arm_wsadhz (v4hi, v4hi)
6341 v4hi __builtin_arm_wshufh (v4hi, int)
6342 long long __builtin_arm_wslld (long long, long long)
6343 long long __builtin_arm_wslldi (long long, int)
6344 v4hi __builtin_arm_wsllh (v4hi, long long)
6345 v4hi __builtin_arm_wsllhi (v4hi, int)
6346 v2si __builtin_arm_wsllw (v2si, long long)
6347 v2si __builtin_arm_wsllwi (v2si, int)
6348 long long __builtin_arm_wsrad (long long, long long)
6349 long long __builtin_arm_wsradi (long long, int)
6350 v4hi __builtin_arm_wsrah (v4hi, long long)
6351 v4hi __builtin_arm_wsrahi (v4hi, int)
6352 v2si __builtin_arm_wsraw (v2si, long long)
6353 v2si __builtin_arm_wsrawi (v2si, int)
6354 long long __builtin_arm_wsrld (long long, long long)
6355 long long __builtin_arm_wsrldi (long long, int)
6356 v4hi __builtin_arm_wsrlh (v4hi, long long)
6357 v4hi __builtin_arm_wsrlhi (v4hi, int)
6358 v2si __builtin_arm_wsrlw (v2si, long long)
6359 v2si __builtin_arm_wsrlwi (v2si, int)
6360 v8qi __builtin_arm_wsubb (v8qi, v8qi)
6361 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
6362 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
6363 v4hi __builtin_arm_wsubh (v4hi, v4hi)
6364 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
6365 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
6366 v2si __builtin_arm_wsubw (v2si, v2si)
6367 v2si __builtin_arm_wsubwss (v2si, v2si)
6368 v2si __builtin_arm_wsubwus (v2si, v2si)
6369 v4hi __builtin_arm_wunpckehsb (v8qi)
6370 v2si __builtin_arm_wunpckehsh (v4hi)
6371 long long __builtin_arm_wunpckehsw (v2si)
6372 v4hi __builtin_arm_wunpckehub (v8qi)
6373 v2si __builtin_arm_wunpckehuh (v4hi)
6374 long long __builtin_arm_wunpckehuw (v2si)
6375 v4hi __builtin_arm_wunpckelsb (v8qi)
6376 v2si __builtin_arm_wunpckelsh (v4hi)
6377 long long __builtin_arm_wunpckelsw (v2si)
6378 v4hi __builtin_arm_wunpckelub (v8qi)
6379 v2si __builtin_arm_wunpckeluh (v4hi)
6380 long long __builtin_arm_wunpckeluw (v2si)
6381 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
6382 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
6383 v2si __builtin_arm_wunpckihw (v2si, v2si)
6384 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
6385 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
6386 v2si __builtin_arm_wunpckilw (v2si, v2si)
6387 long long __builtin_arm_wxor (long long, long long)
6388 long long __builtin_arm_wzero ()
6391 @node Blackfin Built-in Functions
6392 @subsection Blackfin Built-in Functions
6394 Currently, there are two Blackfin-specific built-in functions. These are
6395 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
6396 using inline assembly; by using these built-in functions the compiler can
6397 automatically add workarounds for hardware errata involving these
6398 instructions. These functions are named as follows:
6401 void __builtin_bfin_csync (void)
6402 void __builtin_bfin_ssync (void)
6405 @node FR-V Built-in Functions
6406 @subsection FR-V Built-in Functions
6408 GCC provides many FR-V-specific built-in functions. In general,
6409 these functions are intended to be compatible with those described
6410 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
6411 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
6412 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
6413 pointer rather than by value.
6415 Most of the functions are named after specific FR-V instructions.
6416 Such functions are said to be ``directly mapped'' and are summarized
6417 here in tabular form.
6421 * Directly-mapped Integer Functions::
6422 * Directly-mapped Media Functions::
6423 * Raw read/write Functions::
6424 * Other Built-in Functions::
6427 @node Argument Types
6428 @subsubsection Argument Types
6430 The arguments to the built-in functions can be divided into three groups:
6431 register numbers, compile-time constants and run-time values. In order
6432 to make this classification clear at a glance, the arguments and return
6433 values are given the following pseudo types:
6435 @multitable @columnfractions .20 .30 .15 .35
6436 @item Pseudo type @tab Real C type @tab Constant? @tab Description
6437 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
6438 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
6439 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
6440 @item @code{uw2} @tab @code{unsigned long long} @tab No
6441 @tab an unsigned doubleword
6442 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
6443 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
6444 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
6445 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
6448 These pseudo types are not defined by GCC, they are simply a notational
6449 convenience used in this manual.
6451 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
6452 and @code{sw2} are evaluated at run time. They correspond to
6453 register operands in the underlying FR-V instructions.
6455 @code{const} arguments represent immediate operands in the underlying
6456 FR-V instructions. They must be compile-time constants.
6458 @code{acc} arguments are evaluated at compile time and specify the number
6459 of an accumulator register. For example, an @code{acc} argument of 2
6460 will select the ACC2 register.
6462 @code{iacc} arguments are similar to @code{acc} arguments but specify the
6463 number of an IACC register. See @pxref{Other Built-in Functions}
6466 @node Directly-mapped Integer Functions
6467 @subsubsection Directly-mapped Integer Functions
6469 The functions listed below map directly to FR-V I-type instructions.
6471 @multitable @columnfractions .45 .32 .23
6472 @item Function prototype @tab Example usage @tab Assembly output
6473 @item @code{sw1 __ADDSS (sw1, sw1)}
6474 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
6475 @tab @code{ADDSS @var{a},@var{b},@var{c}}
6476 @item @code{sw1 __SCAN (sw1, sw1)}
6477 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
6478 @tab @code{SCAN @var{a},@var{b},@var{c}}
6479 @item @code{sw1 __SCUTSS (sw1)}
6480 @tab @code{@var{b} = __SCUTSS (@var{a})}
6481 @tab @code{SCUTSS @var{a},@var{b}}
6482 @item @code{sw1 __SLASS (sw1, sw1)}
6483 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
6484 @tab @code{SLASS @var{a},@var{b},@var{c}}
6485 @item @code{void __SMASS (sw1, sw1)}
6486 @tab @code{__SMASS (@var{a}, @var{b})}
6487 @tab @code{SMASS @var{a},@var{b}}
6488 @item @code{void __SMSSS (sw1, sw1)}
6489 @tab @code{__SMSSS (@var{a}, @var{b})}
6490 @tab @code{SMSSS @var{a},@var{b}}
6491 @item @code{void __SMU (sw1, sw1)}
6492 @tab @code{__SMU (@var{a}, @var{b})}
6493 @tab @code{SMU @var{a},@var{b}}
6494 @item @code{sw2 __SMUL (sw1, sw1)}
6495 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
6496 @tab @code{SMUL @var{a},@var{b},@var{c}}
6497 @item @code{sw1 __SUBSS (sw1, sw1)}
6498 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
6499 @tab @code{SUBSS @var{a},@var{b},@var{c}}
6500 @item @code{uw2 __UMUL (uw1, uw1)}
6501 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
6502 @tab @code{UMUL @var{a},@var{b},@var{c}}
6505 @node Directly-mapped Media Functions
6506 @subsubsection Directly-mapped Media Functions
6508 The functions listed below map directly to FR-V M-type instructions.
6510 @multitable @columnfractions .45 .32 .23
6511 @item Function prototype @tab Example usage @tab Assembly output
6512 @item @code{uw1 __MABSHS (sw1)}
6513 @tab @code{@var{b} = __MABSHS (@var{a})}
6514 @tab @code{MABSHS @var{a},@var{b}}
6515 @item @code{void __MADDACCS (acc, acc)}
6516 @tab @code{__MADDACCS (@var{b}, @var{a})}
6517 @tab @code{MADDACCS @var{a},@var{b}}
6518 @item @code{sw1 __MADDHSS (sw1, sw1)}
6519 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
6520 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
6521 @item @code{uw1 __MADDHUS (uw1, uw1)}
6522 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
6523 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
6524 @item @code{uw1 __MAND (uw1, uw1)}
6525 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
6526 @tab @code{MAND @var{a},@var{b},@var{c}}
6527 @item @code{void __MASACCS (acc, acc)}
6528 @tab @code{__MASACCS (@var{b}, @var{a})}
6529 @tab @code{MASACCS @var{a},@var{b}}
6530 @item @code{uw1 __MAVEH (uw1, uw1)}
6531 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
6532 @tab @code{MAVEH @var{a},@var{b},@var{c}}
6533 @item @code{uw2 __MBTOH (uw1)}
6534 @tab @code{@var{b} = __MBTOH (@var{a})}
6535 @tab @code{MBTOH @var{a},@var{b}}
6536 @item @code{void __MBTOHE (uw1 *, uw1)}
6537 @tab @code{__MBTOHE (&@var{b}, @var{a})}
6538 @tab @code{MBTOHE @var{a},@var{b}}
6539 @item @code{void __MCLRACC (acc)}
6540 @tab @code{__MCLRACC (@var{a})}
6541 @tab @code{MCLRACC @var{a}}
6542 @item @code{void __MCLRACCA (void)}
6543 @tab @code{__MCLRACCA ()}
6544 @tab @code{MCLRACCA}
6545 @item @code{uw1 __Mcop1 (uw1, uw1)}
6546 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
6547 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
6548 @item @code{uw1 __Mcop2 (uw1, uw1)}
6549 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
6550 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
6551 @item @code{uw1 __MCPLHI (uw2, const)}
6552 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
6553 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
6554 @item @code{uw1 __MCPLI (uw2, const)}
6555 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
6556 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
6557 @item @code{void __MCPXIS (acc, sw1, sw1)}
6558 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
6559 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
6560 @item @code{void __MCPXIU (acc, uw1, uw1)}
6561 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
6562 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
6563 @item @code{void __MCPXRS (acc, sw1, sw1)}
6564 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
6565 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
6566 @item @code{void __MCPXRU (acc, uw1, uw1)}
6567 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
6568 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
6569 @item @code{uw1 __MCUT (acc, uw1)}
6570 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
6571 @tab @code{MCUT @var{a},@var{b},@var{c}}
6572 @item @code{uw1 __MCUTSS (acc, sw1)}
6573 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
6574 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
6575 @item @code{void __MDADDACCS (acc, acc)}
6576 @tab @code{__MDADDACCS (@var{b}, @var{a})}
6577 @tab @code{MDADDACCS @var{a},@var{b}}
6578 @item @code{void __MDASACCS (acc, acc)}
6579 @tab @code{__MDASACCS (@var{b}, @var{a})}
6580 @tab @code{MDASACCS @var{a},@var{b}}
6581 @item @code{uw2 __MDCUTSSI (acc, const)}
6582 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
6583 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
6584 @item @code{uw2 __MDPACKH (uw2, uw2)}
6585 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
6586 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
6587 @item @code{uw2 __MDROTLI (uw2, const)}
6588 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
6589 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
6590 @item @code{void __MDSUBACCS (acc, acc)}
6591 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
6592 @tab @code{MDSUBACCS @var{a},@var{b}}
6593 @item @code{void __MDUNPACKH (uw1 *, uw2)}
6594 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
6595 @tab @code{MDUNPACKH @var{a},@var{b}}
6596 @item @code{uw2 __MEXPDHD (uw1, const)}
6597 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
6598 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
6599 @item @code{uw1 __MEXPDHW (uw1, const)}
6600 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
6601 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
6602 @item @code{uw1 __MHDSETH (uw1, const)}
6603 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
6604 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
6605 @item @code{sw1 __MHDSETS (const)}
6606 @tab @code{@var{b} = __MHDSETS (@var{a})}
6607 @tab @code{MHDSETS #@var{a},@var{b}}
6608 @item @code{uw1 __MHSETHIH (uw1, const)}
6609 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
6610 @tab @code{MHSETHIH #@var{a},@var{b}}
6611 @item @code{sw1 __MHSETHIS (sw1, const)}
6612 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
6613 @tab @code{MHSETHIS #@var{a},@var{b}}
6614 @item @code{uw1 __MHSETLOH (uw1, const)}
6615 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
6616 @tab @code{MHSETLOH #@var{a},@var{b}}
6617 @item @code{sw1 __MHSETLOS (sw1, const)}
6618 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
6619 @tab @code{MHSETLOS #@var{a},@var{b}}
6620 @item @code{uw1 __MHTOB (uw2)}
6621 @tab @code{@var{b} = __MHTOB (@var{a})}
6622 @tab @code{MHTOB @var{a},@var{b}}
6623 @item @code{void __MMACHS (acc, sw1, sw1)}
6624 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
6625 @tab @code{MMACHS @var{a},@var{b},@var{c}}
6626 @item @code{void __MMACHU (acc, uw1, uw1)}
6627 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
6628 @tab @code{MMACHU @var{a},@var{b},@var{c}}
6629 @item @code{void __MMRDHS (acc, sw1, sw1)}
6630 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
6631 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
6632 @item @code{void __MMRDHU (acc, uw1, uw1)}
6633 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
6634 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
6635 @item @code{void __MMULHS (acc, sw1, sw1)}
6636 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
6637 @tab @code{MMULHS @var{a},@var{b},@var{c}}
6638 @item @code{void __MMULHU (acc, uw1, uw1)}
6639 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
6640 @tab @code{MMULHU @var{a},@var{b},@var{c}}
6641 @item @code{void __MMULXHS (acc, sw1, sw1)}
6642 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
6643 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
6644 @item @code{void __MMULXHU (acc, uw1, uw1)}
6645 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
6646 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
6647 @item @code{uw1 __MNOT (uw1)}
6648 @tab @code{@var{b} = __MNOT (@var{a})}
6649 @tab @code{MNOT @var{a},@var{b}}
6650 @item @code{uw1 __MOR (uw1, uw1)}
6651 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
6652 @tab @code{MOR @var{a},@var{b},@var{c}}
6653 @item @code{uw1 __MPACKH (uh, uh)}
6654 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
6655 @tab @code{MPACKH @var{a},@var{b},@var{c}}
6656 @item @code{sw2 __MQADDHSS (sw2, sw2)}
6657 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
6658 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
6659 @item @code{uw2 __MQADDHUS (uw2, uw2)}
6660 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
6661 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
6662 @item @code{void __MQCPXIS (acc, sw2, sw2)}
6663 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
6664 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
6665 @item @code{void __MQCPXIU (acc, uw2, uw2)}
6666 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
6667 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
6668 @item @code{void __MQCPXRS (acc, sw2, sw2)}
6669 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
6670 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
6671 @item @code{void __MQCPXRU (acc, uw2, uw2)}
6672 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
6673 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
6674 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
6675 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
6676 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
6677 @item @code{sw2 __MQLMTHS (sw2, sw2)}
6678 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
6679 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
6680 @item @code{void __MQMACHS (acc, sw2, sw2)}
6681 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
6682 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
6683 @item @code{void __MQMACHU (acc, uw2, uw2)}
6684 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
6685 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
6686 @item @code{void __MQMACXHS (acc, sw2, sw2)}
6687 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
6688 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
6689 @item @code{void __MQMULHS (acc, sw2, sw2)}
6690 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
6691 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
6692 @item @code{void __MQMULHU (acc, uw2, uw2)}
6693 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
6694 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
6695 @item @code{void __MQMULXHS (acc, sw2, sw2)}
6696 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
6697 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
6698 @item @code{void __MQMULXHU (acc, uw2, uw2)}
6699 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
6700 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
6701 @item @code{sw2 __MQSATHS (sw2, sw2)}
6702 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
6703 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
6704 @item @code{uw2 __MQSLLHI (uw2, int)}
6705 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
6706 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
6707 @item @code{sw2 __MQSRAHI (sw2, int)}
6708 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
6709 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
6710 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
6711 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
6712 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
6713 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
6714 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
6715 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
6716 @item @code{void __MQXMACHS (acc, sw2, sw2)}
6717 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
6718 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
6719 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
6720 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
6721 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
6722 @item @code{uw1 __MRDACC (acc)}
6723 @tab @code{@var{b} = __MRDACC (@var{a})}
6724 @tab @code{MRDACC @var{a},@var{b}}
6725 @item @code{uw1 __MRDACCG (acc)}
6726 @tab @code{@var{b} = __MRDACCG (@var{a})}
6727 @tab @code{MRDACCG @var{a},@var{b}}
6728 @item @code{uw1 __MROTLI (uw1, const)}
6729 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
6730 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
6731 @item @code{uw1 __MROTRI (uw1, const)}
6732 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
6733 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
6734 @item @code{sw1 __MSATHS (sw1, sw1)}
6735 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
6736 @tab @code{MSATHS @var{a},@var{b},@var{c}}
6737 @item @code{uw1 __MSATHU (uw1, uw1)}
6738 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
6739 @tab @code{MSATHU @var{a},@var{b},@var{c}}
6740 @item @code{uw1 __MSLLHI (uw1, const)}
6741 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
6742 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
6743 @item @code{sw1 __MSRAHI (sw1, const)}
6744 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
6745 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
6746 @item @code{uw1 __MSRLHI (uw1, const)}
6747 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
6748 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
6749 @item @code{void __MSUBACCS (acc, acc)}
6750 @tab @code{__MSUBACCS (@var{b}, @var{a})}
6751 @tab @code{MSUBACCS @var{a},@var{b}}
6752 @item @code{sw1 __MSUBHSS (sw1, sw1)}
6753 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
6754 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
6755 @item @code{uw1 __MSUBHUS (uw1, uw1)}
6756 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
6757 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
6758 @item @code{void __MTRAP (void)}
6759 @tab @code{__MTRAP ()}
6761 @item @code{uw2 __MUNPACKH (uw1)}
6762 @tab @code{@var{b} = __MUNPACKH (@var{a})}
6763 @tab @code{MUNPACKH @var{a},@var{b}}
6764 @item @code{uw1 __MWCUT (uw2, uw1)}
6765 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
6766 @tab @code{MWCUT @var{a},@var{b},@var{c}}
6767 @item @code{void __MWTACC (acc, uw1)}
6768 @tab @code{__MWTACC (@var{b}, @var{a})}
6769 @tab @code{MWTACC @var{a},@var{b}}
6770 @item @code{void __MWTACCG (acc, uw1)}
6771 @tab @code{__MWTACCG (@var{b}, @var{a})}
6772 @tab @code{MWTACCG @var{a},@var{b}}
6773 @item @code{uw1 __MXOR (uw1, uw1)}
6774 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
6775 @tab @code{MXOR @var{a},@var{b},@var{c}}
6778 @node Raw read/write Functions
6779 @subsubsection Raw read/write Functions
6781 This sections describes built-in functions related to read and write
6782 instructions to access memory. These functions generate
6783 @code{membar} instructions to flush the I/O load and stores where
6784 appropriate, as described in Fujitsu's manual described above.
6788 @item unsigned char __builtin_read8 (void *@var{data})
6789 @item unsigned short __builtin_read16 (void *@var{data})
6790 @item unsigned long __builtin_read32 (void *@var{data})
6791 @item unsigned long long __builtin_read64 (void *@var{data})
6793 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
6794 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
6795 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
6796 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
6799 @node Other Built-in Functions
6800 @subsubsection Other Built-in Functions
6802 This section describes built-in functions that are not named after
6803 a specific FR-V instruction.
6806 @item sw2 __IACCreadll (iacc @var{reg})
6807 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
6808 for future expansion and must be 0.
6810 @item sw1 __IACCreadl (iacc @var{reg})
6811 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
6812 Other values of @var{reg} are rejected as invalid.
6814 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
6815 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
6816 is reserved for future expansion and must be 0.
6818 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
6819 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
6820 is 1. Other values of @var{reg} are rejected as invalid.
6822 @item void __data_prefetch0 (const void *@var{x})
6823 Use the @code{dcpl} instruction to load the contents of address @var{x}
6824 into the data cache.
6826 @item void __data_prefetch (const void *@var{x})
6827 Use the @code{nldub} instruction to load the contents of address @var{x}
6828 into the data cache. The instruction will be issued in slot I1@.
6831 @node X86 Built-in Functions
6832 @subsection X86 Built-in Functions
6834 These built-in functions are available for the i386 and x86-64 family
6835 of computers, depending on the command-line switches used.
6837 Note that, if you specify command-line switches such as @option{-msse},
6838 the compiler could use the extended instruction sets even if the built-ins
6839 are not used explicitly in the program. For this reason, applications
6840 which perform runtime CPU detection must compile separate files for each
6841 supported architecture, using the appropriate flags. In particular,
6842 the file containing the CPU detection code should be compiled without
6845 The following machine modes are available for use with MMX built-in functions
6846 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
6847 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
6848 vector of eight 8-bit integers. Some of the built-in functions operate on
6849 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
6851 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
6852 of two 32-bit floating point values.
6854 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
6855 floating point values. Some instructions use a vector of four 32-bit
6856 integers, these use @code{V4SI}. Finally, some instructions operate on an
6857 entire vector register, interpreting it as a 128-bit integer, these use mode
6860 The following built-in functions are made available by @option{-mmmx}.
6861 All of them generate the machine instruction that is part of the name.
6864 v8qi __builtin_ia32_paddb (v8qi, v8qi)
6865 v4hi __builtin_ia32_paddw (v4hi, v4hi)
6866 v2si __builtin_ia32_paddd (v2si, v2si)
6867 v8qi __builtin_ia32_psubb (v8qi, v8qi)
6868 v4hi __builtin_ia32_psubw (v4hi, v4hi)
6869 v2si __builtin_ia32_psubd (v2si, v2si)
6870 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
6871 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
6872 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
6873 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
6874 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
6875 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
6876 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
6877 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
6878 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
6879 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
6880 di __builtin_ia32_pand (di, di)
6881 di __builtin_ia32_pandn (di,di)
6882 di __builtin_ia32_por (di, di)
6883 di __builtin_ia32_pxor (di, di)
6884 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
6885 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
6886 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
6887 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
6888 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
6889 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
6890 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
6891 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
6892 v2si __builtin_ia32_punpckhdq (v2si, v2si)
6893 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
6894 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
6895 v2si __builtin_ia32_punpckldq (v2si, v2si)
6896 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
6897 v4hi __builtin_ia32_packssdw (v2si, v2si)
6898 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
6901 The following built-in functions are made available either with
6902 @option{-msse}, or with a combination of @option{-m3dnow} and
6903 @option{-march=athlon}. All of them generate the machine
6904 instruction that is part of the name.
6907 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
6908 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
6909 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
6910 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
6911 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
6912 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
6913 v8qi __builtin_ia32_pminub (v8qi, v8qi)
6914 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
6915 int __builtin_ia32_pextrw (v4hi, int)
6916 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
6917 int __builtin_ia32_pmovmskb (v8qi)
6918 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
6919 void __builtin_ia32_movntq (di *, di)
6920 void __builtin_ia32_sfence (void)
6923 The following built-in functions are available when @option{-msse} is used.
6924 All of them generate the machine instruction that is part of the name.
6927 int __builtin_ia32_comieq (v4sf, v4sf)
6928 int __builtin_ia32_comineq (v4sf, v4sf)
6929 int __builtin_ia32_comilt (v4sf, v4sf)
6930 int __builtin_ia32_comile (v4sf, v4sf)
6931 int __builtin_ia32_comigt (v4sf, v4sf)
6932 int __builtin_ia32_comige (v4sf, v4sf)
6933 int __builtin_ia32_ucomieq (v4sf, v4sf)
6934 int __builtin_ia32_ucomineq (v4sf, v4sf)
6935 int __builtin_ia32_ucomilt (v4sf, v4sf)
6936 int __builtin_ia32_ucomile (v4sf, v4sf)
6937 int __builtin_ia32_ucomigt (v4sf, v4sf)
6938 int __builtin_ia32_ucomige (v4sf, v4sf)
6939 v4sf __builtin_ia32_addps (v4sf, v4sf)
6940 v4sf __builtin_ia32_subps (v4sf, v4sf)
6941 v4sf __builtin_ia32_mulps (v4sf, v4sf)
6942 v4sf __builtin_ia32_divps (v4sf, v4sf)
6943 v4sf __builtin_ia32_addss (v4sf, v4sf)
6944 v4sf __builtin_ia32_subss (v4sf, v4sf)
6945 v4sf __builtin_ia32_mulss (v4sf, v4sf)
6946 v4sf __builtin_ia32_divss (v4sf, v4sf)
6947 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
6948 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
6949 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
6950 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
6951 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
6952 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
6953 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
6954 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
6955 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
6956 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
6957 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
6958 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
6959 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
6960 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
6961 v4si __builtin_ia32_cmpless (v4sf, v4sf)
6962 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
6963 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
6964 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
6965 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
6966 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
6967 v4sf __builtin_ia32_maxps (v4sf, v4sf)
6968 v4sf __builtin_ia32_maxss (v4sf, v4sf)
6969 v4sf __builtin_ia32_minps (v4sf, v4sf)
6970 v4sf __builtin_ia32_minss (v4sf, v4sf)
6971 v4sf __builtin_ia32_andps (v4sf, v4sf)
6972 v4sf __builtin_ia32_andnps (v4sf, v4sf)
6973 v4sf __builtin_ia32_orps (v4sf, v4sf)
6974 v4sf __builtin_ia32_xorps (v4sf, v4sf)
6975 v4sf __builtin_ia32_movss (v4sf, v4sf)
6976 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
6977 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
6978 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
6979 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
6980 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
6981 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
6982 v2si __builtin_ia32_cvtps2pi (v4sf)
6983 int __builtin_ia32_cvtss2si (v4sf)
6984 v2si __builtin_ia32_cvttps2pi (v4sf)
6985 int __builtin_ia32_cvttss2si (v4sf)
6986 v4sf __builtin_ia32_rcpps (v4sf)
6987 v4sf __builtin_ia32_rsqrtps (v4sf)
6988 v4sf __builtin_ia32_sqrtps (v4sf)
6989 v4sf __builtin_ia32_rcpss (v4sf)
6990 v4sf __builtin_ia32_rsqrtss (v4sf)
6991 v4sf __builtin_ia32_sqrtss (v4sf)
6992 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
6993 void __builtin_ia32_movntps (float *, v4sf)
6994 int __builtin_ia32_movmskps (v4sf)
6997 The following built-in functions are available when @option{-msse} is used.
7000 @item v4sf __builtin_ia32_loadaps (float *)
7001 Generates the @code{movaps} machine instruction as a load from memory.
7002 @item void __builtin_ia32_storeaps (float *, v4sf)
7003 Generates the @code{movaps} machine instruction as a store to memory.
7004 @item v4sf __builtin_ia32_loadups (float *)
7005 Generates the @code{movups} machine instruction as a load from memory.
7006 @item void __builtin_ia32_storeups (float *, v4sf)
7007 Generates the @code{movups} machine instruction as a store to memory.
7008 @item v4sf __builtin_ia32_loadsss (float *)
7009 Generates the @code{movss} machine instruction as a load from memory.
7010 @item void __builtin_ia32_storess (float *, v4sf)
7011 Generates the @code{movss} machine instruction as a store to memory.
7012 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
7013 Generates the @code{movhps} machine instruction as a load from memory.
7014 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
7015 Generates the @code{movlps} machine instruction as a load from memory
7016 @item void __builtin_ia32_storehps (v4sf, v2si *)
7017 Generates the @code{movhps} machine instruction as a store to memory.
7018 @item void __builtin_ia32_storelps (v4sf, v2si *)
7019 Generates the @code{movlps} machine instruction as a store to memory.
7022 The following built-in functions are available when @option{-msse2} is used.
7023 All of them generate the machine instruction that is part of the name.
7026 int __builtin_ia32_comisdeq (v2df, v2df)
7027 int __builtin_ia32_comisdlt (v2df, v2df)
7028 int __builtin_ia32_comisdle (v2df, v2df)
7029 int __builtin_ia32_comisdgt (v2df, v2df)
7030 int __builtin_ia32_comisdge (v2df, v2df)
7031 int __builtin_ia32_comisdneq (v2df, v2df)
7032 int __builtin_ia32_ucomisdeq (v2df, v2df)
7033 int __builtin_ia32_ucomisdlt (v2df, v2df)
7034 int __builtin_ia32_ucomisdle (v2df, v2df)
7035 int __builtin_ia32_ucomisdgt (v2df, v2df)
7036 int __builtin_ia32_ucomisdge (v2df, v2df)
7037 int __builtin_ia32_ucomisdneq (v2df, v2df)
7038 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
7039 v2df __builtin_ia32_cmpltpd (v2df, v2df)
7040 v2df __builtin_ia32_cmplepd (v2df, v2df)
7041 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
7042 v2df __builtin_ia32_cmpgepd (v2df, v2df)
7043 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
7044 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
7045 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
7046 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
7047 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
7048 v2df __builtin_ia32_cmpngepd (v2df, v2df)
7049 v2df __builtin_ia32_cmpordpd (v2df, v2df)
7050 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
7051 v2df __builtin_ia32_cmpltsd (v2df, v2df)
7052 v2df __builtin_ia32_cmplesd (v2df, v2df)
7053 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
7054 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
7055 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
7056 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
7057 v2df __builtin_ia32_cmpordsd (v2df, v2df)
7058 v2di __builtin_ia32_paddq (v2di, v2di)
7059 v2di __builtin_ia32_psubq (v2di, v2di)
7060 v2df __builtin_ia32_addpd (v2df, v2df)
7061 v2df __builtin_ia32_subpd (v2df, v2df)
7062 v2df __builtin_ia32_mulpd (v2df, v2df)
7063 v2df __builtin_ia32_divpd (v2df, v2df)
7064 v2df __builtin_ia32_addsd (v2df, v2df)
7065 v2df __builtin_ia32_subsd (v2df, v2df)
7066 v2df __builtin_ia32_mulsd (v2df, v2df)
7067 v2df __builtin_ia32_divsd (v2df, v2df)
7068 v2df __builtin_ia32_minpd (v2df, v2df)
7069 v2df __builtin_ia32_maxpd (v2df, v2df)
7070 v2df __builtin_ia32_minsd (v2df, v2df)
7071 v2df __builtin_ia32_maxsd (v2df, v2df)
7072 v2df __builtin_ia32_andpd (v2df, v2df)
7073 v2df __builtin_ia32_andnpd (v2df, v2df)
7074 v2df __builtin_ia32_orpd (v2df, v2df)
7075 v2df __builtin_ia32_xorpd (v2df, v2df)
7076 v2df __builtin_ia32_movsd (v2df, v2df)
7077 v2df __builtin_ia32_unpckhpd (v2df, v2df)
7078 v2df __builtin_ia32_unpcklpd (v2df, v2df)
7079 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
7080 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
7081 v4si __builtin_ia32_paddd128 (v4si, v4si)
7082 v2di __builtin_ia32_paddq128 (v2di, v2di)
7083 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
7084 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
7085 v4si __builtin_ia32_psubd128 (v4si, v4si)
7086 v2di __builtin_ia32_psubq128 (v2di, v2di)
7087 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
7088 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
7089 v2di __builtin_ia32_pand128 (v2di, v2di)
7090 v2di __builtin_ia32_pandn128 (v2di, v2di)
7091 v2di __builtin_ia32_por128 (v2di, v2di)
7092 v2di __builtin_ia32_pxor128 (v2di, v2di)
7093 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
7094 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
7095 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
7096 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
7097 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
7098 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
7099 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
7100 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
7101 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
7102 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
7103 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
7104 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
7105 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
7106 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
7107 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
7108 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
7109 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
7110 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
7111 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
7112 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
7113 v16qi __builtin_ia32_packsswb128 (v16qi, v16qi)
7114 v8hi __builtin_ia32_packssdw128 (v8hi, v8hi)
7115 v16qi __builtin_ia32_packuswb128 (v16qi, v16qi)
7116 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
7117 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
7118 v2df __builtin_ia32_loadupd (double *)
7119 void __builtin_ia32_storeupd (double *, v2df)
7120 v2df __builtin_ia32_loadhpd (v2df, double *)
7121 v2df __builtin_ia32_loadlpd (v2df, double *)
7122 int __builtin_ia32_movmskpd (v2df)
7123 int __builtin_ia32_pmovmskb128 (v16qi)
7124 void __builtin_ia32_movnti (int *, int)
7125 void __builtin_ia32_movntpd (double *, v2df)
7126 void __builtin_ia32_movntdq (v2df *, v2df)
7127 v4si __builtin_ia32_pshufd (v4si, int)
7128 v8hi __builtin_ia32_pshuflw (v8hi, int)
7129 v8hi __builtin_ia32_pshufhw (v8hi, int)
7130 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
7131 v2df __builtin_ia32_sqrtpd (v2df)
7132 v2df __builtin_ia32_sqrtsd (v2df)
7133 v2df __builtin_ia32_shufpd (v2df, v2df, int)
7134 v2df __builtin_ia32_cvtdq2pd (v4si)
7135 v4sf __builtin_ia32_cvtdq2ps (v4si)
7136 v4si __builtin_ia32_cvtpd2dq (v2df)
7137 v2si __builtin_ia32_cvtpd2pi (v2df)
7138 v4sf __builtin_ia32_cvtpd2ps (v2df)
7139 v4si __builtin_ia32_cvttpd2dq (v2df)
7140 v2si __builtin_ia32_cvttpd2pi (v2df)
7141 v2df __builtin_ia32_cvtpi2pd (v2si)
7142 int __builtin_ia32_cvtsd2si (v2df)
7143 int __builtin_ia32_cvttsd2si (v2df)
7144 long long __builtin_ia32_cvtsd2si64 (v2df)
7145 long long __builtin_ia32_cvttsd2si64 (v2df)
7146 v4si __builtin_ia32_cvtps2dq (v4sf)
7147 v2df __builtin_ia32_cvtps2pd (v4sf)
7148 v4si __builtin_ia32_cvttps2dq (v4sf)
7149 v2df __builtin_ia32_cvtsi2sd (v2df, int)
7150 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
7151 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
7152 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
7153 void __builtin_ia32_clflush (const void *)
7154 void __builtin_ia32_lfence (void)
7155 void __builtin_ia32_mfence (void)
7156 v16qi __builtin_ia32_loaddqu (const char *)
7157 void __builtin_ia32_storedqu (char *, v16qi)
7158 unsigned long long __builtin_ia32_pmuludq (v2si, v2si)
7159 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
7160 v8hi __builtin_ia32_psllw128 (v8hi, v2di)
7161 v4si __builtin_ia32_pslld128 (v4si, v2di)
7162 v2di __builtin_ia32_psllq128 (v4si, v2di)
7163 v8hi __builtin_ia32_psrlw128 (v8hi, v2di)
7164 v4si __builtin_ia32_psrld128 (v4si, v2di)
7165 v2di __builtin_ia32_psrlq128 (v2di, v2di)
7166 v8hi __builtin_ia32_psraw128 (v8hi, v2di)
7167 v4si __builtin_ia32_psrad128 (v4si, v2di)
7168 v2di __builtin_ia32_pslldqi128 (v2di, int)
7169 v8hi __builtin_ia32_psllwi128 (v8hi, int)
7170 v4si __builtin_ia32_pslldi128 (v4si, int)
7171 v2di __builtin_ia32_psllqi128 (v2di, int)
7172 v2di __builtin_ia32_psrldqi128 (v2di, int)
7173 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
7174 v4si __builtin_ia32_psrldi128 (v4si, int)
7175 v2di __builtin_ia32_psrlqi128 (v2di, int)
7176 v8hi __builtin_ia32_psrawi128 (v8hi, int)
7177 v4si __builtin_ia32_psradi128 (v4si, int)
7178 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
7181 The following built-in functions are available when @option{-msse3} is used.
7182 All of them generate the machine instruction that is part of the name.
7185 v2df __builtin_ia32_addsubpd (v2df, v2df)
7186 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
7187 v2df __builtin_ia32_haddpd (v2df, v2df)
7188 v4sf __builtin_ia32_haddps (v4sf, v4sf)
7189 v2df __builtin_ia32_hsubpd (v2df, v2df)
7190 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
7191 v16qi __builtin_ia32_lddqu (char const *)
7192 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
7193 v2df __builtin_ia32_movddup (v2df)
7194 v4sf __builtin_ia32_movshdup (v4sf)
7195 v4sf __builtin_ia32_movsldup (v4sf)
7196 void __builtin_ia32_mwait (unsigned int, unsigned int)
7199 The following built-in functions are available when @option{-msse3} is used.
7202 @item v2df __builtin_ia32_loadddup (double const *)
7203 Generates the @code{movddup} machine instruction as a load from memory.
7206 The following built-in functions are available when @option{-mssse3} is used.
7207 All of them generate the machine instruction that is part of the name
7211 v2si __builtin_ia32_phaddd (v2si, v2si)
7212 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
7213 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
7214 v2si __builtin_ia32_phsubd (v2si, v2si)
7215 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
7216 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
7217 v8qi __builtin_ia32_pmaddubsw (v8qi, v8qi)
7218 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
7219 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
7220 v8qi __builtin_ia32_psignb (v8qi, v8qi)
7221 v2si __builtin_ia32_psignd (v2si, v2si)
7222 v4hi __builtin_ia32_psignw (v4hi, v4hi)
7223 long long __builtin_ia32_palignr (long long, long long, int)
7224 v8qi __builtin_ia32_pabsb (v8qi)
7225 v2si __builtin_ia32_pabsd (v2si)
7226 v4hi __builtin_ia32_pabsw (v4hi)
7229 The following built-in functions are available when @option{-mssse3} is used.
7230 All of them generate the machine instruction that is part of the name
7234 v4si __builtin_ia32_phaddd128 (v4si, v4si)
7235 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
7236 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
7237 v4si __builtin_ia32_phsubd128 (v4si, v4si)
7238 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
7239 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
7240 v16qi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
7241 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
7242 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
7243 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
7244 v4si __builtin_ia32_psignd128 (v4si, v4si)
7245 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
7246 v2di __builtin_ia32_palignr (v2di, v2di, int)
7247 v16qi __builtin_ia32_pabsb128 (v16qi)
7248 v4si __builtin_ia32_pabsd128 (v4si)
7249 v8hi __builtin_ia32_pabsw128 (v8hi)
7252 The following built-in functions are available when @option{-m3dnow} is used.
7253 All of them generate the machine instruction that is part of the name.
7256 void __builtin_ia32_femms (void)
7257 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
7258 v2si __builtin_ia32_pf2id (v2sf)
7259 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
7260 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
7261 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
7262 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
7263 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
7264 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
7265 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
7266 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
7267 v2sf __builtin_ia32_pfrcp (v2sf)
7268 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
7269 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
7270 v2sf __builtin_ia32_pfrsqrt (v2sf)
7271 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
7272 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
7273 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
7274 v2sf __builtin_ia32_pi2fd (v2si)
7275 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
7278 The following built-in functions are available when both @option{-m3dnow}
7279 and @option{-march=athlon} are used. All of them generate the machine
7280 instruction that is part of the name.
7283 v2si __builtin_ia32_pf2iw (v2sf)
7284 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
7285 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
7286 v2sf __builtin_ia32_pi2fw (v2si)
7287 v2sf __builtin_ia32_pswapdsf (v2sf)
7288 v2si __builtin_ia32_pswapdsi (v2si)
7291 @node MIPS DSP Built-in Functions
7292 @subsection MIPS DSP Built-in Functions
7294 The MIPS DSP Application-Specific Extension (ASE) includes new
7295 instructions that are designed to improve the performance of DSP and
7296 media applications. It provides instructions that operate on packed
7297 8-bit integer data, Q15 fractional data and Q31 fractional data.
7299 GCC supports MIPS DSP operations using both the generic
7300 vector extensions (@pxref{Vector Extensions}) and a collection of
7301 MIPS-specific built-in functions. Both kinds of support are
7302 enabled by the @option{-mdsp} command-line option.
7304 At present, GCC only provides support for operations on 32-bit
7305 vectors. The vector type associated with 8-bit integer data is
7306 usually called @code{v4i8} and the vector type associated with Q15 is
7307 usually called @code{v2q15}. They can be defined in C as follows:
7310 typedef char v4i8 __attribute__ ((vector_size(4)));
7311 typedef short v2q15 __attribute__ ((vector_size(4)));
7314 @code{v4i8} and @code{v2q15} values are initialized in the same way as
7315 aggregates. For example:
7318 v4i8 a = @{1, 2, 3, 4@};
7320 b = (v4i8) @{5, 6, 7, 8@};
7322 v2q15 c = @{0x0fcb, 0x3a75@};
7324 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
7327 @emph{Note:} The CPU's endianness determines the order in which values
7328 are packed. On little-endian targets, the first value is the least
7329 significant and the last value is the most significant. The opposite
7330 order applies to big-endian targets. For example, the code above will
7331 set the lowest byte of @code{a} to @code{1} on little-endian targets
7332 and @code{4} on big-endian targets.
7334 @emph{Note:} Q15 and Q31 values must be initialized with their integer
7335 representation. As shown in this example, the integer representation
7336 of a Q15 value can be obtained by multiplying the fractional value by
7337 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
7340 The table below lists the @code{v4i8} and @code{v2q15} operations for which
7341 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
7342 and @code{c} and @code{d} are @code{v2q15} values.
7344 @multitable @columnfractions .50 .50
7345 @item C code @tab MIPS instruction
7346 @item @code{a + b} @tab @code{addu.qb}
7347 @item @code{c + d} @tab @code{addq.ph}
7348 @item @code{a - b} @tab @code{subu.qb}
7349 @item @code{c - d} @tab @code{subq.ph}
7352 It is easier to describe the DSP built-in functions if we first define
7353 the following types:
7358 typedef long long a64;
7361 @code{q31} and @code{i32} are actually the same as @code{int}, but we
7362 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
7363 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
7364 @code{long long}, but we use @code{a64} to indicate values that will
7365 be placed in one of the four DSP accumulators (@code{$ac0},
7366 @code{$ac1}, @code{$ac2} or @code{$ac3}).
7368 Also, some built-in functions prefer or require immediate numbers as
7369 parameters, because the corresponding DSP instructions accept both immediate
7370 numbers and register operands, or accept immediate numbers only. The
7371 immediate parameters are listed as follows.
7379 imm_n32_31: -32 to 31.
7380 imm_n512_511: -512 to 511.
7383 The following built-in functions map directly to a particular MIPS DSP
7384 instruction. Please refer to the architecture specification
7385 for details on what each instruction does.
7388 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
7389 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
7390 q31 __builtin_mips_addq_s_w (q31, q31)
7391 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
7392 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
7393 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
7394 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
7395 q31 __builtin_mips_subq_s_w (q31, q31)
7396 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
7397 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
7398 i32 __builtin_mips_addsc (i32, i32)
7399 i32 __builtin_mips_addwc (i32, i32)
7400 i32 __builtin_mips_modsub (i32, i32)
7401 i32 __builtin_mips_raddu_w_qb (v4i8)
7402 v2q15 __builtin_mips_absq_s_ph (v2q15)
7403 q31 __builtin_mips_absq_s_w (q31)
7404 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
7405 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
7406 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
7407 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
7408 q31 __builtin_mips_preceq_w_phl (v2q15)
7409 q31 __builtin_mips_preceq_w_phr (v2q15)
7410 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
7411 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
7412 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
7413 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
7414 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
7415 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
7416 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
7417 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
7418 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
7419 v4i8 __builtin_mips_shll_qb (v4i8, i32)
7420 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
7421 v2q15 __builtin_mips_shll_ph (v2q15, i32)
7422 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
7423 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
7424 q31 __builtin_mips_shll_s_w (q31, imm0_31)
7425 q31 __builtin_mips_shll_s_w (q31, i32)
7426 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
7427 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
7428 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
7429 v2q15 __builtin_mips_shra_ph (v2q15, i32)
7430 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
7431 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
7432 q31 __builtin_mips_shra_r_w (q31, imm0_31)
7433 q31 __builtin_mips_shra_r_w (q31, i32)
7434 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
7435 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
7436 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
7437 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
7438 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
7439 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
7440 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
7441 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
7442 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
7443 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
7444 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
7445 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
7446 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
7447 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
7448 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
7449 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
7450 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
7451 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
7452 i32 __builtin_mips_bitrev (i32)
7453 i32 __builtin_mips_insv (i32, i32)
7454 v4i8 __builtin_mips_repl_qb (imm0_255)
7455 v4i8 __builtin_mips_repl_qb (i32)
7456 v2q15 __builtin_mips_repl_ph (imm_n512_511)
7457 v2q15 __builtin_mips_repl_ph (i32)
7458 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
7459 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
7460 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
7461 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
7462 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
7463 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
7464 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
7465 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
7466 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
7467 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
7468 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
7469 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
7470 i32 __builtin_mips_extr_w (a64, imm0_31)
7471 i32 __builtin_mips_extr_w (a64, i32)
7472 i32 __builtin_mips_extr_r_w (a64, imm0_31)
7473 i32 __builtin_mips_extr_s_h (a64, i32)
7474 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
7475 i32 __builtin_mips_extr_rs_w (a64, i32)
7476 i32 __builtin_mips_extr_s_h (a64, imm0_31)
7477 i32 __builtin_mips_extr_r_w (a64, i32)
7478 i32 __builtin_mips_extp (a64, imm0_31)
7479 i32 __builtin_mips_extp (a64, i32)
7480 i32 __builtin_mips_extpdp (a64, imm0_31)
7481 i32 __builtin_mips_extpdp (a64, i32)
7482 a64 __builtin_mips_shilo (a64, imm_n32_31)
7483 a64 __builtin_mips_shilo (a64, i32)
7484 a64 __builtin_mips_mthlip (a64, i32)
7485 void __builtin_mips_wrdsp (i32, imm0_63)
7486 i32 __builtin_mips_rddsp (imm0_63)
7487 i32 __builtin_mips_lbux (void *, i32)
7488 i32 __builtin_mips_lhx (void *, i32)
7489 i32 __builtin_mips_lwx (void *, i32)
7490 i32 __builtin_mips_bposge32 (void)
7493 @node MIPS Paired-Single Support
7494 @subsection MIPS Paired-Single Support
7496 The MIPS64 architecture includes a number of instructions that
7497 operate on pairs of single-precision floating-point values.
7498 Each pair is packed into a 64-bit floating-point register,
7499 with one element being designated the ``upper half'' and
7500 the other being designated the ``lower half''.
7502 GCC supports paired-single operations using both the generic
7503 vector extensions (@pxref{Vector Extensions}) and a collection of
7504 MIPS-specific built-in functions. Both kinds of support are
7505 enabled by the @option{-mpaired-single} command-line option.
7507 The vector type associated with paired-single values is usually
7508 called @code{v2sf}. It can be defined in C as follows:
7511 typedef float v2sf __attribute__ ((vector_size (8)));
7514 @code{v2sf} values are initialized in the same way as aggregates.
7518 v2sf a = @{1.5, 9.1@};
7521 b = (v2sf) @{e, f@};
7524 @emph{Note:} The CPU's endianness determines which value is stored in
7525 the upper half of a register and which value is stored in the lower half.
7526 On little-endian targets, the first value is the lower one and the second
7527 value is the upper one. The opposite order applies to big-endian targets.
7528 For example, the code above will set the lower half of @code{a} to
7529 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
7532 * Paired-Single Arithmetic::
7533 * Paired-Single Built-in Functions::
7534 * MIPS-3D Built-in Functions::
7537 @node Paired-Single Arithmetic
7538 @subsubsection Paired-Single Arithmetic
7540 The table below lists the @code{v2sf} operations for which hardware
7541 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
7542 values and @code{x} is an integral value.
7544 @multitable @columnfractions .50 .50
7545 @item C code @tab MIPS instruction
7546 @item @code{a + b} @tab @code{add.ps}
7547 @item @code{a - b} @tab @code{sub.ps}
7548 @item @code{-a} @tab @code{neg.ps}
7549 @item @code{a * b} @tab @code{mul.ps}
7550 @item @code{a * b + c} @tab @code{madd.ps}
7551 @item @code{a * b - c} @tab @code{msub.ps}
7552 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
7553 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
7554 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
7557 Note that the multiply-accumulate instructions can be disabled
7558 using the command-line option @code{-mno-fused-madd}.
7560 @node Paired-Single Built-in Functions
7561 @subsubsection Paired-Single Built-in Functions
7563 The following paired-single functions map directly to a particular
7564 MIPS instruction. Please refer to the architecture specification
7565 for details on what each instruction does.
7568 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
7569 Pair lower lower (@code{pll.ps}).
7571 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
7572 Pair upper lower (@code{pul.ps}).
7574 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
7575 Pair lower upper (@code{plu.ps}).
7577 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
7578 Pair upper upper (@code{puu.ps}).
7580 @item v2sf __builtin_mips_cvt_ps_s (float, float)
7581 Convert pair to paired single (@code{cvt.ps.s}).
7583 @item float __builtin_mips_cvt_s_pl (v2sf)
7584 Convert pair lower to single (@code{cvt.s.pl}).
7586 @item float __builtin_mips_cvt_s_pu (v2sf)
7587 Convert pair upper to single (@code{cvt.s.pu}).
7589 @item v2sf __builtin_mips_abs_ps (v2sf)
7590 Absolute value (@code{abs.ps}).
7592 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
7593 Align variable (@code{alnv.ps}).
7595 @emph{Note:} The value of the third parameter must be 0 or 4
7596 modulo 8, otherwise the result will be unpredictable. Please read the
7597 instruction description for details.
7600 The following multi-instruction functions are also available.
7601 In each case, @var{cond} can be any of the 16 floating-point conditions:
7602 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
7603 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
7604 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
7607 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7608 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7609 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
7610 @code{movt.ps}/@code{movf.ps}).
7612 The @code{movt} functions return the value @var{x} computed by:
7615 c.@var{cond}.ps @var{cc},@var{a},@var{b}
7616 mov.ps @var{x},@var{c}
7617 movt.ps @var{x},@var{d},@var{cc}
7620 The @code{movf} functions are similar but use @code{movf.ps} instead
7623 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7624 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7625 Comparison of two paired-single values (@code{c.@var{cond}.ps},
7626 @code{bc1t}/@code{bc1f}).
7628 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
7629 and return either the upper or lower half of the result. For example:
7633 if (__builtin_mips_upper_c_eq_ps (a, b))
7634 upper_halves_are_equal ();
7636 upper_halves_are_unequal ();
7638 if (__builtin_mips_lower_c_eq_ps (a, b))
7639 lower_halves_are_equal ();
7641 lower_halves_are_unequal ();
7645 @node MIPS-3D Built-in Functions
7646 @subsubsection MIPS-3D Built-in Functions
7648 The MIPS-3D Application-Specific Extension (ASE) includes additional
7649 paired-single instructions that are designed to improve the performance
7650 of 3D graphics operations. Support for these instructions is controlled
7651 by the @option{-mips3d} command-line option.
7653 The functions listed below map directly to a particular MIPS-3D
7654 instruction. Please refer to the architecture specification for
7655 more details on what each instruction does.
7658 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
7659 Reduction add (@code{addr.ps}).
7661 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
7662 Reduction multiply (@code{mulr.ps}).
7664 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
7665 Convert paired single to paired word (@code{cvt.pw.ps}).
7667 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
7668 Convert paired word to paired single (@code{cvt.ps.pw}).
7670 @item float __builtin_mips_recip1_s (float)
7671 @itemx double __builtin_mips_recip1_d (double)
7672 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
7673 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
7675 @item float __builtin_mips_recip2_s (float, float)
7676 @itemx double __builtin_mips_recip2_d (double, double)
7677 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
7678 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
7680 @item float __builtin_mips_rsqrt1_s (float)
7681 @itemx double __builtin_mips_rsqrt1_d (double)
7682 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
7683 Reduced precision reciprocal square root (sequence step 1)
7684 (@code{rsqrt1.@var{fmt}}).
7686 @item float __builtin_mips_rsqrt2_s (float, float)
7687 @itemx double __builtin_mips_rsqrt2_d (double, double)
7688 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
7689 Reduced precision reciprocal square root (sequence step 2)
7690 (@code{rsqrt2.@var{fmt}}).
7693 The following multi-instruction functions are also available.
7694 In each case, @var{cond} can be any of the 16 floating-point conditions:
7695 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
7696 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
7697 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
7700 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
7701 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
7702 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
7703 @code{bc1t}/@code{bc1f}).
7705 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
7706 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
7711 if (__builtin_mips_cabs_eq_s (a, b))
7717 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7718 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7719 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
7720 @code{bc1t}/@code{bc1f}).
7722 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
7723 and return either the upper or lower half of the result. For example:
7727 if (__builtin_mips_upper_cabs_eq_ps (a, b))
7728 upper_halves_are_equal ();
7730 upper_halves_are_unequal ();
7732 if (__builtin_mips_lower_cabs_eq_ps (a, b))
7733 lower_halves_are_equal ();
7735 lower_halves_are_unequal ();
7738 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7739 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7740 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
7741 @code{movt.ps}/@code{movf.ps}).
7743 The @code{movt} functions return the value @var{x} computed by:
7746 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
7747 mov.ps @var{x},@var{c}
7748 movt.ps @var{x},@var{d},@var{cc}
7751 The @code{movf} functions are similar but use @code{movf.ps} instead
7754 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7755 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7756 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7757 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7758 Comparison of two paired-single values
7759 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
7760 @code{bc1any2t}/@code{bc1any2f}).
7762 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
7763 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
7764 result is true and the @code{all} forms return true if both results are true.
7769 if (__builtin_mips_any_c_eq_ps (a, b))
7774 if (__builtin_mips_all_c_eq_ps (a, b))
7780 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7781 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7782 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7783 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7784 Comparison of four paired-single values
7785 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
7786 @code{bc1any4t}/@code{bc1any4f}).
7788 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
7789 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
7790 The @code{any} forms return true if any of the four results are true
7791 and the @code{all} forms return true if all four results are true.
7796 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
7801 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
7808 @node PowerPC AltiVec Built-in Functions
7809 @subsection PowerPC AltiVec Built-in Functions
7811 GCC provides an interface for the PowerPC family of processors to access
7812 the AltiVec operations described in Motorola's AltiVec Programming
7813 Interface Manual. The interface is made available by including
7814 @code{<altivec.h>} and using @option{-maltivec} and
7815 @option{-mabi=altivec}. The interface supports the following vector
7819 vector unsigned char
7823 vector unsigned short
7834 GCC's implementation of the high-level language interface available from
7835 C and C++ code differs from Motorola's documentation in several ways.
7840 A vector constant is a list of constant expressions within curly braces.
7843 A vector initializer requires no cast if the vector constant is of the
7844 same type as the variable it is initializing.
7847 If @code{signed} or @code{unsigned} is omitted, the signedness of the
7848 vector type is the default signedness of the base type. The default
7849 varies depending on the operating system, so a portable program should
7850 always specify the signedness.
7853 Compiling with @option{-maltivec} adds keywords @code{__vector},
7854 @code{__pixel}, and @code{__bool}. Macros @option{vector},
7855 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
7859 GCC allows using a @code{typedef} name as the type specifier for a
7863 For C, overloaded functions are implemented with macros so the following
7867 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
7870 Since @code{vec_add} is a macro, the vector constant in the example
7871 is treated as four separate arguments. Wrap the entire argument in
7872 parentheses for this to work.
7875 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
7876 Internally, GCC uses built-in functions to achieve the functionality in
7877 the aforementioned header file, but they are not supported and are
7878 subject to change without notice.
7880 The following interfaces are supported for the generic and specific
7881 AltiVec operations and the AltiVec predicates. In cases where there
7882 is a direct mapping between generic and specific operations, only the
7883 generic names are shown here, although the specific operations can also
7886 Arguments that are documented as @code{const int} require literal
7887 integral values within the range required for that operation.
7890 vector signed char vec_abs (vector signed char);
7891 vector signed short vec_abs (vector signed short);
7892 vector signed int vec_abs (vector signed int);
7893 vector float vec_abs (vector float);
7895 vector signed char vec_abss (vector signed char);
7896 vector signed short vec_abss (vector signed short);
7897 vector signed int vec_abss (vector signed int);
7899 vector signed char vec_add (vector bool char, vector signed char);
7900 vector signed char vec_add (vector signed char, vector bool char);
7901 vector signed char vec_add (vector signed char, vector signed char);
7902 vector unsigned char vec_add (vector bool char, vector unsigned char);
7903 vector unsigned char vec_add (vector unsigned char, vector bool char);
7904 vector unsigned char vec_add (vector unsigned char,
7905 vector unsigned char);
7906 vector signed short vec_add (vector bool short, vector signed short);
7907 vector signed short vec_add (vector signed short, vector bool short);
7908 vector signed short vec_add (vector signed short, vector signed short);
7909 vector unsigned short vec_add (vector bool short,
7910 vector unsigned short);
7911 vector unsigned short vec_add (vector unsigned short,
7913 vector unsigned short vec_add (vector unsigned short,
7914 vector unsigned short);
7915 vector signed int vec_add (vector bool int, vector signed int);
7916 vector signed int vec_add (vector signed int, vector bool int);
7917 vector signed int vec_add (vector signed int, vector signed int);
7918 vector unsigned int vec_add (vector bool int, vector unsigned int);
7919 vector unsigned int vec_add (vector unsigned int, vector bool int);
7920 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
7921 vector float vec_add (vector float, vector float);
7923 vector float vec_vaddfp (vector float, vector float);
7925 vector signed int vec_vadduwm (vector bool int, vector signed int);
7926 vector signed int vec_vadduwm (vector signed int, vector bool int);
7927 vector signed int vec_vadduwm (vector signed int, vector signed int);
7928 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
7929 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
7930 vector unsigned int vec_vadduwm (vector unsigned int,
7931 vector unsigned int);
7933 vector signed short vec_vadduhm (vector bool short,
7934 vector signed short);
7935 vector signed short vec_vadduhm (vector signed short,
7937 vector signed short vec_vadduhm (vector signed short,
7938 vector signed short);
7939 vector unsigned short vec_vadduhm (vector bool short,
7940 vector unsigned short);
7941 vector unsigned short vec_vadduhm (vector unsigned short,
7943 vector unsigned short vec_vadduhm (vector unsigned short,
7944 vector unsigned short);
7946 vector signed char vec_vaddubm (vector bool char, vector signed char);
7947 vector signed char vec_vaddubm (vector signed char, vector bool char);
7948 vector signed char vec_vaddubm (vector signed char, vector signed char);
7949 vector unsigned char vec_vaddubm (vector bool char,
7950 vector unsigned char);
7951 vector unsigned char vec_vaddubm (vector unsigned char,
7953 vector unsigned char vec_vaddubm (vector unsigned char,
7954 vector unsigned char);
7956 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
7958 vector unsigned char vec_adds (vector bool char, vector unsigned char);
7959 vector unsigned char vec_adds (vector unsigned char, vector bool char);
7960 vector unsigned char vec_adds (vector unsigned char,
7961 vector unsigned char);
7962 vector signed char vec_adds (vector bool char, vector signed char);
7963 vector signed char vec_adds (vector signed char, vector bool char);
7964 vector signed char vec_adds (vector signed char, vector signed char);
7965 vector unsigned short vec_adds (vector bool short,
7966 vector unsigned short);
7967 vector unsigned short vec_adds (vector unsigned short,
7969 vector unsigned short vec_adds (vector unsigned short,
7970 vector unsigned short);
7971 vector signed short vec_adds (vector bool short, vector signed short);
7972 vector signed short vec_adds (vector signed short, vector bool short);
7973 vector signed short vec_adds (vector signed short, vector signed short);
7974 vector unsigned int vec_adds (vector bool int, vector unsigned int);
7975 vector unsigned int vec_adds (vector unsigned int, vector bool int);
7976 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
7977 vector signed int vec_adds (vector bool int, vector signed int);
7978 vector signed int vec_adds (vector signed int, vector bool int);
7979 vector signed int vec_adds (vector signed int, vector signed int);
7981 vector signed int vec_vaddsws (vector bool int, vector signed int);
7982 vector signed int vec_vaddsws (vector signed int, vector bool int);
7983 vector signed int vec_vaddsws (vector signed int, vector signed int);
7985 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
7986 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
7987 vector unsigned int vec_vadduws (vector unsigned int,
7988 vector unsigned int);
7990 vector signed short vec_vaddshs (vector bool short,
7991 vector signed short);
7992 vector signed short vec_vaddshs (vector signed short,
7994 vector signed short vec_vaddshs (vector signed short,
7995 vector signed short);
7997 vector unsigned short vec_vadduhs (vector bool short,
7998 vector unsigned short);
7999 vector unsigned short vec_vadduhs (vector unsigned short,
8001 vector unsigned short vec_vadduhs (vector unsigned short,
8002 vector unsigned short);
8004 vector signed char vec_vaddsbs (vector bool char, vector signed char);
8005 vector signed char vec_vaddsbs (vector signed char, vector bool char);
8006 vector signed char vec_vaddsbs (vector signed char, vector signed char);
8008 vector unsigned char vec_vaddubs (vector bool char,
8009 vector unsigned char);
8010 vector unsigned char vec_vaddubs (vector unsigned char,
8012 vector unsigned char vec_vaddubs (vector unsigned char,
8013 vector unsigned char);
8015 vector float vec_and (vector float, vector float);
8016 vector float vec_and (vector float, vector bool int);
8017 vector float vec_and (vector bool int, vector float);
8018 vector bool int vec_and (vector bool int, vector bool int);
8019 vector signed int vec_and (vector bool int, vector signed int);
8020 vector signed int vec_and (vector signed int, vector bool int);
8021 vector signed int vec_and (vector signed int, vector signed int);
8022 vector unsigned int vec_and (vector bool int, vector unsigned int);
8023 vector unsigned int vec_and (vector unsigned int, vector bool int);
8024 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
8025 vector bool short vec_and (vector bool short, vector bool short);
8026 vector signed short vec_and (vector bool short, vector signed short);
8027 vector signed short vec_and (vector signed short, vector bool short);
8028 vector signed short vec_and (vector signed short, vector signed short);
8029 vector unsigned short vec_and (vector bool short,
8030 vector unsigned short);
8031 vector unsigned short vec_and (vector unsigned short,
8033 vector unsigned short vec_and (vector unsigned short,
8034 vector unsigned short);
8035 vector signed char vec_and (vector bool char, vector signed char);
8036 vector bool char vec_and (vector bool char, vector bool char);
8037 vector signed char vec_and (vector signed char, vector bool char);
8038 vector signed char vec_and (vector signed char, vector signed char);
8039 vector unsigned char vec_and (vector bool char, vector unsigned char);
8040 vector unsigned char vec_and (vector unsigned char, vector bool char);
8041 vector unsigned char vec_and (vector unsigned char,
8042 vector unsigned char);
8044 vector float vec_andc (vector float, vector float);
8045 vector float vec_andc (vector float, vector bool int);
8046 vector float vec_andc (vector bool int, vector float);
8047 vector bool int vec_andc (vector bool int, vector bool int);
8048 vector signed int vec_andc (vector bool int, vector signed int);
8049 vector signed int vec_andc (vector signed int, vector bool int);
8050 vector signed int vec_andc (vector signed int, vector signed int);
8051 vector unsigned int vec_andc (vector bool int, vector unsigned int);
8052 vector unsigned int vec_andc (vector unsigned int, vector bool int);
8053 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
8054 vector bool short vec_andc (vector bool short, vector bool short);
8055 vector signed short vec_andc (vector bool short, vector signed short);
8056 vector signed short vec_andc (vector signed short, vector bool short);
8057 vector signed short vec_andc (vector signed short, vector signed short);
8058 vector unsigned short vec_andc (vector bool short,
8059 vector unsigned short);
8060 vector unsigned short vec_andc (vector unsigned short,
8062 vector unsigned short vec_andc (vector unsigned short,
8063 vector unsigned short);
8064 vector signed char vec_andc (vector bool char, vector signed char);
8065 vector bool char vec_andc (vector bool char, vector bool char);
8066 vector signed char vec_andc (vector signed char, vector bool char);
8067 vector signed char vec_andc (vector signed char, vector signed char);
8068 vector unsigned char vec_andc (vector bool char, vector unsigned char);
8069 vector unsigned char vec_andc (vector unsigned char, vector bool char);
8070 vector unsigned char vec_andc (vector unsigned char,
8071 vector unsigned char);
8073 vector unsigned char vec_avg (vector unsigned char,
8074 vector unsigned char);
8075 vector signed char vec_avg (vector signed char, vector signed char);
8076 vector unsigned short vec_avg (vector unsigned short,
8077 vector unsigned short);
8078 vector signed short vec_avg (vector signed short, vector signed short);
8079 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
8080 vector signed int vec_avg (vector signed int, vector signed int);
8082 vector signed int vec_vavgsw (vector signed int, vector signed int);
8084 vector unsigned int vec_vavguw (vector unsigned int,
8085 vector unsigned int);
8087 vector signed short vec_vavgsh (vector signed short,
8088 vector signed short);
8090 vector unsigned short vec_vavguh (vector unsigned short,
8091 vector unsigned short);
8093 vector signed char vec_vavgsb (vector signed char, vector signed char);
8095 vector unsigned char vec_vavgub (vector unsigned char,
8096 vector unsigned char);
8098 vector float vec_ceil (vector float);
8100 vector signed int vec_cmpb (vector float, vector float);
8102 vector bool char vec_cmpeq (vector signed char, vector signed char);
8103 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
8104 vector bool short vec_cmpeq (vector signed short, vector signed short);
8105 vector bool short vec_cmpeq (vector unsigned short,
8106 vector unsigned short);
8107 vector bool int vec_cmpeq (vector signed int, vector signed int);
8108 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
8109 vector bool int vec_cmpeq (vector float, vector float);
8111 vector bool int vec_vcmpeqfp (vector float, vector float);
8113 vector bool int vec_vcmpequw (vector signed int, vector signed int);
8114 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
8116 vector bool short vec_vcmpequh (vector signed short,
8117 vector signed short);
8118 vector bool short vec_vcmpequh (vector unsigned short,
8119 vector unsigned short);
8121 vector bool char vec_vcmpequb (vector signed char, vector signed char);
8122 vector bool char vec_vcmpequb (vector unsigned char,
8123 vector unsigned char);
8125 vector bool int vec_cmpge (vector float, vector float);
8127 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
8128 vector bool char vec_cmpgt (vector signed char, vector signed char);
8129 vector bool short vec_cmpgt (vector unsigned short,
8130 vector unsigned short);
8131 vector bool short vec_cmpgt (vector signed short, vector signed short);
8132 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
8133 vector bool int vec_cmpgt (vector signed int, vector signed int);
8134 vector bool int vec_cmpgt (vector float, vector float);
8136 vector bool int vec_vcmpgtfp (vector float, vector float);
8138 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
8140 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
8142 vector bool short vec_vcmpgtsh (vector signed short,
8143 vector signed short);
8145 vector bool short vec_vcmpgtuh (vector unsigned short,
8146 vector unsigned short);
8148 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
8150 vector bool char vec_vcmpgtub (vector unsigned char,
8151 vector unsigned char);
8153 vector bool int vec_cmple (vector float, vector float);
8155 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
8156 vector bool char vec_cmplt (vector signed char, vector signed char);
8157 vector bool short vec_cmplt (vector unsigned short,
8158 vector unsigned short);
8159 vector bool short vec_cmplt (vector signed short, vector signed short);
8160 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
8161 vector bool int vec_cmplt (vector signed int, vector signed int);
8162 vector bool int vec_cmplt (vector float, vector float);
8164 vector float vec_ctf (vector unsigned int, const int);
8165 vector float vec_ctf (vector signed int, const int);
8167 vector float vec_vcfsx (vector signed int, const int);
8169 vector float vec_vcfux (vector unsigned int, const int);
8171 vector signed int vec_cts (vector float, const int);
8173 vector unsigned int vec_ctu (vector float, const int);
8175 void vec_dss (const int);
8177 void vec_dssall (void);
8179 void vec_dst (const vector unsigned char *, int, const int);
8180 void vec_dst (const vector signed char *, int, const int);
8181 void vec_dst (const vector bool char *, int, const int);
8182 void vec_dst (const vector unsigned short *, int, const int);
8183 void vec_dst (const vector signed short *, int, const int);
8184 void vec_dst (const vector bool short *, int, const int);
8185 void vec_dst (const vector pixel *, int, const int);
8186 void vec_dst (const vector unsigned int *, int, const int);
8187 void vec_dst (const vector signed int *, int, const int);
8188 void vec_dst (const vector bool int *, int, const int);
8189 void vec_dst (const vector float *, int, const int);
8190 void vec_dst (const unsigned char *, int, const int);
8191 void vec_dst (const signed char *, int, const int);
8192 void vec_dst (const unsigned short *, int, const int);
8193 void vec_dst (const short *, int, const int);
8194 void vec_dst (const unsigned int *, int, const int);
8195 void vec_dst (const int *, int, const int);
8196 void vec_dst (const unsigned long *, int, const int);
8197 void vec_dst (const long *, int, const int);
8198 void vec_dst (const float *, int, const int);
8200 void vec_dstst (const vector unsigned char *, int, const int);
8201 void vec_dstst (const vector signed char *, int, const int);
8202 void vec_dstst (const vector bool char *, int, const int);
8203 void vec_dstst (const vector unsigned short *, int, const int);
8204 void vec_dstst (const vector signed short *, int, const int);
8205 void vec_dstst (const vector bool short *, int, const int);
8206 void vec_dstst (const vector pixel *, int, const int);
8207 void vec_dstst (const vector unsigned int *, int, const int);
8208 void vec_dstst (const vector signed int *, int, const int);
8209 void vec_dstst (const vector bool int *, int, const int);
8210 void vec_dstst (const vector float *, int, const int);
8211 void vec_dstst (const unsigned char *, int, const int);
8212 void vec_dstst (const signed char *, int, const int);
8213 void vec_dstst (const unsigned short *, int, const int);
8214 void vec_dstst (const short *, int, const int);
8215 void vec_dstst (const unsigned int *, int, const int);
8216 void vec_dstst (const int *, int, const int);
8217 void vec_dstst (const unsigned long *, int, const int);
8218 void vec_dstst (const long *, int, const int);
8219 void vec_dstst (const float *, int, const int);
8221 void vec_dststt (const vector unsigned char *, int, const int);
8222 void vec_dststt (const vector signed char *, int, const int);
8223 void vec_dststt (const vector bool char *, int, const int);
8224 void vec_dststt (const vector unsigned short *, int, const int);
8225 void vec_dststt (const vector signed short *, int, const int);
8226 void vec_dststt (const vector bool short *, int, const int);
8227 void vec_dststt (const vector pixel *, int, const int);
8228 void vec_dststt (const vector unsigned int *, int, const int);
8229 void vec_dststt (const vector signed int *, int, const int);
8230 void vec_dststt (const vector bool int *, int, const int);
8231 void vec_dststt (const vector float *, int, const int);
8232 void vec_dststt (const unsigned char *, int, const int);
8233 void vec_dststt (const signed char *, int, const int);
8234 void vec_dststt (const unsigned short *, int, const int);
8235 void vec_dststt (const short *, int, const int);
8236 void vec_dststt (const unsigned int *, int, const int);
8237 void vec_dststt (const int *, int, const int);
8238 void vec_dststt (const unsigned long *, int, const int);
8239 void vec_dststt (const long *, int, const int);
8240 void vec_dststt (const float *, int, const int);
8242 void vec_dstt (const vector unsigned char *, int, const int);
8243 void vec_dstt (const vector signed char *, int, const int);
8244 void vec_dstt (const vector bool char *, int, const int);
8245 void vec_dstt (const vector unsigned short *, int, const int);
8246 void vec_dstt (const vector signed short *, int, const int);
8247 void vec_dstt (const vector bool short *, int, const int);
8248 void vec_dstt (const vector pixel *, int, const int);
8249 void vec_dstt (const vector unsigned int *, int, const int);
8250 void vec_dstt (const vector signed int *, int, const int);
8251 void vec_dstt (const vector bool int *, int, const int);
8252 void vec_dstt (const vector float *, int, const int);
8253 void vec_dstt (const unsigned char *, int, const int);
8254 void vec_dstt (const signed char *, int, const int);
8255 void vec_dstt (const unsigned short *, int, const int);
8256 void vec_dstt (const short *, int, const int);
8257 void vec_dstt (const unsigned int *, int, const int);
8258 void vec_dstt (const int *, int, const int);
8259 void vec_dstt (const unsigned long *, int, const int);
8260 void vec_dstt (const long *, int, const int);
8261 void vec_dstt (const float *, int, const int);
8263 vector float vec_expte (vector float);
8265 vector float vec_floor (vector float);
8267 vector float vec_ld (int, const vector float *);
8268 vector float vec_ld (int, const float *);
8269 vector bool int vec_ld (int, const vector bool int *);
8270 vector signed int vec_ld (int, const vector signed int *);
8271 vector signed int vec_ld (int, const int *);
8272 vector signed int vec_ld (int, const long *);
8273 vector unsigned int vec_ld (int, const vector unsigned int *);
8274 vector unsigned int vec_ld (int, const unsigned int *);
8275 vector unsigned int vec_ld (int, const unsigned long *);
8276 vector bool short vec_ld (int, const vector bool short *);
8277 vector pixel vec_ld (int, const vector pixel *);
8278 vector signed short vec_ld (int, const vector signed short *);
8279 vector signed short vec_ld (int, const short *);
8280 vector unsigned short vec_ld (int, const vector unsigned short *);
8281 vector unsigned short vec_ld (int, const unsigned short *);
8282 vector bool char vec_ld (int, const vector bool char *);
8283 vector signed char vec_ld (int, const vector signed char *);
8284 vector signed char vec_ld (int, const signed char *);
8285 vector unsigned char vec_ld (int, const vector unsigned char *);
8286 vector unsigned char vec_ld (int, const unsigned char *);
8288 vector signed char vec_lde (int, const signed char *);
8289 vector unsigned char vec_lde (int, const unsigned char *);
8290 vector signed short vec_lde (int, const short *);
8291 vector unsigned short vec_lde (int, const unsigned short *);
8292 vector float vec_lde (int, const float *);
8293 vector signed int vec_lde (int, const int *);
8294 vector unsigned int vec_lde (int, const unsigned int *);
8295 vector signed int vec_lde (int, const long *);
8296 vector unsigned int vec_lde (int, const unsigned long *);
8298 vector float vec_lvewx (int, float *);
8299 vector signed int vec_lvewx (int, int *);
8300 vector unsigned int vec_lvewx (int, unsigned int *);
8301 vector signed int vec_lvewx (int, long *);
8302 vector unsigned int vec_lvewx (int, unsigned long *);
8304 vector signed short vec_lvehx (int, short *);
8305 vector unsigned short vec_lvehx (int, unsigned short *);
8307 vector signed char vec_lvebx (int, char *);
8308 vector unsigned char vec_lvebx (int, unsigned char *);
8310 vector float vec_ldl (int, const vector float *);
8311 vector float vec_ldl (int, const float *);
8312 vector bool int vec_ldl (int, const vector bool int *);
8313 vector signed int vec_ldl (int, const vector signed int *);
8314 vector signed int vec_ldl (int, const int *);
8315 vector signed int vec_ldl (int, const long *);
8316 vector unsigned int vec_ldl (int, const vector unsigned int *);
8317 vector unsigned int vec_ldl (int, const unsigned int *);
8318 vector unsigned int vec_ldl (int, const unsigned long *);
8319 vector bool short vec_ldl (int, const vector bool short *);
8320 vector pixel vec_ldl (int, const vector pixel *);
8321 vector signed short vec_ldl (int, const vector signed short *);
8322 vector signed short vec_ldl (int, const short *);
8323 vector unsigned short vec_ldl (int, const vector unsigned short *);
8324 vector unsigned short vec_ldl (int, const unsigned short *);
8325 vector bool char vec_ldl (int, const vector bool char *);
8326 vector signed char vec_ldl (int, const vector signed char *);
8327 vector signed char vec_ldl (int, const signed char *);
8328 vector unsigned char vec_ldl (int, const vector unsigned char *);
8329 vector unsigned char vec_ldl (int, const unsigned char *);
8331 vector float vec_loge (vector float);
8333 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
8334 vector unsigned char vec_lvsl (int, const volatile signed char *);
8335 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
8336 vector unsigned char vec_lvsl (int, const volatile short *);
8337 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
8338 vector unsigned char vec_lvsl (int, const volatile int *);
8339 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
8340 vector unsigned char vec_lvsl (int, const volatile long *);
8341 vector unsigned char vec_lvsl (int, const volatile float *);
8343 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
8344 vector unsigned char vec_lvsr (int, const volatile signed char *);
8345 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
8346 vector unsigned char vec_lvsr (int, const volatile short *);
8347 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
8348 vector unsigned char vec_lvsr (int, const volatile int *);
8349 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
8350 vector unsigned char vec_lvsr (int, const volatile long *);
8351 vector unsigned char vec_lvsr (int, const volatile float *);
8353 vector float vec_madd (vector float, vector float, vector float);
8355 vector signed short vec_madds (vector signed short,
8356 vector signed short,
8357 vector signed short);
8359 vector unsigned char vec_max (vector bool char, vector unsigned char);
8360 vector unsigned char vec_max (vector unsigned char, vector bool char);
8361 vector unsigned char vec_max (vector unsigned char,
8362 vector unsigned char);
8363 vector signed char vec_max (vector bool char, vector signed char);
8364 vector signed char vec_max (vector signed char, vector bool char);
8365 vector signed char vec_max (vector signed char, vector signed char);
8366 vector unsigned short vec_max (vector bool short,
8367 vector unsigned short);
8368 vector unsigned short vec_max (vector unsigned short,
8370 vector unsigned short vec_max (vector unsigned short,
8371 vector unsigned short);
8372 vector signed short vec_max (vector bool short, vector signed short);
8373 vector signed short vec_max (vector signed short, vector bool short);
8374 vector signed short vec_max (vector signed short, vector signed short);
8375 vector unsigned int vec_max (vector bool int, vector unsigned int);
8376 vector unsigned int vec_max (vector unsigned int, vector bool int);
8377 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
8378 vector signed int vec_max (vector bool int, vector signed int);
8379 vector signed int vec_max (vector signed int, vector bool int);
8380 vector signed int vec_max (vector signed int, vector signed int);
8381 vector float vec_max (vector float, vector float);
8383 vector float vec_vmaxfp (vector float, vector float);
8385 vector signed int vec_vmaxsw (vector bool int, vector signed int);
8386 vector signed int vec_vmaxsw (vector signed int, vector bool int);
8387 vector signed int vec_vmaxsw (vector signed int, vector signed int);
8389 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
8390 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
8391 vector unsigned int vec_vmaxuw (vector unsigned int,
8392 vector unsigned int);
8394 vector signed short vec_vmaxsh (vector bool short, vector signed short);
8395 vector signed short vec_vmaxsh (vector signed short, vector bool short);
8396 vector signed short vec_vmaxsh (vector signed short,
8397 vector signed short);
8399 vector unsigned short vec_vmaxuh (vector bool short,
8400 vector unsigned short);
8401 vector unsigned short vec_vmaxuh (vector unsigned short,
8403 vector unsigned short vec_vmaxuh (vector unsigned short,
8404 vector unsigned short);
8406 vector signed char vec_vmaxsb (vector bool char, vector signed char);
8407 vector signed char vec_vmaxsb (vector signed char, vector bool char);
8408 vector signed char vec_vmaxsb (vector signed char, vector signed char);
8410 vector unsigned char vec_vmaxub (vector bool char,
8411 vector unsigned char);
8412 vector unsigned char vec_vmaxub (vector unsigned char,
8414 vector unsigned char vec_vmaxub (vector unsigned char,
8415 vector unsigned char);
8417 vector bool char vec_mergeh (vector bool char, vector bool char);
8418 vector signed char vec_mergeh (vector signed char, vector signed char);
8419 vector unsigned char vec_mergeh (vector unsigned char,
8420 vector unsigned char);
8421 vector bool short vec_mergeh (vector bool short, vector bool short);
8422 vector pixel vec_mergeh (vector pixel, vector pixel);
8423 vector signed short vec_mergeh (vector signed short,
8424 vector signed short);
8425 vector unsigned short vec_mergeh (vector unsigned short,
8426 vector unsigned short);
8427 vector float vec_mergeh (vector float, vector float);
8428 vector bool int vec_mergeh (vector bool int, vector bool int);
8429 vector signed int vec_mergeh (vector signed int, vector signed int);
8430 vector unsigned int vec_mergeh (vector unsigned int,
8431 vector unsigned int);
8433 vector float vec_vmrghw (vector float, vector float);
8434 vector bool int vec_vmrghw (vector bool int, vector bool int);
8435 vector signed int vec_vmrghw (vector signed int, vector signed int);
8436 vector unsigned int vec_vmrghw (vector unsigned int,
8437 vector unsigned int);
8439 vector bool short vec_vmrghh (vector bool short, vector bool short);
8440 vector signed short vec_vmrghh (vector signed short,
8441 vector signed short);
8442 vector unsigned short vec_vmrghh (vector unsigned short,
8443 vector unsigned short);
8444 vector pixel vec_vmrghh (vector pixel, vector pixel);
8446 vector bool char vec_vmrghb (vector bool char, vector bool char);
8447 vector signed char vec_vmrghb (vector signed char, vector signed char);
8448 vector unsigned char vec_vmrghb (vector unsigned char,
8449 vector unsigned char);
8451 vector bool char vec_mergel (vector bool char, vector bool char);
8452 vector signed char vec_mergel (vector signed char, vector signed char);
8453 vector unsigned char vec_mergel (vector unsigned char,
8454 vector unsigned char);
8455 vector bool short vec_mergel (vector bool short, vector bool short);
8456 vector pixel vec_mergel (vector pixel, vector pixel);
8457 vector signed short vec_mergel (vector signed short,
8458 vector signed short);
8459 vector unsigned short vec_mergel (vector unsigned short,
8460 vector unsigned short);
8461 vector float vec_mergel (vector float, vector float);
8462 vector bool int vec_mergel (vector bool int, vector bool int);
8463 vector signed int vec_mergel (vector signed int, vector signed int);
8464 vector unsigned int vec_mergel (vector unsigned int,
8465 vector unsigned int);
8467 vector float vec_vmrglw (vector float, vector float);
8468 vector signed int vec_vmrglw (vector signed int, vector signed int);
8469 vector unsigned int vec_vmrglw (vector unsigned int,
8470 vector unsigned int);
8471 vector bool int vec_vmrglw (vector bool int, vector bool int);
8473 vector bool short vec_vmrglh (vector bool short, vector bool short);
8474 vector signed short vec_vmrglh (vector signed short,
8475 vector signed short);
8476 vector unsigned short vec_vmrglh (vector unsigned short,
8477 vector unsigned short);
8478 vector pixel vec_vmrglh (vector pixel, vector pixel);
8480 vector bool char vec_vmrglb (vector bool char, vector bool char);
8481 vector signed char vec_vmrglb (vector signed char, vector signed char);
8482 vector unsigned char vec_vmrglb (vector unsigned char,
8483 vector unsigned char);
8485 vector unsigned short vec_mfvscr (void);
8487 vector unsigned char vec_min (vector bool char, vector unsigned char);
8488 vector unsigned char vec_min (vector unsigned char, vector bool char);
8489 vector unsigned char vec_min (vector unsigned char,
8490 vector unsigned char);
8491 vector signed char vec_min (vector bool char, vector signed char);
8492 vector signed char vec_min (vector signed char, vector bool char);
8493 vector signed char vec_min (vector signed char, vector signed char);
8494 vector unsigned short vec_min (vector bool short,
8495 vector unsigned short);
8496 vector unsigned short vec_min (vector unsigned short,
8498 vector unsigned short vec_min (vector unsigned short,
8499 vector unsigned short);
8500 vector signed short vec_min (vector bool short, vector signed short);
8501 vector signed short vec_min (vector signed short, vector bool short);
8502 vector signed short vec_min (vector signed short, vector signed short);
8503 vector unsigned int vec_min (vector bool int, vector unsigned int);
8504 vector unsigned int vec_min (vector unsigned int, vector bool int);
8505 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
8506 vector signed int vec_min (vector bool int, vector signed int);
8507 vector signed int vec_min (vector signed int, vector bool int);
8508 vector signed int vec_min (vector signed int, vector signed int);
8509 vector float vec_min (vector float, vector float);
8511 vector float vec_vminfp (vector float, vector float);
8513 vector signed int vec_vminsw (vector bool int, vector signed int);
8514 vector signed int vec_vminsw (vector signed int, vector bool int);
8515 vector signed int vec_vminsw (vector signed int, vector signed int);
8517 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
8518 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
8519 vector unsigned int vec_vminuw (vector unsigned int,
8520 vector unsigned int);
8522 vector signed short vec_vminsh (vector bool short, vector signed short);
8523 vector signed short vec_vminsh (vector signed short, vector bool short);
8524 vector signed short vec_vminsh (vector signed short,
8525 vector signed short);
8527 vector unsigned short vec_vminuh (vector bool short,
8528 vector unsigned short);
8529 vector unsigned short vec_vminuh (vector unsigned short,
8531 vector unsigned short vec_vminuh (vector unsigned short,
8532 vector unsigned short);
8534 vector signed char vec_vminsb (vector bool char, vector signed char);
8535 vector signed char vec_vminsb (vector signed char, vector bool char);
8536 vector signed char vec_vminsb (vector signed char, vector signed char);
8538 vector unsigned char vec_vminub (vector bool char,
8539 vector unsigned char);
8540 vector unsigned char vec_vminub (vector unsigned char,
8542 vector unsigned char vec_vminub (vector unsigned char,
8543 vector unsigned char);
8545 vector signed short vec_mladd (vector signed short,
8546 vector signed short,
8547 vector signed short);
8548 vector signed short vec_mladd (vector signed short,
8549 vector unsigned short,
8550 vector unsigned short);
8551 vector signed short vec_mladd (vector unsigned short,
8552 vector signed short,
8553 vector signed short);
8554 vector unsigned short vec_mladd (vector unsigned short,
8555 vector unsigned short,
8556 vector unsigned short);
8558 vector signed short vec_mradds (vector signed short,
8559 vector signed short,
8560 vector signed short);
8562 vector unsigned int vec_msum (vector unsigned char,
8563 vector unsigned char,
8564 vector unsigned int);
8565 vector signed int vec_msum (vector signed char,
8566 vector unsigned char,
8568 vector unsigned int vec_msum (vector unsigned short,
8569 vector unsigned short,
8570 vector unsigned int);
8571 vector signed int vec_msum (vector signed short,
8572 vector signed short,
8575 vector signed int vec_vmsumshm (vector signed short,
8576 vector signed short,
8579 vector unsigned int vec_vmsumuhm (vector unsigned short,
8580 vector unsigned short,
8581 vector unsigned int);
8583 vector signed int vec_vmsummbm (vector signed char,
8584 vector unsigned char,
8587 vector unsigned int vec_vmsumubm (vector unsigned char,
8588 vector unsigned char,
8589 vector unsigned int);
8591 vector unsigned int vec_msums (vector unsigned short,
8592 vector unsigned short,
8593 vector unsigned int);
8594 vector signed int vec_msums (vector signed short,
8595 vector signed short,
8598 vector signed int vec_vmsumshs (vector signed short,
8599 vector signed short,
8602 vector unsigned int vec_vmsumuhs (vector unsigned short,
8603 vector unsigned short,
8604 vector unsigned int);
8606 void vec_mtvscr (vector signed int);
8607 void vec_mtvscr (vector unsigned int);
8608 void vec_mtvscr (vector bool int);
8609 void vec_mtvscr (vector signed short);
8610 void vec_mtvscr (vector unsigned short);
8611 void vec_mtvscr (vector bool short);
8612 void vec_mtvscr (vector pixel);
8613 void vec_mtvscr (vector signed char);
8614 void vec_mtvscr (vector unsigned char);
8615 void vec_mtvscr (vector bool char);
8617 vector unsigned short vec_mule (vector unsigned char,
8618 vector unsigned char);
8619 vector signed short vec_mule (vector signed char,
8620 vector signed char);
8621 vector unsigned int vec_mule (vector unsigned short,
8622 vector unsigned short);
8623 vector signed int vec_mule (vector signed short, vector signed short);
8625 vector signed int vec_vmulesh (vector signed short,
8626 vector signed short);
8628 vector unsigned int vec_vmuleuh (vector unsigned short,
8629 vector unsigned short);
8631 vector signed short vec_vmulesb (vector signed char,
8632 vector signed char);
8634 vector unsigned short vec_vmuleub (vector unsigned char,
8635 vector unsigned char);
8637 vector unsigned short vec_mulo (vector unsigned char,
8638 vector unsigned char);
8639 vector signed short vec_mulo (vector signed char, vector signed char);
8640 vector unsigned int vec_mulo (vector unsigned short,
8641 vector unsigned short);
8642 vector signed int vec_mulo (vector signed short, vector signed short);
8644 vector signed int vec_vmulosh (vector signed short,
8645 vector signed short);
8647 vector unsigned int vec_vmulouh (vector unsigned short,
8648 vector unsigned short);
8650 vector signed short vec_vmulosb (vector signed char,
8651 vector signed char);
8653 vector unsigned short vec_vmuloub (vector unsigned char,
8654 vector unsigned char);
8656 vector float vec_nmsub (vector float, vector float, vector float);
8658 vector float vec_nor (vector float, vector float);
8659 vector signed int vec_nor (vector signed int, vector signed int);
8660 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
8661 vector bool int vec_nor (vector bool int, vector bool int);
8662 vector signed short vec_nor (vector signed short, vector signed short);
8663 vector unsigned short vec_nor (vector unsigned short,
8664 vector unsigned short);
8665 vector bool short vec_nor (vector bool short, vector bool short);
8666 vector signed char vec_nor (vector signed char, vector signed char);
8667 vector unsigned char vec_nor (vector unsigned char,
8668 vector unsigned char);
8669 vector bool char vec_nor (vector bool char, vector bool char);
8671 vector float vec_or (vector float, vector float);
8672 vector float vec_or (vector float, vector bool int);
8673 vector float vec_or (vector bool int, vector float);
8674 vector bool int vec_or (vector bool int, vector bool int);
8675 vector signed int vec_or (vector bool int, vector signed int);
8676 vector signed int vec_or (vector signed int, vector bool int);
8677 vector signed int vec_or (vector signed int, vector signed int);
8678 vector unsigned int vec_or (vector bool int, vector unsigned int);
8679 vector unsigned int vec_or (vector unsigned int, vector bool int);
8680 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
8681 vector bool short vec_or (vector bool short, vector bool short);
8682 vector signed short vec_or (vector bool short, vector signed short);
8683 vector signed short vec_or (vector signed short, vector bool short);
8684 vector signed short vec_or (vector signed short, vector signed short);
8685 vector unsigned short vec_or (vector bool short, vector unsigned short);
8686 vector unsigned short vec_or (vector unsigned short, vector bool short);
8687 vector unsigned short vec_or (vector unsigned short,
8688 vector unsigned short);
8689 vector signed char vec_or (vector bool char, vector signed char);
8690 vector bool char vec_or (vector bool char, vector bool char);
8691 vector signed char vec_or (vector signed char, vector bool char);
8692 vector signed char vec_or (vector signed char, vector signed char);
8693 vector unsigned char vec_or (vector bool char, vector unsigned char);
8694 vector unsigned char vec_or (vector unsigned char, vector bool char);
8695 vector unsigned char vec_or (vector unsigned char,
8696 vector unsigned char);
8698 vector signed char vec_pack (vector signed short, vector signed short);
8699 vector unsigned char vec_pack (vector unsigned short,
8700 vector unsigned short);
8701 vector bool char vec_pack (vector bool short, vector bool short);
8702 vector signed short vec_pack (vector signed int, vector signed int);
8703 vector unsigned short vec_pack (vector unsigned int,
8704 vector unsigned int);
8705 vector bool short vec_pack (vector bool int, vector bool int);
8707 vector bool short vec_vpkuwum (vector bool int, vector bool int);
8708 vector signed short vec_vpkuwum (vector signed int, vector signed int);
8709 vector unsigned short vec_vpkuwum (vector unsigned int,
8710 vector unsigned int);
8712 vector bool char vec_vpkuhum (vector bool short, vector bool short);
8713 vector signed char vec_vpkuhum (vector signed short,
8714 vector signed short);
8715 vector unsigned char vec_vpkuhum (vector unsigned short,
8716 vector unsigned short);
8718 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
8720 vector unsigned char vec_packs (vector unsigned short,
8721 vector unsigned short);
8722 vector signed char vec_packs (vector signed short, vector signed short);
8723 vector unsigned short vec_packs (vector unsigned int,
8724 vector unsigned int);
8725 vector signed short vec_packs (vector signed int, vector signed int);
8727 vector signed short vec_vpkswss (vector signed int, vector signed int);
8729 vector unsigned short vec_vpkuwus (vector unsigned int,
8730 vector unsigned int);
8732 vector signed char vec_vpkshss (vector signed short,
8733 vector signed short);
8735 vector unsigned char vec_vpkuhus (vector unsigned short,
8736 vector unsigned short);
8738 vector unsigned char vec_packsu (vector unsigned short,
8739 vector unsigned short);
8740 vector unsigned char vec_packsu (vector signed short,
8741 vector signed short);
8742 vector unsigned short vec_packsu (vector unsigned int,
8743 vector unsigned int);
8744 vector unsigned short vec_packsu (vector signed int, vector signed int);
8746 vector unsigned short vec_vpkswus (vector signed int,
8749 vector unsigned char vec_vpkshus (vector signed short,
8750 vector signed short);
8752 vector float vec_perm (vector float,
8754 vector unsigned char);
8755 vector signed int vec_perm (vector signed int,
8757 vector unsigned char);
8758 vector unsigned int vec_perm (vector unsigned int,
8759 vector unsigned int,
8760 vector unsigned char);
8761 vector bool int vec_perm (vector bool int,
8763 vector unsigned char);
8764 vector signed short vec_perm (vector signed short,
8765 vector signed short,
8766 vector unsigned char);
8767 vector unsigned short vec_perm (vector unsigned short,
8768 vector unsigned short,
8769 vector unsigned char);
8770 vector bool short vec_perm (vector bool short,
8772 vector unsigned char);
8773 vector pixel vec_perm (vector pixel,
8775 vector unsigned char);
8776 vector signed char vec_perm (vector signed char,
8778 vector unsigned char);
8779 vector unsigned char vec_perm (vector unsigned char,
8780 vector unsigned char,
8781 vector unsigned char);
8782 vector bool char vec_perm (vector bool char,
8784 vector unsigned char);
8786 vector float vec_re (vector float);
8788 vector signed char vec_rl (vector signed char,
8789 vector unsigned char);
8790 vector unsigned char vec_rl (vector unsigned char,
8791 vector unsigned char);
8792 vector signed short vec_rl (vector signed short, vector unsigned short);
8793 vector unsigned short vec_rl (vector unsigned short,
8794 vector unsigned short);
8795 vector signed int vec_rl (vector signed int, vector unsigned int);
8796 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
8798 vector signed int vec_vrlw (vector signed int, vector unsigned int);
8799 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
8801 vector signed short vec_vrlh (vector signed short,
8802 vector unsigned short);
8803 vector unsigned short vec_vrlh (vector unsigned short,
8804 vector unsigned short);
8806 vector signed char vec_vrlb (vector signed char, vector unsigned char);
8807 vector unsigned char vec_vrlb (vector unsigned char,
8808 vector unsigned char);
8810 vector float vec_round (vector float);
8812 vector float vec_rsqrte (vector float);
8814 vector float vec_sel (vector float, vector float, vector bool int);
8815 vector float vec_sel (vector float, vector float, vector unsigned int);
8816 vector signed int vec_sel (vector signed int,
8819 vector signed int vec_sel (vector signed int,
8821 vector unsigned int);
8822 vector unsigned int vec_sel (vector unsigned int,
8823 vector unsigned int,
8825 vector unsigned int vec_sel (vector unsigned int,
8826 vector unsigned int,
8827 vector unsigned int);
8828 vector bool int vec_sel (vector bool int,
8831 vector bool int vec_sel (vector bool int,
8833 vector unsigned int);
8834 vector signed short vec_sel (vector signed short,
8835 vector signed short,
8837 vector signed short vec_sel (vector signed short,
8838 vector signed short,
8839 vector unsigned short);
8840 vector unsigned short vec_sel (vector unsigned short,
8841 vector unsigned short,
8843 vector unsigned short vec_sel (vector unsigned short,
8844 vector unsigned short,
8845 vector unsigned short);
8846 vector bool short vec_sel (vector bool short,
8849 vector bool short vec_sel (vector bool short,
8851 vector unsigned short);
8852 vector signed char vec_sel (vector signed char,
8855 vector signed char vec_sel (vector signed char,
8857 vector unsigned char);
8858 vector unsigned char vec_sel (vector unsigned char,
8859 vector unsigned char,
8861 vector unsigned char vec_sel (vector unsigned char,
8862 vector unsigned char,
8863 vector unsigned char);
8864 vector bool char vec_sel (vector bool char,
8867 vector bool char vec_sel (vector bool char,
8869 vector unsigned char);
8871 vector signed char vec_sl (vector signed char,
8872 vector unsigned char);
8873 vector unsigned char vec_sl (vector unsigned char,
8874 vector unsigned char);
8875 vector signed short vec_sl (vector signed short, vector unsigned short);
8876 vector unsigned short vec_sl (vector unsigned short,
8877 vector unsigned short);
8878 vector signed int vec_sl (vector signed int, vector unsigned int);
8879 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
8881 vector signed int vec_vslw (vector signed int, vector unsigned int);
8882 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
8884 vector signed short vec_vslh (vector signed short,
8885 vector unsigned short);
8886 vector unsigned short vec_vslh (vector unsigned short,
8887 vector unsigned short);
8889 vector signed char vec_vslb (vector signed char, vector unsigned char);
8890 vector unsigned char vec_vslb (vector unsigned char,
8891 vector unsigned char);
8893 vector float vec_sld (vector float, vector float, const int);
8894 vector signed int vec_sld (vector signed int,
8897 vector unsigned int vec_sld (vector unsigned int,
8898 vector unsigned int,
8900 vector bool int vec_sld (vector bool int,
8903 vector signed short vec_sld (vector signed short,
8904 vector signed short,
8906 vector unsigned short vec_sld (vector unsigned short,
8907 vector unsigned short,
8909 vector bool short vec_sld (vector bool short,
8912 vector pixel vec_sld (vector pixel,
8915 vector signed char vec_sld (vector signed char,
8918 vector unsigned char vec_sld (vector unsigned char,
8919 vector unsigned char,
8921 vector bool char vec_sld (vector bool char,
8925 vector signed int vec_sll (vector signed int,
8926 vector unsigned int);
8927 vector signed int vec_sll (vector signed int,
8928 vector unsigned short);
8929 vector signed int vec_sll (vector signed int,
8930 vector unsigned char);
8931 vector unsigned int vec_sll (vector unsigned int,
8932 vector unsigned int);
8933 vector unsigned int vec_sll (vector unsigned int,
8934 vector unsigned short);
8935 vector unsigned int vec_sll (vector unsigned int,
8936 vector unsigned char);
8937 vector bool int vec_sll (vector bool int,
8938 vector unsigned int);
8939 vector bool int vec_sll (vector bool int,
8940 vector unsigned short);
8941 vector bool int vec_sll (vector bool int,
8942 vector unsigned char);
8943 vector signed short vec_sll (vector signed short,
8944 vector unsigned int);
8945 vector signed short vec_sll (vector signed short,
8946 vector unsigned short);
8947 vector signed short vec_sll (vector signed short,
8948 vector unsigned char);
8949 vector unsigned short vec_sll (vector unsigned short,
8950 vector unsigned int);
8951 vector unsigned short vec_sll (vector unsigned short,
8952 vector unsigned short);
8953 vector unsigned short vec_sll (vector unsigned short,
8954 vector unsigned char);
8955 vector bool short vec_sll (vector bool short, vector unsigned int);
8956 vector bool short vec_sll (vector bool short, vector unsigned short);
8957 vector bool short vec_sll (vector bool short, vector unsigned char);
8958 vector pixel vec_sll (vector pixel, vector unsigned int);
8959 vector pixel vec_sll (vector pixel, vector unsigned short);
8960 vector pixel vec_sll (vector pixel, vector unsigned char);
8961 vector signed char vec_sll (vector signed char, vector unsigned int);
8962 vector signed char vec_sll (vector signed char, vector unsigned short);
8963 vector signed char vec_sll (vector signed char, vector unsigned char);
8964 vector unsigned char vec_sll (vector unsigned char,
8965 vector unsigned int);
8966 vector unsigned char vec_sll (vector unsigned char,
8967 vector unsigned short);
8968 vector unsigned char vec_sll (vector unsigned char,
8969 vector unsigned char);
8970 vector bool char vec_sll (vector bool char, vector unsigned int);
8971 vector bool char vec_sll (vector bool char, vector unsigned short);
8972 vector bool char vec_sll (vector bool char, vector unsigned char);
8974 vector float vec_slo (vector float, vector signed char);
8975 vector float vec_slo (vector float, vector unsigned char);
8976 vector signed int vec_slo (vector signed int, vector signed char);
8977 vector signed int vec_slo (vector signed int, vector unsigned char);
8978 vector unsigned int vec_slo (vector unsigned int, vector signed char);
8979 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
8980 vector signed short vec_slo (vector signed short, vector signed char);
8981 vector signed short vec_slo (vector signed short, vector unsigned char);
8982 vector unsigned short vec_slo (vector unsigned short,
8983 vector signed char);
8984 vector unsigned short vec_slo (vector unsigned short,
8985 vector unsigned char);
8986 vector pixel vec_slo (vector pixel, vector signed char);
8987 vector pixel vec_slo (vector pixel, vector unsigned char);
8988 vector signed char vec_slo (vector signed char, vector signed char);
8989 vector signed char vec_slo (vector signed char, vector unsigned char);
8990 vector unsigned char vec_slo (vector unsigned char, vector signed char);
8991 vector unsigned char vec_slo (vector unsigned char,
8992 vector unsigned char);
8994 vector signed char vec_splat (vector signed char, const int);
8995 vector unsigned char vec_splat (vector unsigned char, const int);
8996 vector bool char vec_splat (vector bool char, const int);
8997 vector signed short vec_splat (vector signed short, const int);
8998 vector unsigned short vec_splat (vector unsigned short, const int);
8999 vector bool short vec_splat (vector bool short, const int);
9000 vector pixel vec_splat (vector pixel, const int);
9001 vector float vec_splat (vector float, const int);
9002 vector signed int vec_splat (vector signed int, const int);
9003 vector unsigned int vec_splat (vector unsigned int, const int);
9004 vector bool int vec_splat (vector bool int, const int);
9006 vector float vec_vspltw (vector float, const int);
9007 vector signed int vec_vspltw (vector signed int, const int);
9008 vector unsigned int vec_vspltw (vector unsigned int, const int);
9009 vector bool int vec_vspltw (vector bool int, const int);
9011 vector bool short vec_vsplth (vector bool short, const int);
9012 vector signed short vec_vsplth (vector signed short, const int);
9013 vector unsigned short vec_vsplth (vector unsigned short, const int);
9014 vector pixel vec_vsplth (vector pixel, const int);
9016 vector signed char vec_vspltb (vector signed char, const int);
9017 vector unsigned char vec_vspltb (vector unsigned char, const int);
9018 vector bool char vec_vspltb (vector bool char, const int);
9020 vector signed char vec_splat_s8 (const int);
9022 vector signed short vec_splat_s16 (const int);
9024 vector signed int vec_splat_s32 (const int);
9026 vector unsigned char vec_splat_u8 (const int);
9028 vector unsigned short vec_splat_u16 (const int);
9030 vector unsigned int vec_splat_u32 (const int);
9032 vector signed char vec_sr (vector signed char, vector unsigned char);
9033 vector unsigned char vec_sr (vector unsigned char,
9034 vector unsigned char);
9035 vector signed short vec_sr (vector signed short,
9036 vector unsigned short);
9037 vector unsigned short vec_sr (vector unsigned short,
9038 vector unsigned short);
9039 vector signed int vec_sr (vector signed int, vector unsigned int);
9040 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
9042 vector signed int vec_vsrw (vector signed int, vector unsigned int);
9043 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
9045 vector signed short vec_vsrh (vector signed short,
9046 vector unsigned short);
9047 vector unsigned short vec_vsrh (vector unsigned short,
9048 vector unsigned short);
9050 vector signed char vec_vsrb (vector signed char, vector unsigned char);
9051 vector unsigned char vec_vsrb (vector unsigned char,
9052 vector unsigned char);
9054 vector signed char vec_sra (vector signed char, vector unsigned char);
9055 vector unsigned char vec_sra (vector unsigned char,
9056 vector unsigned char);
9057 vector signed short vec_sra (vector signed short,
9058 vector unsigned short);
9059 vector unsigned short vec_sra (vector unsigned short,
9060 vector unsigned short);
9061 vector signed int vec_sra (vector signed int, vector unsigned int);
9062 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
9064 vector signed int vec_vsraw (vector signed int, vector unsigned int);
9065 vector unsigned int vec_vsraw (vector unsigned int,
9066 vector unsigned int);
9068 vector signed short vec_vsrah (vector signed short,
9069 vector unsigned short);
9070 vector unsigned short vec_vsrah (vector unsigned short,
9071 vector unsigned short);
9073 vector signed char vec_vsrab (vector signed char, vector unsigned char);
9074 vector unsigned char vec_vsrab (vector unsigned char,
9075 vector unsigned char);
9077 vector signed int vec_srl (vector signed int, vector unsigned int);
9078 vector signed int vec_srl (vector signed int, vector unsigned short);
9079 vector signed int vec_srl (vector signed int, vector unsigned char);
9080 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
9081 vector unsigned int vec_srl (vector unsigned int,
9082 vector unsigned short);
9083 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
9084 vector bool int vec_srl (vector bool int, vector unsigned int);
9085 vector bool int vec_srl (vector bool int, vector unsigned short);
9086 vector bool int vec_srl (vector bool int, vector unsigned char);
9087 vector signed short vec_srl (vector signed short, vector unsigned int);
9088 vector signed short vec_srl (vector signed short,
9089 vector unsigned short);
9090 vector signed short vec_srl (vector signed short, vector unsigned char);
9091 vector unsigned short vec_srl (vector unsigned short,
9092 vector unsigned int);
9093 vector unsigned short vec_srl (vector unsigned short,
9094 vector unsigned short);
9095 vector unsigned short vec_srl (vector unsigned short,
9096 vector unsigned char);
9097 vector bool short vec_srl (vector bool short, vector unsigned int);
9098 vector bool short vec_srl (vector bool short, vector unsigned short);
9099 vector bool short vec_srl (vector bool short, vector unsigned char);
9100 vector pixel vec_srl (vector pixel, vector unsigned int);
9101 vector pixel vec_srl (vector pixel, vector unsigned short);
9102 vector pixel vec_srl (vector pixel, vector unsigned char);
9103 vector signed char vec_srl (vector signed char, vector unsigned int);
9104 vector signed char vec_srl (vector signed char, vector unsigned short);
9105 vector signed char vec_srl (vector signed char, vector unsigned char);
9106 vector unsigned char vec_srl (vector unsigned char,
9107 vector unsigned int);
9108 vector unsigned char vec_srl (vector unsigned char,
9109 vector unsigned short);
9110 vector unsigned char vec_srl (vector unsigned char,
9111 vector unsigned char);
9112 vector bool char vec_srl (vector bool char, vector unsigned int);
9113 vector bool char vec_srl (vector bool char, vector unsigned short);
9114 vector bool char vec_srl (vector bool char, vector unsigned char);
9116 vector float vec_sro (vector float, vector signed char);
9117 vector float vec_sro (vector float, vector unsigned char);
9118 vector signed int vec_sro (vector signed int, vector signed char);
9119 vector signed int vec_sro (vector signed int, vector unsigned char);
9120 vector unsigned int vec_sro (vector unsigned int, vector signed char);
9121 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
9122 vector signed short vec_sro (vector signed short, vector signed char);
9123 vector signed short vec_sro (vector signed short, vector unsigned char);
9124 vector unsigned short vec_sro (vector unsigned short,
9125 vector signed char);
9126 vector unsigned short vec_sro (vector unsigned short,
9127 vector unsigned char);
9128 vector pixel vec_sro (vector pixel, vector signed char);
9129 vector pixel vec_sro (vector pixel, vector unsigned char);
9130 vector signed char vec_sro (vector signed char, vector signed char);
9131 vector signed char vec_sro (vector signed char, vector unsigned char);
9132 vector unsigned char vec_sro (vector unsigned char, vector signed char);
9133 vector unsigned char vec_sro (vector unsigned char,
9134 vector unsigned char);
9136 void vec_st (vector float, int, vector float *);
9137 void vec_st (vector float, int, float *);
9138 void vec_st (vector signed int, int, vector signed int *);
9139 void vec_st (vector signed int, int, int *);
9140 void vec_st (vector unsigned int, int, vector unsigned int *);
9141 void vec_st (vector unsigned int, int, unsigned int *);
9142 void vec_st (vector bool int, int, vector bool int *);
9143 void vec_st (vector bool int, int, unsigned int *);
9144 void vec_st (vector bool int, int, int *);
9145 void vec_st (vector signed short, int, vector signed short *);
9146 void vec_st (vector signed short, int, short *);
9147 void vec_st (vector unsigned short, int, vector unsigned short *);
9148 void vec_st (vector unsigned short, int, unsigned short *);
9149 void vec_st (vector bool short, int, vector bool short *);
9150 void vec_st (vector bool short, int, unsigned short *);
9151 void vec_st (vector pixel, int, vector pixel *);
9152 void vec_st (vector pixel, int, unsigned short *);
9153 void vec_st (vector pixel, int, short *);
9154 void vec_st (vector bool short, int, short *);
9155 void vec_st (vector signed char, int, vector signed char *);
9156 void vec_st (vector signed char, int, signed char *);
9157 void vec_st (vector unsigned char, int, vector unsigned char *);
9158 void vec_st (vector unsigned char, int, unsigned char *);
9159 void vec_st (vector bool char, int, vector bool char *);
9160 void vec_st (vector bool char, int, unsigned char *);
9161 void vec_st (vector bool char, int, signed char *);
9163 void vec_ste (vector signed char, int, signed char *);
9164 void vec_ste (vector unsigned char, int, unsigned char *);
9165 void vec_ste (vector bool char, int, signed char *);
9166 void vec_ste (vector bool char, int, unsigned char *);
9167 void vec_ste (vector signed short, int, short *);
9168 void vec_ste (vector unsigned short, int, unsigned short *);
9169 void vec_ste (vector bool short, int, short *);
9170 void vec_ste (vector bool short, int, unsigned short *);
9171 void vec_ste (vector pixel, int, short *);
9172 void vec_ste (vector pixel, int, unsigned short *);
9173 void vec_ste (vector float, int, float *);
9174 void vec_ste (vector signed int, int, int *);
9175 void vec_ste (vector unsigned int, int, unsigned int *);
9176 void vec_ste (vector bool int, int, int *);
9177 void vec_ste (vector bool int, int, unsigned int *);
9179 void vec_stvewx (vector float, int, float *);
9180 void vec_stvewx (vector signed int, int, int *);
9181 void vec_stvewx (vector unsigned int, int, unsigned int *);
9182 void vec_stvewx (vector bool int, int, int *);
9183 void vec_stvewx (vector bool int, int, unsigned int *);
9185 void vec_stvehx (vector signed short, int, short *);
9186 void vec_stvehx (vector unsigned short, int, unsigned short *);
9187 void vec_stvehx (vector bool short, int, short *);
9188 void vec_stvehx (vector bool short, int, unsigned short *);
9189 void vec_stvehx (vector pixel, int, short *);
9190 void vec_stvehx (vector pixel, int, unsigned short *);
9192 void vec_stvebx (vector signed char, int, signed char *);
9193 void vec_stvebx (vector unsigned char, int, unsigned char *);
9194 void vec_stvebx (vector bool char, int, signed char *);
9195 void vec_stvebx (vector bool char, int, unsigned char *);
9197 void vec_stl (vector float, int, vector float *);
9198 void vec_stl (vector float, int, float *);
9199 void vec_stl (vector signed int, int, vector signed int *);
9200 void vec_stl (vector signed int, int, int *);
9201 void vec_stl (vector unsigned int, int, vector unsigned int *);
9202 void vec_stl (vector unsigned int, int, unsigned int *);
9203 void vec_stl (vector bool int, int, vector bool int *);
9204 void vec_stl (vector bool int, int, unsigned int *);
9205 void vec_stl (vector bool int, int, int *);
9206 void vec_stl (vector signed short, int, vector signed short *);
9207 void vec_stl (vector signed short, int, short *);
9208 void vec_stl (vector unsigned short, int, vector unsigned short *);
9209 void vec_stl (vector unsigned short, int, unsigned short *);
9210 void vec_stl (vector bool short, int, vector bool short *);
9211 void vec_stl (vector bool short, int, unsigned short *);
9212 void vec_stl (vector bool short, int, short *);
9213 void vec_stl (vector pixel, int, vector pixel *);
9214 void vec_stl (vector pixel, int, unsigned short *);
9215 void vec_stl (vector pixel, int, short *);
9216 void vec_stl (vector signed char, int, vector signed char *);
9217 void vec_stl (vector signed char, int, signed char *);
9218 void vec_stl (vector unsigned char, int, vector unsigned char *);
9219 void vec_stl (vector unsigned char, int, unsigned char *);
9220 void vec_stl (vector bool char, int, vector bool char *);
9221 void vec_stl (vector bool char, int, unsigned char *);
9222 void vec_stl (vector bool char, int, signed char *);
9224 vector signed char vec_sub (vector bool char, vector signed char);
9225 vector signed char vec_sub (vector signed char, vector bool char);
9226 vector signed char vec_sub (vector signed char, vector signed char);
9227 vector unsigned char vec_sub (vector bool char, vector unsigned char);
9228 vector unsigned char vec_sub (vector unsigned char, vector bool char);
9229 vector unsigned char vec_sub (vector unsigned char,
9230 vector unsigned char);
9231 vector signed short vec_sub (vector bool short, vector signed short);
9232 vector signed short vec_sub (vector signed short, vector bool short);
9233 vector signed short vec_sub (vector signed short, vector signed short);
9234 vector unsigned short vec_sub (vector bool short,
9235 vector unsigned short);
9236 vector unsigned short vec_sub (vector unsigned short,
9238 vector unsigned short vec_sub (vector unsigned short,
9239 vector unsigned short);
9240 vector signed int vec_sub (vector bool int, vector signed int);
9241 vector signed int vec_sub (vector signed int, vector bool int);
9242 vector signed int vec_sub (vector signed int, vector signed int);
9243 vector unsigned int vec_sub (vector bool int, vector unsigned int);
9244 vector unsigned int vec_sub (vector unsigned int, vector bool int);
9245 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
9246 vector float vec_sub (vector float, vector float);
9248 vector float vec_vsubfp (vector float, vector float);
9250 vector signed int vec_vsubuwm (vector bool int, vector signed int);
9251 vector signed int vec_vsubuwm (vector signed int, vector bool int);
9252 vector signed int vec_vsubuwm (vector signed int, vector signed int);
9253 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
9254 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
9255 vector unsigned int vec_vsubuwm (vector unsigned int,
9256 vector unsigned int);
9258 vector signed short vec_vsubuhm (vector bool short,
9259 vector signed short);
9260 vector signed short vec_vsubuhm (vector signed short,
9262 vector signed short vec_vsubuhm (vector signed short,
9263 vector signed short);
9264 vector unsigned short vec_vsubuhm (vector bool short,
9265 vector unsigned short);
9266 vector unsigned short vec_vsubuhm (vector unsigned short,
9268 vector unsigned short vec_vsubuhm (vector unsigned short,
9269 vector unsigned short);
9271 vector signed char vec_vsububm (vector bool char, vector signed char);
9272 vector signed char vec_vsububm (vector signed char, vector bool char);
9273 vector signed char vec_vsububm (vector signed char, vector signed char);
9274 vector unsigned char vec_vsububm (vector bool char,
9275 vector unsigned char);
9276 vector unsigned char vec_vsububm (vector unsigned char,
9278 vector unsigned char vec_vsububm (vector unsigned char,
9279 vector unsigned char);
9281 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
9283 vector unsigned char vec_subs (vector bool char, vector unsigned char);
9284 vector unsigned char vec_subs (vector unsigned char, vector bool char);
9285 vector unsigned char vec_subs (vector unsigned char,
9286 vector unsigned char);
9287 vector signed char vec_subs (vector bool char, vector signed char);
9288 vector signed char vec_subs (vector signed char, vector bool char);
9289 vector signed char vec_subs (vector signed char, vector signed char);
9290 vector unsigned short vec_subs (vector bool short,
9291 vector unsigned short);
9292 vector unsigned short vec_subs (vector unsigned short,
9294 vector unsigned short vec_subs (vector unsigned short,
9295 vector unsigned short);
9296 vector signed short vec_subs (vector bool short, vector signed short);
9297 vector signed short vec_subs (vector signed short, vector bool short);
9298 vector signed short vec_subs (vector signed short, vector signed short);
9299 vector unsigned int vec_subs (vector bool int, vector unsigned int);
9300 vector unsigned int vec_subs (vector unsigned int, vector bool int);
9301 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
9302 vector signed int vec_subs (vector bool int, vector signed int);
9303 vector signed int vec_subs (vector signed int, vector bool int);
9304 vector signed int vec_subs (vector signed int, vector signed int);
9306 vector signed int vec_vsubsws (vector bool int, vector signed int);
9307 vector signed int vec_vsubsws (vector signed int, vector bool int);
9308 vector signed int vec_vsubsws (vector signed int, vector signed int);
9310 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
9311 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
9312 vector unsigned int vec_vsubuws (vector unsigned int,
9313 vector unsigned int);
9315 vector signed short vec_vsubshs (vector bool short,
9316 vector signed short);
9317 vector signed short vec_vsubshs (vector signed short,
9319 vector signed short vec_vsubshs (vector signed short,
9320 vector signed short);
9322 vector unsigned short vec_vsubuhs (vector bool short,
9323 vector unsigned short);
9324 vector unsigned short vec_vsubuhs (vector unsigned short,
9326 vector unsigned short vec_vsubuhs (vector unsigned short,
9327 vector unsigned short);
9329 vector signed char vec_vsubsbs (vector bool char, vector signed char);
9330 vector signed char vec_vsubsbs (vector signed char, vector bool char);
9331 vector signed char vec_vsubsbs (vector signed char, vector signed char);
9333 vector unsigned char vec_vsububs (vector bool char,
9334 vector unsigned char);
9335 vector unsigned char vec_vsububs (vector unsigned char,
9337 vector unsigned char vec_vsububs (vector unsigned char,
9338 vector unsigned char);
9340 vector unsigned int vec_sum4s (vector unsigned char,
9341 vector unsigned int);
9342 vector signed int vec_sum4s (vector signed char, vector signed int);
9343 vector signed int vec_sum4s (vector signed short, vector signed int);
9345 vector signed int vec_vsum4shs (vector signed short, vector signed int);
9347 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
9349 vector unsigned int vec_vsum4ubs (vector unsigned char,
9350 vector unsigned int);
9352 vector signed int vec_sum2s (vector signed int, vector signed int);
9354 vector signed int vec_sums (vector signed int, vector signed int);
9356 vector float vec_trunc (vector float);
9358 vector signed short vec_unpackh (vector signed char);
9359 vector bool short vec_unpackh (vector bool char);
9360 vector signed int vec_unpackh (vector signed short);
9361 vector bool int vec_unpackh (vector bool short);
9362 vector unsigned int vec_unpackh (vector pixel);
9364 vector bool int vec_vupkhsh (vector bool short);
9365 vector signed int vec_vupkhsh (vector signed short);
9367 vector unsigned int vec_vupkhpx (vector pixel);
9369 vector bool short vec_vupkhsb (vector bool char);
9370 vector signed short vec_vupkhsb (vector signed char);
9372 vector signed short vec_unpackl (vector signed char);
9373 vector bool short vec_unpackl (vector bool char);
9374 vector unsigned int vec_unpackl (vector pixel);
9375 vector signed int vec_unpackl (vector signed short);
9376 vector bool int vec_unpackl (vector bool short);
9378 vector unsigned int vec_vupklpx (vector pixel);
9380 vector bool int vec_vupklsh (vector bool short);
9381 vector signed int vec_vupklsh (vector signed short);
9383 vector bool short vec_vupklsb (vector bool char);
9384 vector signed short vec_vupklsb (vector signed char);
9386 vector float vec_xor (vector float, vector float);
9387 vector float vec_xor (vector float, vector bool int);
9388 vector float vec_xor (vector bool int, vector float);
9389 vector bool int vec_xor (vector bool int, vector bool int);
9390 vector signed int vec_xor (vector bool int, vector signed int);
9391 vector signed int vec_xor (vector signed int, vector bool int);
9392 vector signed int vec_xor (vector signed int, vector signed int);
9393 vector unsigned int vec_xor (vector bool int, vector unsigned int);
9394 vector unsigned int vec_xor (vector unsigned int, vector bool int);
9395 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
9396 vector bool short vec_xor (vector bool short, vector bool short);
9397 vector signed short vec_xor (vector bool short, vector signed short);
9398 vector signed short vec_xor (vector signed short, vector bool short);
9399 vector signed short vec_xor (vector signed short, vector signed short);
9400 vector unsigned short vec_xor (vector bool short,
9401 vector unsigned short);
9402 vector unsigned short vec_xor (vector unsigned short,
9404 vector unsigned short vec_xor (vector unsigned short,
9405 vector unsigned short);
9406 vector signed char vec_xor (vector bool char, vector signed char);
9407 vector bool char vec_xor (vector bool char, vector bool char);
9408 vector signed char vec_xor (vector signed char, vector bool char);
9409 vector signed char vec_xor (vector signed char, vector signed char);
9410 vector unsigned char vec_xor (vector bool char, vector unsigned char);
9411 vector unsigned char vec_xor (vector unsigned char, vector bool char);
9412 vector unsigned char vec_xor (vector unsigned char,
9413 vector unsigned char);
9415 int vec_all_eq (vector signed char, vector bool char);
9416 int vec_all_eq (vector signed char, vector signed char);
9417 int vec_all_eq (vector unsigned char, vector bool char);
9418 int vec_all_eq (vector unsigned char, vector unsigned char);
9419 int vec_all_eq (vector bool char, vector bool char);
9420 int vec_all_eq (vector bool char, vector unsigned char);
9421 int vec_all_eq (vector bool char, vector signed char);
9422 int vec_all_eq (vector signed short, vector bool short);
9423 int vec_all_eq (vector signed short, vector signed short);
9424 int vec_all_eq (vector unsigned short, vector bool short);
9425 int vec_all_eq (vector unsigned short, vector unsigned short);
9426 int vec_all_eq (vector bool short, vector bool short);
9427 int vec_all_eq (vector bool short, vector unsigned short);
9428 int vec_all_eq (vector bool short, vector signed short);
9429 int vec_all_eq (vector pixel, vector pixel);
9430 int vec_all_eq (vector signed int, vector bool int);
9431 int vec_all_eq (vector signed int, vector signed int);
9432 int vec_all_eq (vector unsigned int, vector bool int);
9433 int vec_all_eq (vector unsigned int, vector unsigned int);
9434 int vec_all_eq (vector bool int, vector bool int);
9435 int vec_all_eq (vector bool int, vector unsigned int);
9436 int vec_all_eq (vector bool int, vector signed int);
9437 int vec_all_eq (vector float, vector float);
9439 int vec_all_ge (vector bool char, vector unsigned char);
9440 int vec_all_ge (vector unsigned char, vector bool char);
9441 int vec_all_ge (vector unsigned char, vector unsigned char);
9442 int vec_all_ge (vector bool char, vector signed char);
9443 int vec_all_ge (vector signed char, vector bool char);
9444 int vec_all_ge (vector signed char, vector signed char);
9445 int vec_all_ge (vector bool short, vector unsigned short);
9446 int vec_all_ge (vector unsigned short, vector bool short);
9447 int vec_all_ge (vector unsigned short, vector unsigned short);
9448 int vec_all_ge (vector signed short, vector signed short);
9449 int vec_all_ge (vector bool short, vector signed short);
9450 int vec_all_ge (vector signed short, vector bool short);
9451 int vec_all_ge (vector bool int, vector unsigned int);
9452 int vec_all_ge (vector unsigned int, vector bool int);
9453 int vec_all_ge (vector unsigned int, vector unsigned int);
9454 int vec_all_ge (vector bool int, vector signed int);
9455 int vec_all_ge (vector signed int, vector bool int);
9456 int vec_all_ge (vector signed int, vector signed int);
9457 int vec_all_ge (vector float, vector float);
9459 int vec_all_gt (vector bool char, vector unsigned char);
9460 int vec_all_gt (vector unsigned char, vector bool char);
9461 int vec_all_gt (vector unsigned char, vector unsigned char);
9462 int vec_all_gt (vector bool char, vector signed char);
9463 int vec_all_gt (vector signed char, vector bool char);
9464 int vec_all_gt (vector signed char, vector signed char);
9465 int vec_all_gt (vector bool short, vector unsigned short);
9466 int vec_all_gt (vector unsigned short, vector bool short);
9467 int vec_all_gt (vector unsigned short, vector unsigned short);
9468 int vec_all_gt (vector bool short, vector signed short);
9469 int vec_all_gt (vector signed short, vector bool short);
9470 int vec_all_gt (vector signed short, vector signed short);
9471 int vec_all_gt (vector bool int, vector unsigned int);
9472 int vec_all_gt (vector unsigned int, vector bool int);
9473 int vec_all_gt (vector unsigned int, vector unsigned int);
9474 int vec_all_gt (vector bool int, vector signed int);
9475 int vec_all_gt (vector signed int, vector bool int);
9476 int vec_all_gt (vector signed int, vector signed int);
9477 int vec_all_gt (vector float, vector float);
9479 int vec_all_in (vector float, vector float);
9481 int vec_all_le (vector bool char, vector unsigned char);
9482 int vec_all_le (vector unsigned char, vector bool char);
9483 int vec_all_le (vector unsigned char, vector unsigned char);
9484 int vec_all_le (vector bool char, vector signed char);
9485 int vec_all_le (vector signed char, vector bool char);
9486 int vec_all_le (vector signed char, vector signed char);
9487 int vec_all_le (vector bool short, vector unsigned short);
9488 int vec_all_le (vector unsigned short, vector bool short);
9489 int vec_all_le (vector unsigned short, vector unsigned short);
9490 int vec_all_le (vector bool short, vector signed short);
9491 int vec_all_le (vector signed short, vector bool short);
9492 int vec_all_le (vector signed short, vector signed short);
9493 int vec_all_le (vector bool int, vector unsigned int);
9494 int vec_all_le (vector unsigned int, vector bool int);
9495 int vec_all_le (vector unsigned int, vector unsigned int);
9496 int vec_all_le (vector bool int, vector signed int);
9497 int vec_all_le (vector signed int, vector bool int);
9498 int vec_all_le (vector signed int, vector signed int);
9499 int vec_all_le (vector float, vector float);
9501 int vec_all_lt (vector bool char, vector unsigned char);
9502 int vec_all_lt (vector unsigned char, vector bool char);
9503 int vec_all_lt (vector unsigned char, vector unsigned char);
9504 int vec_all_lt (vector bool char, vector signed char);
9505 int vec_all_lt (vector signed char, vector bool char);
9506 int vec_all_lt (vector signed char, vector signed char);
9507 int vec_all_lt (vector bool short, vector unsigned short);
9508 int vec_all_lt (vector unsigned short, vector bool short);
9509 int vec_all_lt (vector unsigned short, vector unsigned short);
9510 int vec_all_lt (vector bool short, vector signed short);
9511 int vec_all_lt (vector signed short, vector bool short);
9512 int vec_all_lt (vector signed short, vector signed short);
9513 int vec_all_lt (vector bool int, vector unsigned int);
9514 int vec_all_lt (vector unsigned int, vector bool int);
9515 int vec_all_lt (vector unsigned int, vector unsigned int);
9516 int vec_all_lt (vector bool int, vector signed int);
9517 int vec_all_lt (vector signed int, vector bool int);
9518 int vec_all_lt (vector signed int, vector signed int);
9519 int vec_all_lt (vector float, vector float);
9521 int vec_all_nan (vector float);
9523 int vec_all_ne (vector signed char, vector bool char);
9524 int vec_all_ne (vector signed char, vector signed char);
9525 int vec_all_ne (vector unsigned char, vector bool char);
9526 int vec_all_ne (vector unsigned char, vector unsigned char);
9527 int vec_all_ne (vector bool char, vector bool char);
9528 int vec_all_ne (vector bool char, vector unsigned char);
9529 int vec_all_ne (vector bool char, vector signed char);
9530 int vec_all_ne (vector signed short, vector bool short);
9531 int vec_all_ne (vector signed short, vector signed short);
9532 int vec_all_ne (vector unsigned short, vector bool short);
9533 int vec_all_ne (vector unsigned short, vector unsigned short);
9534 int vec_all_ne (vector bool short, vector bool short);
9535 int vec_all_ne (vector bool short, vector unsigned short);
9536 int vec_all_ne (vector bool short, vector signed short);
9537 int vec_all_ne (vector pixel, vector pixel);
9538 int vec_all_ne (vector signed int, vector bool int);
9539 int vec_all_ne (vector signed int, vector signed int);
9540 int vec_all_ne (vector unsigned int, vector bool int);
9541 int vec_all_ne (vector unsigned int, vector unsigned int);
9542 int vec_all_ne (vector bool int, vector bool int);
9543 int vec_all_ne (vector bool int, vector unsigned int);
9544 int vec_all_ne (vector bool int, vector signed int);
9545 int vec_all_ne (vector float, vector float);
9547 int vec_all_nge (vector float, vector float);
9549 int vec_all_ngt (vector float, vector float);
9551 int vec_all_nle (vector float, vector float);
9553 int vec_all_nlt (vector float, vector float);
9555 int vec_all_numeric (vector float);
9557 int vec_any_eq (vector signed char, vector bool char);
9558 int vec_any_eq (vector signed char, vector signed char);
9559 int vec_any_eq (vector unsigned char, vector bool char);
9560 int vec_any_eq (vector unsigned char, vector unsigned char);
9561 int vec_any_eq (vector bool char, vector bool char);
9562 int vec_any_eq (vector bool char, vector unsigned char);
9563 int vec_any_eq (vector bool char, vector signed char);
9564 int vec_any_eq (vector signed short, vector bool short);
9565 int vec_any_eq (vector signed short, vector signed short);
9566 int vec_any_eq (vector unsigned short, vector bool short);
9567 int vec_any_eq (vector unsigned short, vector unsigned short);
9568 int vec_any_eq (vector bool short, vector bool short);
9569 int vec_any_eq (vector bool short, vector unsigned short);
9570 int vec_any_eq (vector bool short, vector signed short);
9571 int vec_any_eq (vector pixel, vector pixel);
9572 int vec_any_eq (vector signed int, vector bool int);
9573 int vec_any_eq (vector signed int, vector signed int);
9574 int vec_any_eq (vector unsigned int, vector bool int);
9575 int vec_any_eq (vector unsigned int, vector unsigned int);
9576 int vec_any_eq (vector bool int, vector bool int);
9577 int vec_any_eq (vector bool int, vector unsigned int);
9578 int vec_any_eq (vector bool int, vector signed int);
9579 int vec_any_eq (vector float, vector float);
9581 int vec_any_ge (vector signed char, vector bool char);
9582 int vec_any_ge (vector unsigned char, vector bool char);
9583 int vec_any_ge (vector unsigned char, vector unsigned char);
9584 int vec_any_ge (vector signed char, vector signed char);
9585 int vec_any_ge (vector bool char, vector unsigned char);
9586 int vec_any_ge (vector bool char, vector signed char);
9587 int vec_any_ge (vector unsigned short, vector bool short);
9588 int vec_any_ge (vector unsigned short, vector unsigned short);
9589 int vec_any_ge (vector signed short, vector signed short);
9590 int vec_any_ge (vector signed short, vector bool short);
9591 int vec_any_ge (vector bool short, vector unsigned short);
9592 int vec_any_ge (vector bool short, vector signed short);
9593 int vec_any_ge (vector signed int, vector bool int);
9594 int vec_any_ge (vector unsigned int, vector bool int);
9595 int vec_any_ge (vector unsigned int, vector unsigned int);
9596 int vec_any_ge (vector signed int, vector signed int);
9597 int vec_any_ge (vector bool int, vector unsigned int);
9598 int vec_any_ge (vector bool int, vector signed int);
9599 int vec_any_ge (vector float, vector float);
9601 int vec_any_gt (vector bool char, vector unsigned char);
9602 int vec_any_gt (vector unsigned char, vector bool char);
9603 int vec_any_gt (vector unsigned char, vector unsigned char);
9604 int vec_any_gt (vector bool char, vector signed char);
9605 int vec_any_gt (vector signed char, vector bool char);
9606 int vec_any_gt (vector signed char, vector signed char);
9607 int vec_any_gt (vector bool short, vector unsigned short);
9608 int vec_any_gt (vector unsigned short, vector bool short);
9609 int vec_any_gt (vector unsigned short, vector unsigned short);
9610 int vec_any_gt (vector bool short, vector signed short);
9611 int vec_any_gt (vector signed short, vector bool short);
9612 int vec_any_gt (vector signed short, vector signed short);
9613 int vec_any_gt (vector bool int, vector unsigned int);
9614 int vec_any_gt (vector unsigned int, vector bool int);
9615 int vec_any_gt (vector unsigned int, vector unsigned int);
9616 int vec_any_gt (vector bool int, vector signed int);
9617 int vec_any_gt (vector signed int, vector bool int);
9618 int vec_any_gt (vector signed int, vector signed int);
9619 int vec_any_gt (vector float, vector float);
9621 int vec_any_le (vector bool char, vector unsigned char);
9622 int vec_any_le (vector unsigned char, vector bool char);
9623 int vec_any_le (vector unsigned char, vector unsigned char);
9624 int vec_any_le (vector bool char, vector signed char);
9625 int vec_any_le (vector signed char, vector bool char);
9626 int vec_any_le (vector signed char, vector signed char);
9627 int vec_any_le (vector bool short, vector unsigned short);
9628 int vec_any_le (vector unsigned short, vector bool short);
9629 int vec_any_le (vector unsigned short, vector unsigned short);
9630 int vec_any_le (vector bool short, vector signed short);
9631 int vec_any_le (vector signed short, vector bool short);
9632 int vec_any_le (vector signed short, vector signed short);
9633 int vec_any_le (vector bool int, vector unsigned int);
9634 int vec_any_le (vector unsigned int, vector bool int);
9635 int vec_any_le (vector unsigned int, vector unsigned int);
9636 int vec_any_le (vector bool int, vector signed int);
9637 int vec_any_le (vector signed int, vector bool int);
9638 int vec_any_le (vector signed int, vector signed int);
9639 int vec_any_le (vector float, vector float);
9641 int vec_any_lt (vector bool char, vector unsigned char);
9642 int vec_any_lt (vector unsigned char, vector bool char);
9643 int vec_any_lt (vector unsigned char, vector unsigned char);
9644 int vec_any_lt (vector bool char, vector signed char);
9645 int vec_any_lt (vector signed char, vector bool char);
9646 int vec_any_lt (vector signed char, vector signed char);
9647 int vec_any_lt (vector bool short, vector unsigned short);
9648 int vec_any_lt (vector unsigned short, vector bool short);
9649 int vec_any_lt (vector unsigned short, vector unsigned short);
9650 int vec_any_lt (vector bool short, vector signed short);
9651 int vec_any_lt (vector signed short, vector bool short);
9652 int vec_any_lt (vector signed short, vector signed short);
9653 int vec_any_lt (vector bool int, vector unsigned int);
9654 int vec_any_lt (vector unsigned int, vector bool int);
9655 int vec_any_lt (vector unsigned int, vector unsigned int);
9656 int vec_any_lt (vector bool int, vector signed int);
9657 int vec_any_lt (vector signed int, vector bool int);
9658 int vec_any_lt (vector signed int, vector signed int);
9659 int vec_any_lt (vector float, vector float);
9661 int vec_any_nan (vector float);
9663 int vec_any_ne (vector signed char, vector bool char);
9664 int vec_any_ne (vector signed char, vector signed char);
9665 int vec_any_ne (vector unsigned char, vector bool char);
9666 int vec_any_ne (vector unsigned char, vector unsigned char);
9667 int vec_any_ne (vector bool char, vector bool char);
9668 int vec_any_ne (vector bool char, vector unsigned char);
9669 int vec_any_ne (vector bool char, vector signed char);
9670 int vec_any_ne (vector signed short, vector bool short);
9671 int vec_any_ne (vector signed short, vector signed short);
9672 int vec_any_ne (vector unsigned short, vector bool short);
9673 int vec_any_ne (vector unsigned short, vector unsigned short);
9674 int vec_any_ne (vector bool short, vector bool short);
9675 int vec_any_ne (vector bool short, vector unsigned short);
9676 int vec_any_ne (vector bool short, vector signed short);
9677 int vec_any_ne (vector pixel, vector pixel);
9678 int vec_any_ne (vector signed int, vector bool int);
9679 int vec_any_ne (vector signed int, vector signed int);
9680 int vec_any_ne (vector unsigned int, vector bool int);
9681 int vec_any_ne (vector unsigned int, vector unsigned int);
9682 int vec_any_ne (vector bool int, vector bool int);
9683 int vec_any_ne (vector bool int, vector unsigned int);
9684 int vec_any_ne (vector bool int, vector signed int);
9685 int vec_any_ne (vector float, vector float);
9687 int vec_any_nge (vector float, vector float);
9689 int vec_any_ngt (vector float, vector float);
9691 int vec_any_nle (vector float, vector float);
9693 int vec_any_nlt (vector float, vector float);
9695 int vec_any_numeric (vector float);
9697 int vec_any_out (vector float, vector float);
9700 @node SPARC VIS Built-in Functions
9701 @subsection SPARC VIS Built-in Functions
9703 GCC supports SIMD operations on the SPARC using both the generic vector
9704 extensions (@pxref{Vector Extensions}) as well as built-in functions for
9705 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
9706 switch, the VIS extension is exposed as the following built-in functions:
9709 typedef int v2si __attribute__ ((vector_size (8)));
9710 typedef short v4hi __attribute__ ((vector_size (8)));
9711 typedef short v2hi __attribute__ ((vector_size (4)));
9712 typedef char v8qi __attribute__ ((vector_size (8)));
9713 typedef char v4qi __attribute__ ((vector_size (4)));
9715 void * __builtin_vis_alignaddr (void *, long);
9716 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
9717 v2si __builtin_vis_faligndatav2si (v2si, v2si);
9718 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
9719 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
9721 v4hi __builtin_vis_fexpand (v4qi);
9723 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
9724 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
9725 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
9726 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
9727 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
9728 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
9729 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
9731 v4qi __builtin_vis_fpack16 (v4hi);
9732 v8qi __builtin_vis_fpack32 (v2si, v2si);
9733 v2hi __builtin_vis_fpackfix (v2si);
9734 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
9736 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
9739 @node Target Format Checks
9740 @section Format Checks Specific to Particular Target Machines
9742 For some target machines, GCC supports additional options to the
9744 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
9747 * Solaris Format Checks::
9750 @node Solaris Format Checks
9751 @subsection Solaris Format Checks
9753 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
9754 check. @code{cmn_err} accepts a subset of the standard @code{printf}
9755 conversions, and the two-argument @code{%b} conversion for displaying
9756 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
9759 @section Pragmas Accepted by GCC
9763 GCC supports several types of pragmas, primarily in order to compile
9764 code originally written for other compilers. Note that in general
9765 we do not recommend the use of pragmas; @xref{Function Attributes},
9766 for further explanation.
9771 * RS/6000 and PowerPC Pragmas::
9774 * Symbol-Renaming Pragmas::
9775 * Structure-Packing Pragmas::
9777 * Diagnostic Pragmas::
9778 * Visibility Pragmas::
9782 @subsection ARM Pragmas
9784 The ARM target defines pragmas for controlling the default addition of
9785 @code{long_call} and @code{short_call} attributes to functions.
9786 @xref{Function Attributes}, for information about the effects of these
9791 @cindex pragma, long_calls
9792 Set all subsequent functions to have the @code{long_call} attribute.
9795 @cindex pragma, no_long_calls
9796 Set all subsequent functions to have the @code{short_call} attribute.
9798 @item long_calls_off
9799 @cindex pragma, long_calls_off
9800 Do not affect the @code{long_call} or @code{short_call} attributes of
9801 subsequent functions.
9805 @subsection M32C Pragmas
9808 @item memregs @var{number}
9809 @cindex pragma, memregs
9810 Overrides the command line option @code{-memregs=} for the current
9811 file. Use with care! This pragma must be before any function in the
9812 file, and mixing different memregs values in different objects may
9813 make them incompatible. This pragma is useful when a
9814 performance-critical function uses a memreg for temporary values,
9815 as it may allow you to reduce the number of memregs used.
9819 @node RS/6000 and PowerPC Pragmas
9820 @subsection RS/6000 and PowerPC Pragmas
9822 The RS/6000 and PowerPC targets define one pragma for controlling
9823 whether or not the @code{longcall} attribute is added to function
9824 declarations by default. This pragma overrides the @option{-mlongcall}
9825 option, but not the @code{longcall} and @code{shortcall} attributes.
9826 @xref{RS/6000 and PowerPC Options}, for more information about when long
9827 calls are and are not necessary.
9831 @cindex pragma, longcall
9832 Apply the @code{longcall} attribute to all subsequent function
9836 Do not apply the @code{longcall} attribute to subsequent function
9840 @c Describe c4x pragmas here.
9841 @c Describe h8300 pragmas here.
9842 @c Describe sh pragmas here.
9843 @c Describe v850 pragmas here.
9845 @node Darwin Pragmas
9846 @subsection Darwin Pragmas
9848 The following pragmas are available for all architectures running the
9849 Darwin operating system. These are useful for compatibility with other
9853 @item mark @var{tokens}@dots{}
9854 @cindex pragma, mark
9855 This pragma is accepted, but has no effect.
9857 @item options align=@var{alignment}
9858 @cindex pragma, options align
9859 This pragma sets the alignment of fields in structures. The values of
9860 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
9861 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
9862 properly; to restore the previous setting, use @code{reset} for the
9865 @item segment @var{tokens}@dots{}
9866 @cindex pragma, segment
9867 This pragma is accepted, but has no effect.
9869 @item unused (@var{var} [, @var{var}]@dots{})
9870 @cindex pragma, unused
9871 This pragma declares variables to be possibly unused. GCC will not
9872 produce warnings for the listed variables. The effect is similar to
9873 that of the @code{unused} attribute, except that this pragma may appear
9874 anywhere within the variables' scopes.
9877 @node Solaris Pragmas
9878 @subsection Solaris Pragmas
9880 The Solaris target supports @code{#pragma redefine_extname}
9881 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
9882 @code{#pragma} directives for compatibility with the system compiler.
9885 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
9886 @cindex pragma, align
9888 Increase the minimum alignment of each @var{variable} to @var{alignment}.
9889 This is the same as GCC's @code{aligned} attribute @pxref{Variable
9890 Attributes}). Macro expansion occurs on the arguments to this pragma
9891 when compiling C and Objective-C. It does not currently occur when
9892 compiling C++, but this is a bug which may be fixed in a future
9895 @item fini (@var{function} [, @var{function}]...)
9896 @cindex pragma, fini
9898 This pragma causes each listed @var{function} to be called after
9899 main, or during shared module unloading, by adding a call to the
9900 @code{.fini} section.
9902 @item init (@var{function} [, @var{function}]...)
9903 @cindex pragma, init
9905 This pragma causes each listed @var{function} to be called during
9906 initialization (before @code{main}) or during shared module loading, by
9907 adding a call to the @code{.init} section.
9911 @node Symbol-Renaming Pragmas
9912 @subsection Symbol-Renaming Pragmas
9914 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
9915 supports two @code{#pragma} directives which change the name used in
9916 assembly for a given declaration. These pragmas are only available on
9917 platforms whose system headers need them. To get this effect on all
9918 platforms supported by GCC, use the asm labels extension (@pxref{Asm
9922 @item redefine_extname @var{oldname} @var{newname}
9923 @cindex pragma, redefine_extname
9925 This pragma gives the C function @var{oldname} the assembly symbol
9926 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
9927 will be defined if this pragma is available (currently only on
9930 @item extern_prefix @var{string}
9931 @cindex pragma, extern_prefix
9933 This pragma causes all subsequent external function and variable
9934 declarations to have @var{string} prepended to their assembly symbols.
9935 This effect may be terminated with another @code{extern_prefix} pragma
9936 whose argument is an empty string. The preprocessor macro
9937 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
9938 available (currently only on Tru64 UNIX)@.
9941 These pragmas and the asm labels extension interact in a complicated
9942 manner. Here are some corner cases you may want to be aware of.
9945 @item Both pragmas silently apply only to declarations with external
9946 linkage. Asm labels do not have this restriction.
9948 @item In C++, both pragmas silently apply only to declarations with
9949 ``C'' linkage. Again, asm labels do not have this restriction.
9951 @item If any of the three ways of changing the assembly name of a
9952 declaration is applied to a declaration whose assembly name has
9953 already been determined (either by a previous use of one of these
9954 features, or because the compiler needed the assembly name in order to
9955 generate code), and the new name is different, a warning issues and
9956 the name does not change.
9958 @item The @var{oldname} used by @code{#pragma redefine_extname} is
9959 always the C-language name.
9961 @item If @code{#pragma extern_prefix} is in effect, and a declaration
9962 occurs with an asm label attached, the prefix is silently ignored for
9965 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
9966 apply to the same declaration, whichever triggered first wins, and a
9967 warning issues if they contradict each other. (We would like to have
9968 @code{#pragma redefine_extname} always win, for consistency with asm
9969 labels, but if @code{#pragma extern_prefix} triggers first we have no
9970 way of knowing that that happened.)
9973 @node Structure-Packing Pragmas
9974 @subsection Structure-Packing Pragmas
9976 For compatibility with Win32, GCC supports a set of @code{#pragma}
9977 directives which change the maximum alignment of members of structures
9978 (other than zero-width bitfields), unions, and classes subsequently
9979 defined. The @var{n} value below always is required to be a small power
9980 of two and specifies the new alignment in bytes.
9983 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
9984 @item @code{#pragma pack()} sets the alignment to the one that was in
9985 effect when compilation started (see also command line option
9986 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
9987 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
9988 setting on an internal stack and then optionally sets the new alignment.
9989 @item @code{#pragma pack(pop)} restores the alignment setting to the one
9990 saved at the top of the internal stack (and removes that stack entry).
9991 Note that @code{#pragma pack([@var{n}])} does not influence this internal
9992 stack; thus it is possible to have @code{#pragma pack(push)} followed by
9993 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
9994 @code{#pragma pack(pop)}.
9997 Some targets, e.g. i386 and powerpc, support the @code{ms_struct}
9998 @code{#pragma} which lays out a structure as the documented
9999 @code{__attribute__ ((ms_struct))}.
10001 @item @code{#pragma ms_struct on} turns on the layout for structures
10003 @item @code{#pragma ms_struct off} turns off the layout for structures
10005 @item @code{#pragma ms_struct reset} goes back to the default layout.
10009 @subsection Weak Pragmas
10011 For compatibility with SVR4, GCC supports a set of @code{#pragma}
10012 directives for declaring symbols to be weak, and defining weak
10016 @item #pragma weak @var{symbol}
10017 @cindex pragma, weak
10018 This pragma declares @var{symbol} to be weak, as if the declaration
10019 had the attribute of the same name. The pragma may appear before
10020 or after the declaration of @var{symbol}, but must appear before
10021 either its first use or its definition. It is not an error for
10022 @var{symbol} to never be defined at all.
10024 @item #pragma weak @var{symbol1} = @var{symbol2}
10025 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
10026 It is an error if @var{symbol2} is not defined in the current
10030 @node Diagnostic Pragmas
10031 @subsection Diagnostic Pragmas
10033 GCC allows the user to selectively enable or disable certain types of
10034 diagnostics, and change the kind of the diagnostic. For example, a
10035 project's policy might require that all sources compile with
10036 @option{-Werror} but certain files might have exceptions allowing
10037 specific types of warnings. Or, a project might selectively enable
10038 diagnostics and treat them as errors depending on which preprocessor
10039 macros are defined.
10042 @item #pragma GCC diagnostic @var{kind} @var{option}
10043 @cindex pragma, diagnostic
10045 Modifies the disposition of a diagnostic. Note that not all
10046 diagnostics are modifyiable; at the moment only warnings (normally
10047 controlled by @samp{-W...}) can be controlled, and not all of them.
10048 Use @option{-fdiagnostics-show-option} to determine which diagnostics
10049 are controllable and which option controls them.
10051 @var{kind} is @samp{error} to treat this diagnostic as an error,
10052 @samp{warning} to treat it like a warning (even if @option{-Werror} is
10053 in effect), or @samp{ignored} if the diagnostic is to be ignored.
10054 @var{option} is a double quoted string which matches the command line
10058 #pragma GCC diagnostic warning "-Wformat"
10059 #pragma GCC diagnostic error "-Walways-true"
10060 #pragma GCC diagnostic ignored "-Walways-true"
10063 Note that these pragmas override any command line options. Also,
10064 while it is syntactically valid to put these pragmas anywhere in your
10065 sources, the only supported location for them is before any data or
10066 functions are defined. Doing otherwise may result in unpredictable
10067 results depending on how the optimizer manages your sources. If the
10068 same option is listed multiple times, the last one specified is the
10069 one that is in effect. This pragma is not intended to be a general
10070 purpose replacement for command line options, but for implementing
10071 strict control over project policies.
10075 @node Visibility Pragmas
10076 @subsection Visibility Pragmas
10079 @item #pragma GCC visibility push(@var{visibility})
10080 @itemx #pragma GCC visibility pop
10081 @cindex pragma, visibility
10083 This pragma allows the user to set the visibility for multiple
10084 declarations without having to give each a visibility attribute
10085 @xref{Function Attributes}, for more information about visibility and
10086 the attribute syntax.
10088 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
10089 declarations. Class members and template specializations are not
10090 affected; if you want to override the visibility for a particular
10091 member or instantiation, you must use an attribute.
10095 @node Unnamed Fields
10096 @section Unnamed struct/union fields within structs/unions
10100 For compatibility with other compilers, GCC allows you to define
10101 a structure or union that contains, as fields, structures and unions
10102 without names. For example:
10115 In this example, the user would be able to access members of the unnamed
10116 union with code like @samp{foo.b}. Note that only unnamed structs and
10117 unions are allowed, you may not have, for example, an unnamed
10120 You must never create such structures that cause ambiguous field definitions.
10121 For example, this structure:
10132 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
10133 Such constructs are not supported and must be avoided. In the future,
10134 such constructs may be detected and treated as compilation errors.
10136 @opindex fms-extensions
10137 Unless @option{-fms-extensions} is used, the unnamed field must be a
10138 structure or union definition without a tag (for example, @samp{struct
10139 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
10140 also be a definition with a tag such as @samp{struct foo @{ int a;
10141 @};}, a reference to a previously defined structure or union such as
10142 @samp{struct foo;}, or a reference to a @code{typedef} name for a
10143 previously defined structure or union type.
10146 @section Thread-Local Storage
10147 @cindex Thread-Local Storage
10148 @cindex @acronym{TLS}
10151 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
10152 are allocated such that there is one instance of the variable per extant
10153 thread. The run-time model GCC uses to implement this originates
10154 in the IA-64 processor-specific ABI, but has since been migrated
10155 to other processors as well. It requires significant support from
10156 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
10157 system libraries (@file{libc.so} and @file{libpthread.so}), so it
10158 is not available everywhere.
10160 At the user level, the extension is visible with a new storage
10161 class keyword: @code{__thread}. For example:
10165 extern __thread struct state s;
10166 static __thread char *p;
10169 The @code{__thread} specifier may be used alone, with the @code{extern}
10170 or @code{static} specifiers, but with no other storage class specifier.
10171 When used with @code{extern} or @code{static}, @code{__thread} must appear
10172 immediately after the other storage class specifier.
10174 The @code{__thread} specifier may be applied to any global, file-scoped
10175 static, function-scoped static, or static data member of a class. It may
10176 not be applied to block-scoped automatic or non-static data member.
10178 When the address-of operator is applied to a thread-local variable, it is
10179 evaluated at run-time and returns the address of the current thread's
10180 instance of that variable. An address so obtained may be used by any
10181 thread. When a thread terminates, any pointers to thread-local variables
10182 in that thread become invalid.
10184 No static initialization may refer to the address of a thread-local variable.
10186 In C++, if an initializer is present for a thread-local variable, it must
10187 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
10190 See @uref{http://people.redhat.com/drepper/tls.pdf,
10191 ELF Handling For Thread-Local Storage} for a detailed explanation of
10192 the four thread-local storage addressing models, and how the run-time
10193 is expected to function.
10196 * C99 Thread-Local Edits::
10197 * C++98 Thread-Local Edits::
10200 @node C99 Thread-Local Edits
10201 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
10203 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
10204 that document the exact semantics of the language extension.
10208 @cite{5.1.2 Execution environments}
10210 Add new text after paragraph 1
10213 Within either execution environment, a @dfn{thread} is a flow of
10214 control within a program. It is implementation defined whether
10215 or not there may be more than one thread associated with a program.
10216 It is implementation defined how threads beyond the first are
10217 created, the name and type of the function called at thread
10218 startup, and how threads may be terminated. However, objects
10219 with thread storage duration shall be initialized before thread
10224 @cite{6.2.4 Storage durations of objects}
10226 Add new text before paragraph 3
10229 An object whose identifier is declared with the storage-class
10230 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
10231 Its lifetime is the entire execution of the thread, and its
10232 stored value is initialized only once, prior to thread startup.
10236 @cite{6.4.1 Keywords}
10238 Add @code{__thread}.
10241 @cite{6.7.1 Storage-class specifiers}
10243 Add @code{__thread} to the list of storage class specifiers in
10246 Change paragraph 2 to
10249 With the exception of @code{__thread}, at most one storage-class
10250 specifier may be given [@dots{}]. The @code{__thread} specifier may
10251 be used alone, or immediately following @code{extern} or
10255 Add new text after paragraph 6
10258 The declaration of an identifier for a variable that has
10259 block scope that specifies @code{__thread} shall also
10260 specify either @code{extern} or @code{static}.
10262 The @code{__thread} specifier shall be used only with
10267 @node C++98 Thread-Local Edits
10268 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
10270 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
10271 that document the exact semantics of the language extension.
10275 @b{[intro.execution]}
10277 New text after paragraph 4
10280 A @dfn{thread} is a flow of control within the abstract machine.
10281 It is implementation defined whether or not there may be more than
10285 New text after paragraph 7
10288 It is unspecified whether additional action must be taken to
10289 ensure when and whether side effects are visible to other threads.
10295 Add @code{__thread}.
10298 @b{[basic.start.main]}
10300 Add after paragraph 5
10303 The thread that begins execution at the @code{main} function is called
10304 the @dfn{main thread}. It is implementation defined how functions
10305 beginning threads other than the main thread are designated or typed.
10306 A function so designated, as well as the @code{main} function, is called
10307 a @dfn{thread startup function}. It is implementation defined what
10308 happens if a thread startup function returns. It is implementation
10309 defined what happens to other threads when any thread calls @code{exit}.
10313 @b{[basic.start.init]}
10315 Add after paragraph 4
10318 The storage for an object of thread storage duration shall be
10319 statically initialized before the first statement of the thread startup
10320 function. An object of thread storage duration shall not require
10321 dynamic initialization.
10325 @b{[basic.start.term]}
10327 Add after paragraph 3
10330 The type of an object with thread storage duration shall not have a
10331 non-trivial destructor, nor shall it be an array type whose elements
10332 (directly or indirectly) have non-trivial destructors.
10338 Add ``thread storage duration'' to the list in paragraph 1.
10343 Thread, static, and automatic storage durations are associated with
10344 objects introduced by declarations [@dots{}].
10347 Add @code{__thread} to the list of specifiers in paragraph 3.
10350 @b{[basic.stc.thread]}
10352 New section before @b{[basic.stc.static]}
10355 The keyword @code{__thread} applied to a non-local object gives the
10356 object thread storage duration.
10358 A local variable or class data member declared both @code{static}
10359 and @code{__thread} gives the variable or member thread storage
10364 @b{[basic.stc.static]}
10369 All objects which have neither thread storage duration, dynamic
10370 storage duration nor are local [@dots{}].
10376 Add @code{__thread} to the list in paragraph 1.
10381 With the exception of @code{__thread}, at most one
10382 @var{storage-class-specifier} shall appear in a given
10383 @var{decl-specifier-seq}. The @code{__thread} specifier may
10384 be used alone, or immediately following the @code{extern} or
10385 @code{static} specifiers. [@dots{}]
10388 Add after paragraph 5
10391 The @code{__thread} specifier can be applied only to the names of objects
10392 and to anonymous unions.
10398 Add after paragraph 6
10401 Non-@code{static} members shall not be @code{__thread}.
10405 @node C++ Extensions
10406 @chapter Extensions to the C++ Language
10407 @cindex extensions, C++ language
10408 @cindex C++ language extensions
10410 The GNU compiler provides these extensions to the C++ language (and you
10411 can also use most of the C language extensions in your C++ programs). If you
10412 want to write code that checks whether these features are available, you can
10413 test for the GNU compiler the same way as for C programs: check for a
10414 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
10415 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
10416 Predefined Macros,cpp,The GNU C Preprocessor}).
10419 * Volatiles:: What constitutes an access to a volatile object.
10420 * Restricted Pointers:: C99 restricted pointers and references.
10421 * Vague Linkage:: Where G++ puts inlines, vtables and such.
10422 * C++ Interface:: You can use a single C++ header file for both
10423 declarations and definitions.
10424 * Template Instantiation:: Methods for ensuring that exactly one copy of
10425 each needed template instantiation is emitted.
10426 * Bound member functions:: You can extract a function pointer to the
10427 method denoted by a @samp{->*} or @samp{.*} expression.
10428 * C++ Attributes:: Variable, function, and type attributes for C++ only.
10429 * Namespace Association:: Strong using-directives for namespace association.
10430 * Java Exceptions:: Tweaking exception handling to work with Java.
10431 * Deprecated Features:: Things will disappear from g++.
10432 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
10436 @section When is a Volatile Object Accessed?
10437 @cindex accessing volatiles
10438 @cindex volatile read
10439 @cindex volatile write
10440 @cindex volatile access
10442 Both the C and C++ standard have the concept of volatile objects. These
10443 are normally accessed by pointers and used for accessing hardware. The
10444 standards encourage compilers to refrain from optimizations
10445 concerning accesses to volatile objects that it might perform on
10446 non-volatile objects. The C standard leaves it implementation defined
10447 as to what constitutes a volatile access. The C++ standard omits to
10448 specify this, except to say that C++ should behave in a similar manner
10449 to C with respect to volatiles, where possible. The minimum either
10450 standard specifies is that at a sequence point all previous accesses to
10451 volatile objects have stabilized and no subsequent accesses have
10452 occurred. Thus an implementation is free to reorder and combine
10453 volatile accesses which occur between sequence points, but cannot do so
10454 for accesses across a sequence point. The use of volatiles does not
10455 allow you to violate the restriction on updating objects multiple times
10456 within a sequence point.
10458 In most expressions, it is intuitively obvious what is a read and what is
10459 a write. For instance
10462 volatile int *dst = @var{somevalue};
10463 volatile int *src = @var{someothervalue};
10468 will cause a read of the volatile object pointed to by @var{src} and stores the
10469 value into the volatile object pointed to by @var{dst}. There is no
10470 guarantee that these reads and writes are atomic, especially for objects
10471 larger than @code{int}.
10473 Less obvious expressions are where something which looks like an access
10474 is used in a void context. An example would be,
10477 volatile int *src = @var{somevalue};
10481 With C, such expressions are rvalues, and as rvalues cause a read of
10482 the object, GCC interprets this as a read of the volatile being pointed
10483 to. The C++ standard specifies that such expressions do not undergo
10484 lvalue to rvalue conversion, and that the type of the dereferenced
10485 object may be incomplete. The C++ standard does not specify explicitly
10486 that it is this lvalue to rvalue conversion which is responsible for
10487 causing an access. However, there is reason to believe that it is,
10488 because otherwise certain simple expressions become undefined. However,
10489 because it would surprise most programmers, G++ treats dereferencing a
10490 pointer to volatile object of complete type in a void context as a read
10491 of the object. When the object has incomplete type, G++ issues a
10496 struct T @{int m;@};
10497 volatile S *ptr1 = @var{somevalue};
10498 volatile T *ptr2 = @var{somevalue};
10503 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
10504 causes a read of the object pointed to. If you wish to force an error on
10505 the first case, you must force a conversion to rvalue with, for instance
10506 a static cast, @code{static_cast<S>(*ptr1)}.
10508 When using a reference to volatile, G++ does not treat equivalent
10509 expressions as accesses to volatiles, but instead issues a warning that
10510 no volatile is accessed. The rationale for this is that otherwise it
10511 becomes difficult to determine where volatile access occur, and not
10512 possible to ignore the return value from functions returning volatile
10513 references. Again, if you wish to force a read, cast the reference to
10516 @node Restricted Pointers
10517 @section Restricting Pointer Aliasing
10518 @cindex restricted pointers
10519 @cindex restricted references
10520 @cindex restricted this pointer
10522 As with the C front end, G++ understands the C99 feature of restricted pointers,
10523 specified with the @code{__restrict__}, or @code{__restrict} type
10524 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
10525 language flag, @code{restrict} is not a keyword in C++.
10527 In addition to allowing restricted pointers, you can specify restricted
10528 references, which indicate that the reference is not aliased in the local
10532 void fn (int *__restrict__ rptr, int &__restrict__ rref)
10539 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
10540 @var{rref} refers to a (different) unaliased integer.
10542 You may also specify whether a member function's @var{this} pointer is
10543 unaliased by using @code{__restrict__} as a member function qualifier.
10546 void T::fn () __restrict__
10553 Within the body of @code{T::fn}, @var{this} will have the effective
10554 definition @code{T *__restrict__ const this}. Notice that the
10555 interpretation of a @code{__restrict__} member function qualifier is
10556 different to that of @code{const} or @code{volatile} qualifier, in that it
10557 is applied to the pointer rather than the object. This is consistent with
10558 other compilers which implement restricted pointers.
10560 As with all outermost parameter qualifiers, @code{__restrict__} is
10561 ignored in function definition matching. This means you only need to
10562 specify @code{__restrict__} in a function definition, rather than
10563 in a function prototype as well.
10565 @node Vague Linkage
10566 @section Vague Linkage
10567 @cindex vague linkage
10569 There are several constructs in C++ which require space in the object
10570 file but are not clearly tied to a single translation unit. We say that
10571 these constructs have ``vague linkage''. Typically such constructs are
10572 emitted wherever they are needed, though sometimes we can be more
10576 @item Inline Functions
10577 Inline functions are typically defined in a header file which can be
10578 included in many different compilations. Hopefully they can usually be
10579 inlined, but sometimes an out-of-line copy is necessary, if the address
10580 of the function is taken or if inlining fails. In general, we emit an
10581 out-of-line copy in all translation units where one is needed. As an
10582 exception, we only emit inline virtual functions with the vtable, since
10583 it will always require a copy.
10585 Local static variables and string constants used in an inline function
10586 are also considered to have vague linkage, since they must be shared
10587 between all inlined and out-of-line instances of the function.
10591 C++ virtual functions are implemented in most compilers using a lookup
10592 table, known as a vtable. The vtable contains pointers to the virtual
10593 functions provided by a class, and each object of the class contains a
10594 pointer to its vtable (or vtables, in some multiple-inheritance
10595 situations). If the class declares any non-inline, non-pure virtual
10596 functions, the first one is chosen as the ``key method'' for the class,
10597 and the vtable is only emitted in the translation unit where the key
10600 @emph{Note:} If the chosen key method is later defined as inline, the
10601 vtable will still be emitted in every translation unit which defines it.
10602 Make sure that any inline virtuals are declared inline in the class
10603 body, even if they are not defined there.
10605 @item type_info objects
10608 C++ requires information about types to be written out in order to
10609 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
10610 For polymorphic classes (classes with virtual functions), the type_info
10611 object is written out along with the vtable so that @samp{dynamic_cast}
10612 can determine the dynamic type of a class object at runtime. For all
10613 other types, we write out the type_info object when it is used: when
10614 applying @samp{typeid} to an expression, throwing an object, or
10615 referring to a type in a catch clause or exception specification.
10617 @item Template Instantiations
10618 Most everything in this section also applies to template instantiations,
10619 but there are other options as well.
10620 @xref{Template Instantiation,,Where's the Template?}.
10624 When used with GNU ld version 2.8 or later on an ELF system such as
10625 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
10626 these constructs will be discarded at link time. This is known as
10629 On targets that don't support COMDAT, but do support weak symbols, GCC
10630 will use them. This way one copy will override all the others, but
10631 the unused copies will still take up space in the executable.
10633 For targets which do not support either COMDAT or weak symbols,
10634 most entities with vague linkage will be emitted as local symbols to
10635 avoid duplicate definition errors from the linker. This will not happen
10636 for local statics in inlines, however, as having multiple copies will
10637 almost certainly break things.
10639 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
10640 another way to control placement of these constructs.
10642 @node C++ Interface
10643 @section #pragma interface and implementation
10645 @cindex interface and implementation headers, C++
10646 @cindex C++ interface and implementation headers
10647 @cindex pragmas, interface and implementation
10649 @code{#pragma interface} and @code{#pragma implementation} provide the
10650 user with a way of explicitly directing the compiler to emit entities
10651 with vague linkage (and debugging information) in a particular
10654 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
10655 most cases, because of COMDAT support and the ``key method'' heuristic
10656 mentioned in @ref{Vague Linkage}. Using them can actually cause your
10657 program to grow due to unnecessary out-of-line copies of inline
10658 functions. Currently (3.4) the only benefit of these
10659 @code{#pragma}s is reduced duplication of debugging information, and
10660 that should be addressed soon on DWARF 2 targets with the use of
10664 @item #pragma interface
10665 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
10666 @kindex #pragma interface
10667 Use this directive in @emph{header files} that define object classes, to save
10668 space in most of the object files that use those classes. Normally,
10669 local copies of certain information (backup copies of inline member
10670 functions, debugging information, and the internal tables that implement
10671 virtual functions) must be kept in each object file that includes class
10672 definitions. You can use this pragma to avoid such duplication. When a
10673 header file containing @samp{#pragma interface} is included in a
10674 compilation, this auxiliary information will not be generated (unless
10675 the main input source file itself uses @samp{#pragma implementation}).
10676 Instead, the object files will contain references to be resolved at link
10679 The second form of this directive is useful for the case where you have
10680 multiple headers with the same name in different directories. If you
10681 use this form, you must specify the same string to @samp{#pragma
10684 @item #pragma implementation
10685 @itemx #pragma implementation "@var{objects}.h"
10686 @kindex #pragma implementation
10687 Use this pragma in a @emph{main input file}, when you want full output from
10688 included header files to be generated (and made globally visible). The
10689 included header file, in turn, should use @samp{#pragma interface}.
10690 Backup copies of inline member functions, debugging information, and the
10691 internal tables used to implement virtual functions are all generated in
10692 implementation files.
10694 @cindex implied @code{#pragma implementation}
10695 @cindex @code{#pragma implementation}, implied
10696 @cindex naming convention, implementation headers
10697 If you use @samp{#pragma implementation} with no argument, it applies to
10698 an include file with the same basename@footnote{A file's @dfn{basename}
10699 was the name stripped of all leading path information and of trailing
10700 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
10701 file. For example, in @file{allclass.cc}, giving just
10702 @samp{#pragma implementation}
10703 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
10705 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
10706 an implementation file whenever you would include it from
10707 @file{allclass.cc} even if you never specified @samp{#pragma
10708 implementation}. This was deemed to be more trouble than it was worth,
10709 however, and disabled.
10711 Use the string argument if you want a single implementation file to
10712 include code from multiple header files. (You must also use
10713 @samp{#include} to include the header file; @samp{#pragma
10714 implementation} only specifies how to use the file---it doesn't actually
10717 There is no way to split up the contents of a single header file into
10718 multiple implementation files.
10721 @cindex inlining and C++ pragmas
10722 @cindex C++ pragmas, effect on inlining
10723 @cindex pragmas in C++, effect on inlining
10724 @samp{#pragma implementation} and @samp{#pragma interface} also have an
10725 effect on function inlining.
10727 If you define a class in a header file marked with @samp{#pragma
10728 interface}, the effect on an inline function defined in that class is
10729 similar to an explicit @code{extern} declaration---the compiler emits
10730 no code at all to define an independent version of the function. Its
10731 definition is used only for inlining with its callers.
10733 @opindex fno-implement-inlines
10734 Conversely, when you include the same header file in a main source file
10735 that declares it as @samp{#pragma implementation}, the compiler emits
10736 code for the function itself; this defines a version of the function
10737 that can be found via pointers (or by callers compiled without
10738 inlining). If all calls to the function can be inlined, you can avoid
10739 emitting the function by compiling with @option{-fno-implement-inlines}.
10740 If any calls were not inlined, you will get linker errors.
10742 @node Template Instantiation
10743 @section Where's the Template?
10744 @cindex template instantiation
10746 C++ templates are the first language feature to require more
10747 intelligence from the environment than one usually finds on a UNIX
10748 system. Somehow the compiler and linker have to make sure that each
10749 template instance occurs exactly once in the executable if it is needed,
10750 and not at all otherwise. There are two basic approaches to this
10751 problem, which are referred to as the Borland model and the Cfront model.
10754 @item Borland model
10755 Borland C++ solved the template instantiation problem by adding the code
10756 equivalent of common blocks to their linker; the compiler emits template
10757 instances in each translation unit that uses them, and the linker
10758 collapses them together. The advantage of this model is that the linker
10759 only has to consider the object files themselves; there is no external
10760 complexity to worry about. This disadvantage is that compilation time
10761 is increased because the template code is being compiled repeatedly.
10762 Code written for this model tends to include definitions of all
10763 templates in the header file, since they must be seen to be
10767 The AT&T C++ translator, Cfront, solved the template instantiation
10768 problem by creating the notion of a template repository, an
10769 automatically maintained place where template instances are stored. A
10770 more modern version of the repository works as follows: As individual
10771 object files are built, the compiler places any template definitions and
10772 instantiations encountered in the repository. At link time, the link
10773 wrapper adds in the objects in the repository and compiles any needed
10774 instances that were not previously emitted. The advantages of this
10775 model are more optimal compilation speed and the ability to use the
10776 system linker; to implement the Borland model a compiler vendor also
10777 needs to replace the linker. The disadvantages are vastly increased
10778 complexity, and thus potential for error; for some code this can be
10779 just as transparent, but in practice it can been very difficult to build
10780 multiple programs in one directory and one program in multiple
10781 directories. Code written for this model tends to separate definitions
10782 of non-inline member templates into a separate file, which should be
10783 compiled separately.
10786 When used with GNU ld version 2.8 or later on an ELF system such as
10787 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
10788 Borland model. On other systems, G++ implements neither automatic
10791 A future version of G++ will support a hybrid model whereby the compiler
10792 will emit any instantiations for which the template definition is
10793 included in the compile, and store template definitions and
10794 instantiation context information into the object file for the rest.
10795 The link wrapper will extract that information as necessary and invoke
10796 the compiler to produce the remaining instantiations. The linker will
10797 then combine duplicate instantiations.
10799 In the mean time, you have the following options for dealing with
10800 template instantiations:
10805 Compile your template-using code with @option{-frepo}. The compiler will
10806 generate files with the extension @samp{.rpo} listing all of the
10807 template instantiations used in the corresponding object files which
10808 could be instantiated there; the link wrapper, @samp{collect2}, will
10809 then update the @samp{.rpo} files to tell the compiler where to place
10810 those instantiations and rebuild any affected object files. The
10811 link-time overhead is negligible after the first pass, as the compiler
10812 will continue to place the instantiations in the same files.
10814 This is your best option for application code written for the Borland
10815 model, as it will just work. Code written for the Cfront model will
10816 need to be modified so that the template definitions are available at
10817 one or more points of instantiation; usually this is as simple as adding
10818 @code{#include <tmethods.cc>} to the end of each template header.
10820 For library code, if you want the library to provide all of the template
10821 instantiations it needs, just try to link all of its object files
10822 together; the link will fail, but cause the instantiations to be
10823 generated as a side effect. Be warned, however, that this may cause
10824 conflicts if multiple libraries try to provide the same instantiations.
10825 For greater control, use explicit instantiation as described in the next
10829 @opindex fno-implicit-templates
10830 Compile your code with @option{-fno-implicit-templates} to disable the
10831 implicit generation of template instances, and explicitly instantiate
10832 all the ones you use. This approach requires more knowledge of exactly
10833 which instances you need than do the others, but it's less
10834 mysterious and allows greater control. You can scatter the explicit
10835 instantiations throughout your program, perhaps putting them in the
10836 translation units where the instances are used or the translation units
10837 that define the templates themselves; you can put all of the explicit
10838 instantiations you need into one big file; or you can create small files
10845 template class Foo<int>;
10846 template ostream& operator <<
10847 (ostream&, const Foo<int>&);
10850 for each of the instances you need, and create a template instantiation
10851 library from those.
10853 If you are using Cfront-model code, you can probably get away with not
10854 using @option{-fno-implicit-templates} when compiling files that don't
10855 @samp{#include} the member template definitions.
10857 If you use one big file to do the instantiations, you may want to
10858 compile it without @option{-fno-implicit-templates} so you get all of the
10859 instances required by your explicit instantiations (but not by any
10860 other files) without having to specify them as well.
10862 G++ has extended the template instantiation syntax given in the ISO
10863 standard to allow forward declaration of explicit instantiations
10864 (with @code{extern}), instantiation of the compiler support data for a
10865 template class (i.e.@: the vtable) without instantiating any of its
10866 members (with @code{inline}), and instantiation of only the static data
10867 members of a template class, without the support data or member
10868 functions (with (@code{static}):
10871 extern template int max (int, int);
10872 inline template class Foo<int>;
10873 static template class Foo<int>;
10877 Do nothing. Pretend G++ does implement automatic instantiation
10878 management. Code written for the Borland model will work fine, but
10879 each translation unit will contain instances of each of the templates it
10880 uses. In a large program, this can lead to an unacceptable amount of code
10884 @node Bound member functions
10885 @section Extracting the function pointer from a bound pointer to member function
10887 @cindex pointer to member function
10888 @cindex bound pointer to member function
10890 In C++, pointer to member functions (PMFs) are implemented using a wide
10891 pointer of sorts to handle all the possible call mechanisms; the PMF
10892 needs to store information about how to adjust the @samp{this} pointer,
10893 and if the function pointed to is virtual, where to find the vtable, and
10894 where in the vtable to look for the member function. If you are using
10895 PMFs in an inner loop, you should really reconsider that decision. If
10896 that is not an option, you can extract the pointer to the function that
10897 would be called for a given object/PMF pair and call it directly inside
10898 the inner loop, to save a bit of time.
10900 Note that you will still be paying the penalty for the call through a
10901 function pointer; on most modern architectures, such a call defeats the
10902 branch prediction features of the CPU@. This is also true of normal
10903 virtual function calls.
10905 The syntax for this extension is
10909 extern int (A::*fp)();
10910 typedef int (*fptr)(A *);
10912 fptr p = (fptr)(a.*fp);
10915 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
10916 no object is needed to obtain the address of the function. They can be
10917 converted to function pointers directly:
10920 fptr p1 = (fptr)(&A::foo);
10923 @opindex Wno-pmf-conversions
10924 You must specify @option{-Wno-pmf-conversions} to use this extension.
10926 @node C++ Attributes
10927 @section C++-Specific Variable, Function, and Type Attributes
10929 Some attributes only make sense for C++ programs.
10932 @item init_priority (@var{priority})
10933 @cindex init_priority attribute
10936 In Standard C++, objects defined at namespace scope are guaranteed to be
10937 initialized in an order in strict accordance with that of their definitions
10938 @emph{in a given translation unit}. No guarantee is made for initializations
10939 across translation units. However, GNU C++ allows users to control the
10940 order of initialization of objects defined at namespace scope with the
10941 @code{init_priority} attribute by specifying a relative @var{priority},
10942 a constant integral expression currently bounded between 101 and 65535
10943 inclusive. Lower numbers indicate a higher priority.
10945 In the following example, @code{A} would normally be created before
10946 @code{B}, but the @code{init_priority} attribute has reversed that order:
10949 Some_Class A __attribute__ ((init_priority (2000)));
10950 Some_Class B __attribute__ ((init_priority (543)));
10954 Note that the particular values of @var{priority} do not matter; only their
10957 @item java_interface
10958 @cindex java_interface attribute
10960 This type attribute informs C++ that the class is a Java interface. It may
10961 only be applied to classes declared within an @code{extern "Java"} block.
10962 Calls to methods declared in this interface will be dispatched using GCJ's
10963 interface table mechanism, instead of regular virtual table dispatch.
10967 See also @xref{Namespace Association}.
10969 @node Namespace Association
10970 @section Namespace Association
10972 @strong{Caution:} The semantics of this extension are not fully
10973 defined. Users should refrain from using this extension as its
10974 semantics may change subtly over time. It is possible that this
10975 extension will be removed in future versions of G++.
10977 A using-directive with @code{__attribute ((strong))} is stronger
10978 than a normal using-directive in two ways:
10982 Templates from the used namespace can be specialized and explicitly
10983 instantiated as though they were members of the using namespace.
10986 The using namespace is considered an associated namespace of all
10987 templates in the used namespace for purposes of argument-dependent
10991 The used namespace must be nested within the using namespace so that
10992 normal unqualified lookup works properly.
10994 This is useful for composing a namespace transparently from
10995 implementation namespaces. For example:
11000 template <class T> struct A @{ @};
11002 using namespace debug __attribute ((__strong__));
11003 template <> struct A<int> @{ @}; // @r{ok to specialize}
11005 template <class T> void f (A<T>);
11010 f (std::A<float>()); // @r{lookup finds} std::f
11015 @node Java Exceptions
11016 @section Java Exceptions
11018 The Java language uses a slightly different exception handling model
11019 from C++. Normally, GNU C++ will automatically detect when you are
11020 writing C++ code that uses Java exceptions, and handle them
11021 appropriately. However, if C++ code only needs to execute destructors
11022 when Java exceptions are thrown through it, GCC will guess incorrectly.
11023 Sample problematic code is:
11026 struct S @{ ~S(); @};
11027 extern void bar(); // @r{is written in Java, and may throw exceptions}
11036 The usual effect of an incorrect guess is a link failure, complaining of
11037 a missing routine called @samp{__gxx_personality_v0}.
11039 You can inform the compiler that Java exceptions are to be used in a
11040 translation unit, irrespective of what it might think, by writing
11041 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
11042 @samp{#pragma} must appear before any functions that throw or catch
11043 exceptions, or run destructors when exceptions are thrown through them.
11045 You cannot mix Java and C++ exceptions in the same translation unit. It
11046 is believed to be safe to throw a C++ exception from one file through
11047 another file compiled for the Java exception model, or vice versa, but
11048 there may be bugs in this area.
11050 @node Deprecated Features
11051 @section Deprecated Features
11053 In the past, the GNU C++ compiler was extended to experiment with new
11054 features, at a time when the C++ language was still evolving. Now that
11055 the C++ standard is complete, some of those features are superseded by
11056 superior alternatives. Using the old features might cause a warning in
11057 some cases that the feature will be dropped in the future. In other
11058 cases, the feature might be gone already.
11060 While the list below is not exhaustive, it documents some of the options
11061 that are now deprecated:
11064 @item -fexternal-templates
11065 @itemx -falt-external-templates
11066 These are two of the many ways for G++ to implement template
11067 instantiation. @xref{Template Instantiation}. The C++ standard clearly
11068 defines how template definitions have to be organized across
11069 implementation units. G++ has an implicit instantiation mechanism that
11070 should work just fine for standard-conforming code.
11072 @item -fstrict-prototype
11073 @itemx -fno-strict-prototype
11074 Previously it was possible to use an empty prototype parameter list to
11075 indicate an unspecified number of parameters (like C), rather than no
11076 parameters, as C++ demands. This feature has been removed, except where
11077 it is required for backwards compatibility @xref{Backwards Compatibility}.
11080 G++ allows a virtual function returning @samp{void *} to be overridden
11081 by one returning a different pointer type. This extension to the
11082 covariant return type rules is now deprecated and will be removed from a
11085 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
11086 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
11087 and will be removed in a future version. Code using these operators
11088 should be modified to use @code{std::min} and @code{std::max} instead.
11090 The named return value extension has been deprecated, and is now
11093 The use of initializer lists with new expressions has been deprecated,
11094 and is now removed from G++.
11096 Floating and complex non-type template parameters have been deprecated,
11097 and are now removed from G++.
11099 The implicit typename extension has been deprecated and is now
11102 The use of default arguments in function pointers, function typedefs and
11103 and other places where they are not permitted by the standard is
11104 deprecated and will be removed from a future version of G++.
11106 G++ allows floating-point literals to appear in integral constant expressions,
11107 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
11108 This extension is deprecated and will be removed from a future version.
11110 G++ allows static data members of const floating-point type to be declared
11111 with an initializer in a class definition. The standard only allows
11112 initializers for static members of const integral types and const
11113 enumeration types so this extension has been deprecated and will be removed
11114 from a future version.
11116 @node Backwards Compatibility
11117 @section Backwards Compatibility
11118 @cindex Backwards Compatibility
11119 @cindex ARM [Annotated C++ Reference Manual]
11121 Now that there is a definitive ISO standard C++, G++ has a specification
11122 to adhere to. The C++ language evolved over time, and features that
11123 used to be acceptable in previous drafts of the standard, such as the ARM
11124 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
11125 compilation of C++ written to such drafts, G++ contains some backwards
11126 compatibilities. @emph{All such backwards compatibility features are
11127 liable to disappear in future versions of G++.} They should be considered
11128 deprecated @xref{Deprecated Features}.
11132 If a variable is declared at for scope, it used to remain in scope until
11133 the end of the scope which contained the for statement (rather than just
11134 within the for scope). G++ retains this, but issues a warning, if such a
11135 variable is accessed outside the for scope.
11137 @item Implicit C language
11138 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
11139 scope to set the language. On such systems, all header files are
11140 implicitly scoped inside a C language scope. Also, an empty prototype
11141 @code{()} will be treated as an unspecified number of arguments, rather
11142 than no arguments, as C++ demands.