1 @c Copyright (C) 1988,1989,1992,1993,1994,1996,1998,1999,2000,2001,2002,2003,2004,2005
2 @c Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
7 @chapter Extensions to the C Language Family
8 @cindex extensions, C language
9 @cindex C language extensions
12 GNU C provides several language features not found in ISO standard C@.
13 (The @option{-pedantic} option directs GCC to print a warning message if
14 any of these features is used.) To test for the availability of these
15 features in conditional compilation, check for a predefined macro
16 @code{__GNUC__}, which is always defined under GCC@.
18 These extensions are available in C and Objective-C@. Most of them are
19 also available in C++. @xref{C++ Extensions,,Extensions to the
20 C++ Language}, for extensions that apply @emph{only} to C++.
22 Some features that are in ISO C99 but not C89 or C++ are also, as
23 extensions, accepted by GCC in C89 mode and in C++.
26 * Statement Exprs:: Putting statements and declarations inside expressions.
27 * Local Labels:: Labels local to a block.
28 * Labels as Values:: Getting pointers to labels, and computed gotos.
29 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
30 * Constructing Calls:: Dispatching a call to another function.
31 * Typeof:: @code{typeof}: referring to the type of an expression.
32 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
33 * Long Long:: Double-word integers---@code{long long int}.
34 * Complex:: Data types for complex numbers.
35 * Hex Floats:: Hexadecimal floating-point constants.
36 * Zero Length:: Zero-length arrays.
37 * Variable Length:: Arrays whose length is computed at run time.
38 * Empty Structures:: Structures with no members.
39 * Variadic Macros:: Macros with a variable number of arguments.
40 * Escaped Newlines:: Slightly looser rules for escaped newlines.
41 * Subscripting:: Any array can be subscripted, even if not an lvalue.
42 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
43 * Initializers:: Non-constant initializers.
44 * Compound Literals:: Compound literals give structures, unions
46 * Designated Inits:: Labeling elements of initializers.
47 * Cast to Union:: Casting to union type from any member of the union.
48 * Case Ranges:: `case 1 ... 9' and such.
49 * Mixed Declarations:: Mixing declarations and code.
50 * Function Attributes:: Declaring that functions have no side effects,
51 or that they can never return.
52 * Attribute Syntax:: Formal syntax for attributes.
53 * Function Prototypes:: Prototype declarations and old-style definitions.
54 * C++ Comments:: C++ comments are recognized.
55 * Dollar Signs:: Dollar sign is allowed in identifiers.
56 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
57 * Variable Attributes:: Specifying attributes of variables.
58 * Type Attributes:: Specifying attributes of types.
59 * Alignment:: Inquiring about the alignment of a type or variable.
60 * Inline:: Defining inline functions (as fast as macros).
61 * Extended Asm:: Assembler instructions with C expressions as operands.
62 (With them you can define ``built-in'' functions.)
63 * Constraints:: Constraints for asm operands
64 * Asm Labels:: Specifying the assembler name to use for a C symbol.
65 * Explicit Reg Vars:: Defining variables residing in specified registers.
66 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
67 * Incomplete Enums:: @code{enum foo;}, with details to follow.
68 * Function Names:: Printable strings which are the name of the current
70 * Return Address:: Getting the return or frame address of a function.
71 * Vector Extensions:: Using vector instructions through built-in functions.
72 * Offsetof:: Special syntax for implementing @code{offsetof}.
73 * Atomic Builtins:: Built-in functions for atomic memory access.
74 * Other Builtins:: Other built-in functions.
75 * Target Builtins:: Built-in functions specific to particular targets.
76 * Target Format Checks:: Format checks specific to particular targets.
77 * Pragmas:: Pragmas accepted by GCC.
78 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
79 * Thread-Local:: Per-thread variables.
83 @section Statements and Declarations in Expressions
84 @cindex statements inside expressions
85 @cindex declarations inside expressions
86 @cindex expressions containing statements
87 @cindex macros, statements in expressions
89 @c the above section title wrapped and causes an underfull hbox.. i
90 @c changed it from "within" to "in". --mew 4feb93
91 A compound statement enclosed in parentheses may appear as an expression
92 in GNU C@. This allows you to use loops, switches, and local variables
95 Recall that a compound statement is a sequence of statements surrounded
96 by braces; in this construct, parentheses go around the braces. For
100 (@{ int y = foo (); int z;
107 is a valid (though slightly more complex than necessary) expression
108 for the absolute value of @code{foo ()}.
110 The last thing in the compound statement should be an expression
111 followed by a semicolon; the value of this subexpression serves as the
112 value of the entire construct. (If you use some other kind of statement
113 last within the braces, the construct has type @code{void}, and thus
114 effectively no value.)
116 This feature is especially useful in making macro definitions ``safe'' (so
117 that they evaluate each operand exactly once). For example, the
118 ``maximum'' function is commonly defined as a macro in standard C as
122 #define max(a,b) ((a) > (b) ? (a) : (b))
126 @cindex side effects, macro argument
127 But this definition computes either @var{a} or @var{b} twice, with bad
128 results if the operand has side effects. In GNU C, if you know the
129 type of the operands (here taken as @code{int}), you can define
130 the macro safely as follows:
133 #define maxint(a,b) \
134 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
137 Embedded statements are not allowed in constant expressions, such as
138 the value of an enumeration constant, the width of a bit-field, or
139 the initial value of a static variable.
141 If you don't know the type of the operand, you can still do this, but you
142 must use @code{typeof} (@pxref{Typeof}).
144 In G++, the result value of a statement expression undergoes array and
145 function pointer decay, and is returned by value to the enclosing
146 expression. For instance, if @code{A} is a class, then
155 will construct a temporary @code{A} object to hold the result of the
156 statement expression, and that will be used to invoke @code{Foo}.
157 Therefore the @code{this} pointer observed by @code{Foo} will not be the
160 Any temporaries created within a statement within a statement expression
161 will be destroyed at the statement's end. This makes statement
162 expressions inside macros slightly different from function calls. In
163 the latter case temporaries introduced during argument evaluation will
164 be destroyed at the end of the statement that includes the function
165 call. In the statement expression case they will be destroyed during
166 the statement expression. For instance,
169 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
170 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
180 will have different places where temporaries are destroyed. For the
181 @code{macro} case, the temporary @code{X} will be destroyed just after
182 the initialization of @code{b}. In the @code{function} case that
183 temporary will be destroyed when the function returns.
185 These considerations mean that it is probably a bad idea to use
186 statement-expressions of this form in header files that are designed to
187 work with C++. (Note that some versions of the GNU C Library contained
188 header files using statement-expression that lead to precisely this
191 Jumping into a statement expression with @code{goto} or using a
192 @code{switch} statement outside the statement expression with a
193 @code{case} or @code{default} label inside the statement expression is
194 not permitted. Jumping into a statement expression with a computed
195 @code{goto} (@pxref{Labels as Values}) yields undefined behavior.
196 Jumping out of a statement expression is permitted, but if the
197 statement expression is part of a larger expression then it is
198 unspecified which other subexpressions of that expression have been
199 evaluated except where the language definition requires certain
200 subexpressions to be evaluated before or after the statement
201 expression. In any case, as with a function call the evaluation of a
202 statement expression is not interleaved with the evaluation of other
203 parts of the containing expression. For example,
206 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
210 will call @code{foo} and @code{bar1} and will not call @code{baz} but
211 may or may not call @code{bar2}. If @code{bar2} is called, it will be
212 called after @code{foo} and before @code{bar1}
215 @section Locally Declared Labels
217 @cindex macros, local labels
219 GCC allows you to declare @dfn{local labels} in any nested block
220 scope. A local label is just like an ordinary label, but you can
221 only reference it (with a @code{goto} statement, or by taking its
222 address) within the block in which it was declared.
224 A local label declaration looks like this:
227 __label__ @var{label};
234 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
237 Local label declarations must come at the beginning of the block,
238 before any ordinary declarations or statements.
240 The label declaration defines the label @emph{name}, but does not define
241 the label itself. You must do this in the usual way, with
242 @code{@var{label}:}, within the statements of the statement expression.
244 The local label feature is useful for complex macros. If a macro
245 contains nested loops, a @code{goto} can be useful for breaking out of
246 them. However, an ordinary label whose scope is the whole function
247 cannot be used: if the macro can be expanded several times in one
248 function, the label will be multiply defined in that function. A
249 local label avoids this problem. For example:
252 #define SEARCH(value, array, target) \
255 typeof (target) _SEARCH_target = (target); \
256 typeof (*(array)) *_SEARCH_array = (array); \
259 for (i = 0; i < max; i++) \
260 for (j = 0; j < max; j++) \
261 if (_SEARCH_array[i][j] == _SEARCH_target) \
262 @{ (value) = i; goto found; @} \
268 This could also be written using a statement-expression:
271 #define SEARCH(array, target) \
274 typeof (target) _SEARCH_target = (target); \
275 typeof (*(array)) *_SEARCH_array = (array); \
278 for (i = 0; i < max; i++) \
279 for (j = 0; j < max; j++) \
280 if (_SEARCH_array[i][j] == _SEARCH_target) \
281 @{ value = i; goto found; @} \
288 Local label declarations also make the labels they declare visible to
289 nested functions, if there are any. @xref{Nested Functions}, for details.
291 @node Labels as Values
292 @section Labels as Values
293 @cindex labels as values
294 @cindex computed gotos
295 @cindex goto with computed label
296 @cindex address of a label
298 You can get the address of a label defined in the current function
299 (or a containing function) with the unary operator @samp{&&}. The
300 value has type @code{void *}. This value is a constant and can be used
301 wherever a constant of that type is valid. For example:
309 To use these values, you need to be able to jump to one. This is done
310 with the computed goto statement@footnote{The analogous feature in
311 Fortran is called an assigned goto, but that name seems inappropriate in
312 C, where one can do more than simply store label addresses in label
313 variables.}, @code{goto *@var{exp};}. For example,
320 Any expression of type @code{void *} is allowed.
322 One way of using these constants is in initializing a static array that
323 will serve as a jump table:
326 static void *array[] = @{ &&foo, &&bar, &&hack @};
329 Then you can select a label with indexing, like this:
336 Note that this does not check whether the subscript is in bounds---array
337 indexing in C never does that.
339 Such an array of label values serves a purpose much like that of the
340 @code{switch} statement. The @code{switch} statement is cleaner, so
341 use that rather than an array unless the problem does not fit a
342 @code{switch} statement very well.
344 Another use of label values is in an interpreter for threaded code.
345 The labels within the interpreter function can be stored in the
346 threaded code for super-fast dispatching.
348 You may not use this mechanism to jump to code in a different function.
349 If you do that, totally unpredictable things will happen. The best way to
350 avoid this is to store the label address only in automatic variables and
351 never pass it as an argument.
353 An alternate way to write the above example is
356 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
358 goto *(&&foo + array[i]);
362 This is more friendly to code living in shared libraries, as it reduces
363 the number of dynamic relocations that are needed, and by consequence,
364 allows the data to be read-only.
366 @node Nested Functions
367 @section Nested Functions
368 @cindex nested functions
369 @cindex downward funargs
372 A @dfn{nested function} is a function defined inside another function.
373 (Nested functions are not supported for GNU C++.) The nested function's
374 name is local to the block where it is defined. For example, here we
375 define a nested function named @code{square}, and call it twice:
379 foo (double a, double b)
381 double square (double z) @{ return z * z; @}
383 return square (a) + square (b);
388 The nested function can access all the variables of the containing
389 function that are visible at the point of its definition. This is
390 called @dfn{lexical scoping}. For example, here we show a nested
391 function which uses an inherited variable named @code{offset}:
395 bar (int *array, int offset, int size)
397 int access (int *array, int index)
398 @{ return array[index + offset]; @}
401 for (i = 0; i < size; i++)
402 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
407 Nested function definitions are permitted within functions in the places
408 where variable definitions are allowed; that is, in any block, mixed
409 with the other declarations and statements in the block.
411 It is possible to call the nested function from outside the scope of its
412 name by storing its address or passing the address to another function:
415 hack (int *array, int size)
417 void store (int index, int value)
418 @{ array[index] = value; @}
420 intermediate (store, size);
424 Here, the function @code{intermediate} receives the address of
425 @code{store} as an argument. If @code{intermediate} calls @code{store},
426 the arguments given to @code{store} are used to store into @code{array}.
427 But this technique works only so long as the containing function
428 (@code{hack}, in this example) does not exit.
430 If you try to call the nested function through its address after the
431 containing function has exited, all hell will break loose. If you try
432 to call it after a containing scope level has exited, and if it refers
433 to some of the variables that are no longer in scope, you may be lucky,
434 but it's not wise to take the risk. If, however, the nested function
435 does not refer to anything that has gone out of scope, you should be
438 GCC implements taking the address of a nested function using a technique
439 called @dfn{trampolines}. A paper describing them is available as
442 @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}.
444 A nested function can jump to a label inherited from a containing
445 function, provided the label was explicitly declared in the containing
446 function (@pxref{Local Labels}). Such a jump returns instantly to the
447 containing function, exiting the nested function which did the
448 @code{goto} and any intermediate functions as well. Here is an example:
452 bar (int *array, int offset, int size)
455 int access (int *array, int index)
459 return array[index + offset];
463 for (i = 0; i < size; i++)
464 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
468 /* @r{Control comes here from @code{access}
469 if it detects an error.} */
476 A nested function always has no linkage. Declaring one with
477 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
478 before its definition, use @code{auto} (which is otherwise meaningless
479 for function declarations).
482 bar (int *array, int offset, int size)
485 auto int access (int *, int);
487 int access (int *array, int index)
491 return array[index + offset];
497 @node Constructing Calls
498 @section Constructing Function Calls
499 @cindex constructing calls
500 @cindex forwarding calls
502 Using the built-in functions described below, you can record
503 the arguments a function received, and call another function
504 with the same arguments, without knowing the number or types
507 You can also record the return value of that function call,
508 and later return that value, without knowing what data type
509 the function tried to return (as long as your caller expects
512 However, these built-in functions may interact badly with some
513 sophisticated features or other extensions of the language. It
514 is, therefore, not recommended to use them outside very simple
515 functions acting as mere forwarders for their arguments.
517 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
518 This built-in function returns a pointer to data
519 describing how to perform a call with the same arguments as were passed
520 to the current function.
522 The function saves the arg pointer register, structure value address,
523 and all registers that might be used to pass arguments to a function
524 into a block of memory allocated on the stack. Then it returns the
525 address of that block.
528 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
529 This built-in function invokes @var{function}
530 with a copy of the parameters described by @var{arguments}
533 The value of @var{arguments} should be the value returned by
534 @code{__builtin_apply_args}. The argument @var{size} specifies the size
535 of the stack argument data, in bytes.
537 This function returns a pointer to data describing
538 how to return whatever value was returned by @var{function}. The data
539 is saved in a block of memory allocated on the stack.
541 It is not always simple to compute the proper value for @var{size}. The
542 value is used by @code{__builtin_apply} to compute the amount of data
543 that should be pushed on the stack and copied from the incoming argument
547 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
548 This built-in function returns the value described by @var{result} from
549 the containing function. You should specify, for @var{result}, a value
550 returned by @code{__builtin_apply}.
554 @section Referring to a Type with @code{typeof}
557 @cindex macros, types of arguments
559 Another way to refer to the type of an expression is with @code{typeof}.
560 The syntax of using of this keyword looks like @code{sizeof}, but the
561 construct acts semantically like a type name defined with @code{typedef}.
563 There are two ways of writing the argument to @code{typeof}: with an
564 expression or with a type. Here is an example with an expression:
571 This assumes that @code{x} is an array of pointers to functions;
572 the type described is that of the values of the functions.
574 Here is an example with a typename as the argument:
581 Here the type described is that of pointers to @code{int}.
583 If you are writing a header file that must work when included in ISO C
584 programs, write @code{__typeof__} instead of @code{typeof}.
585 @xref{Alternate Keywords}.
587 A @code{typeof}-construct can be used anywhere a typedef name could be
588 used. For example, you can use it in a declaration, in a cast, or inside
589 of @code{sizeof} or @code{typeof}.
591 @code{typeof} is often useful in conjunction with the
592 statements-within-expressions feature. Here is how the two together can
593 be used to define a safe ``maximum'' macro that operates on any
594 arithmetic type and evaluates each of its arguments exactly once:
598 (@{ typeof (a) _a = (a); \
599 typeof (b) _b = (b); \
600 _a > _b ? _a : _b; @})
603 @cindex underscores in variables in macros
604 @cindex @samp{_} in variables in macros
605 @cindex local variables in macros
606 @cindex variables, local, in macros
607 @cindex macros, local variables in
609 The reason for using names that start with underscores for the local
610 variables is to avoid conflicts with variable names that occur within the
611 expressions that are substituted for @code{a} and @code{b}. Eventually we
612 hope to design a new form of declaration syntax that allows you to declare
613 variables whose scopes start only after their initializers; this will be a
614 more reliable way to prevent such conflicts.
617 Some more examples of the use of @code{typeof}:
621 This declares @code{y} with the type of what @code{x} points to.
628 This declares @code{y} as an array of such values.
635 This declares @code{y} as an array of pointers to characters:
638 typeof (typeof (char *)[4]) y;
642 It is equivalent to the following traditional C declaration:
648 To see the meaning of the declaration using @code{typeof}, and why it
649 might be a useful way to write, rewrite it with these macros:
652 #define pointer(T) typeof(T *)
653 #define array(T, N) typeof(T [N])
657 Now the declaration can be rewritten this way:
660 array (pointer (char), 4) y;
664 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
665 pointers to @code{char}.
668 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
669 a more limited extension which permitted one to write
672 typedef @var{T} = @var{expr};
676 with the effect of declaring @var{T} to have the type of the expression
677 @var{expr}. This extension does not work with GCC 3 (versions between
678 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
679 relies on it should be rewritten to use @code{typeof}:
682 typedef typeof(@var{expr}) @var{T};
686 This will work with all versions of GCC@.
689 @section Conditionals with Omitted Operands
690 @cindex conditional expressions, extensions
691 @cindex omitted middle-operands
692 @cindex middle-operands, omitted
693 @cindex extensions, @code{?:}
694 @cindex @code{?:} extensions
696 The middle operand in a conditional expression may be omitted. Then
697 if the first operand is nonzero, its value is the value of the conditional
700 Therefore, the expression
707 has the value of @code{x} if that is nonzero; otherwise, the value of
710 This example is perfectly equivalent to
716 @cindex side effect in ?:
717 @cindex ?: side effect
719 In this simple case, the ability to omit the middle operand is not
720 especially useful. When it becomes useful is when the first operand does,
721 or may (if it is a macro argument), contain a side effect. Then repeating
722 the operand in the middle would perform the side effect twice. Omitting
723 the middle operand uses the value already computed without the undesirable
724 effects of recomputing it.
727 @section Double-Word Integers
728 @cindex @code{long long} data types
729 @cindex double-word arithmetic
730 @cindex multiprecision arithmetic
731 @cindex @code{LL} integer suffix
732 @cindex @code{ULL} integer suffix
734 ISO C99 supports data types for integers that are at least 64 bits wide,
735 and as an extension GCC supports them in C89 mode and in C++.
736 Simply write @code{long long int} for a signed integer, or
737 @code{unsigned long long int} for an unsigned integer. To make an
738 integer constant of type @code{long long int}, add the suffix @samp{LL}
739 to the integer. To make an integer constant of type @code{unsigned long
740 long int}, add the suffix @samp{ULL} to the integer.
742 You can use these types in arithmetic like any other integer types.
743 Addition, subtraction, and bitwise boolean operations on these types
744 are open-coded on all types of machines. Multiplication is open-coded
745 if the machine supports fullword-to-doubleword a widening multiply
746 instruction. Division and shifts are open-coded only on machines that
747 provide special support. The operations that are not open-coded use
748 special library routines that come with GCC@.
750 There may be pitfalls when you use @code{long long} types for function
751 arguments, unless you declare function prototypes. If a function
752 expects type @code{int} for its argument, and you pass a value of type
753 @code{long long int}, confusion will result because the caller and the
754 subroutine will disagree about the number of bytes for the argument.
755 Likewise, if the function expects @code{long long int} and you pass
756 @code{int}. The best way to avoid such problems is to use prototypes.
759 @section Complex Numbers
760 @cindex complex numbers
761 @cindex @code{_Complex} keyword
762 @cindex @code{__complex__} keyword
764 ISO C99 supports complex floating data types, and as an extension GCC
765 supports them in C89 mode and in C++, and supports complex integer data
766 types which are not part of ISO C99. You can declare complex types
767 using the keyword @code{_Complex}. As an extension, the older GNU
768 keyword @code{__complex__} is also supported.
770 For example, @samp{_Complex double x;} declares @code{x} as a
771 variable whose real part and imaginary part are both of type
772 @code{double}. @samp{_Complex short int y;} declares @code{y} to
773 have real and imaginary parts of type @code{short int}; this is not
774 likely to be useful, but it shows that the set of complex types is
777 To write a constant with a complex data type, use the suffix @samp{i} or
778 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
779 has type @code{_Complex float} and @code{3i} has type
780 @code{_Complex int}. Such a constant always has a pure imaginary
781 value, but you can form any complex value you like by adding one to a
782 real constant. This is a GNU extension; if you have an ISO C99
783 conforming C library (such as GNU libc), and want to construct complex
784 constants of floating type, you should include @code{<complex.h>} and
785 use the macros @code{I} or @code{_Complex_I} instead.
787 @cindex @code{__real__} keyword
788 @cindex @code{__imag__} keyword
789 To extract the real part of a complex-valued expression @var{exp}, write
790 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
791 extract the imaginary part. This is a GNU extension; for values of
792 floating type, you should use the ISO C99 functions @code{crealf},
793 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
794 @code{cimagl}, declared in @code{<complex.h>} and also provided as
795 built-in functions by GCC@.
797 @cindex complex conjugation
798 The operator @samp{~} performs complex conjugation when used on a value
799 with a complex type. This is a GNU extension; for values of
800 floating type, you should use the ISO C99 functions @code{conjf},
801 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
802 provided as built-in functions by GCC@.
804 GCC can allocate complex automatic variables in a noncontiguous
805 fashion; it's even possible for the real part to be in a register while
806 the imaginary part is on the stack (or vice-versa). Only the DWARF2
807 debug info format can represent this, so use of DWARF2 is recommended.
808 If you are using the stabs debug info format, GCC describes a noncontiguous
809 complex variable as if it were two separate variables of noncomplex type.
810 If the variable's actual name is @code{foo}, the two fictitious
811 variables are named @code{foo$real} and @code{foo$imag}. You can
812 examine and set these two fictitious variables with your debugger.
818 ISO C99 supports floating-point numbers written not only in the usual
819 decimal notation, such as @code{1.55e1}, but also numbers such as
820 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
821 supports this in C89 mode (except in some cases when strictly
822 conforming) and in C++. In that format the
823 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
824 mandatory. The exponent is a decimal number that indicates the power of
825 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
832 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
833 is the same as @code{1.55e1}.
835 Unlike for floating-point numbers in the decimal notation the exponent
836 is always required in the hexadecimal notation. Otherwise the compiler
837 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
838 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
839 extension for floating-point constants of type @code{float}.
842 @section Arrays of Length Zero
843 @cindex arrays of length zero
844 @cindex zero-length arrays
845 @cindex length-zero arrays
846 @cindex flexible array members
848 Zero-length arrays are allowed in GNU C@. They are very useful as the
849 last element of a structure which is really a header for a variable-length
858 struct line *thisline = (struct line *)
859 malloc (sizeof (struct line) + this_length);
860 thisline->length = this_length;
863 In ISO C90, you would have to give @code{contents} a length of 1, which
864 means either you waste space or complicate the argument to @code{malloc}.
866 In ISO C99, you would use a @dfn{flexible array member}, which is
867 slightly different in syntax and semantics:
871 Flexible array members are written as @code{contents[]} without
875 Flexible array members have incomplete type, and so the @code{sizeof}
876 operator may not be applied. As a quirk of the original implementation
877 of zero-length arrays, @code{sizeof} evaluates to zero.
880 Flexible array members may only appear as the last member of a
881 @code{struct} that is otherwise non-empty.
884 A structure containing a flexible array member, or a union containing
885 such a structure (possibly recursively), may not be a member of a
886 structure or an element of an array. (However, these uses are
887 permitted by GCC as extensions.)
890 GCC versions before 3.0 allowed zero-length arrays to be statically
891 initialized, as if they were flexible arrays. In addition to those
892 cases that were useful, it also allowed initializations in situations
893 that would corrupt later data. Non-empty initialization of zero-length
894 arrays is now treated like any case where there are more initializer
895 elements than the array holds, in that a suitable warning about "excess
896 elements in array" is given, and the excess elements (all of them, in
897 this case) are ignored.
899 Instead GCC allows static initialization of flexible array members.
900 This is equivalent to defining a new structure containing the original
901 structure followed by an array of sufficient size to contain the data.
902 I.e.@: in the following, @code{f1} is constructed as if it were declared
908 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
911 struct f1 f1; int data[3];
912 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
916 The convenience of this extension is that @code{f1} has the desired
917 type, eliminating the need to consistently refer to @code{f2.f1}.
919 This has symmetry with normal static arrays, in that an array of
920 unknown size is also written with @code{[]}.
922 Of course, this extension only makes sense if the extra data comes at
923 the end of a top-level object, as otherwise we would be overwriting
924 data at subsequent offsets. To avoid undue complication and confusion
925 with initialization of deeply nested arrays, we simply disallow any
926 non-empty initialization except when the structure is the top-level
930 struct foo @{ int x; int y[]; @};
931 struct bar @{ struct foo z; @};
933 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
934 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
935 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
936 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
939 @node Empty Structures
940 @section Structures With No Members
941 @cindex empty structures
942 @cindex zero-size structures
944 GCC permits a C structure to have no members:
951 The structure will have size zero. In C++, empty structures are part
952 of the language. G++ treats empty structures as if they had a single
953 member of type @code{char}.
955 @node Variable Length
956 @section Arrays of Variable Length
957 @cindex variable-length arrays
958 @cindex arrays of variable length
961 Variable-length automatic arrays are allowed in ISO C99, and as an
962 extension GCC accepts them in C89 mode and in C++. (However, GCC's
963 implementation of variable-length arrays does not yet conform in detail
964 to the ISO C99 standard.) These arrays are
965 declared like any other automatic arrays, but with a length that is not
966 a constant expression. The storage is allocated at the point of
967 declaration and deallocated when the brace-level is exited. For
972 concat_fopen (char *s1, char *s2, char *mode)
974 char str[strlen (s1) + strlen (s2) + 1];
977 return fopen (str, mode);
981 @cindex scope of a variable length array
982 @cindex variable-length array scope
983 @cindex deallocating variable length arrays
984 Jumping or breaking out of the scope of the array name deallocates the
985 storage. Jumping into the scope is not allowed; you get an error
988 @cindex @code{alloca} vs variable-length arrays
989 You can use the function @code{alloca} to get an effect much like
990 variable-length arrays. The function @code{alloca} is available in
991 many other C implementations (but not in all). On the other hand,
992 variable-length arrays are more elegant.
994 There are other differences between these two methods. Space allocated
995 with @code{alloca} exists until the containing @emph{function} returns.
996 The space for a variable-length array is deallocated as soon as the array
997 name's scope ends. (If you use both variable-length arrays and
998 @code{alloca} in the same function, deallocation of a variable-length array
999 will also deallocate anything more recently allocated with @code{alloca}.)
1001 You can also use variable-length arrays as arguments to functions:
1005 tester (int len, char data[len][len])
1011 The length of an array is computed once when the storage is allocated
1012 and is remembered for the scope of the array in case you access it with
1015 If you want to pass the array first and the length afterward, you can
1016 use a forward declaration in the parameter list---another GNU extension.
1020 tester (int len; char data[len][len], int len)
1026 @cindex parameter forward declaration
1027 The @samp{int len} before the semicolon is a @dfn{parameter forward
1028 declaration}, and it serves the purpose of making the name @code{len}
1029 known when the declaration of @code{data} is parsed.
1031 You can write any number of such parameter forward declarations in the
1032 parameter list. They can be separated by commas or semicolons, but the
1033 last one must end with a semicolon, which is followed by the ``real''
1034 parameter declarations. Each forward declaration must match a ``real''
1035 declaration in parameter name and data type. ISO C99 does not support
1036 parameter forward declarations.
1038 @node Variadic Macros
1039 @section Macros with a Variable Number of Arguments.
1040 @cindex variable number of arguments
1041 @cindex macro with variable arguments
1042 @cindex rest argument (in macro)
1043 @cindex variadic macros
1045 In the ISO C standard of 1999, a macro can be declared to accept a
1046 variable number of arguments much as a function can. The syntax for
1047 defining the macro is similar to that of a function. Here is an
1051 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1054 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1055 such a macro, it represents the zero or more tokens until the closing
1056 parenthesis that ends the invocation, including any commas. This set of
1057 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1058 wherever it appears. See the CPP manual for more information.
1060 GCC has long supported variadic macros, and used a different syntax that
1061 allowed you to give a name to the variable arguments just like any other
1062 argument. Here is an example:
1065 #define debug(format, args...) fprintf (stderr, format, args)
1068 This is in all ways equivalent to the ISO C example above, but arguably
1069 more readable and descriptive.
1071 GNU CPP has two further variadic macro extensions, and permits them to
1072 be used with either of the above forms of macro definition.
1074 In standard C, you are not allowed to leave the variable argument out
1075 entirely; but you are allowed to pass an empty argument. For example,
1076 this invocation is invalid in ISO C, because there is no comma after
1083 GNU CPP permits you to completely omit the variable arguments in this
1084 way. In the above examples, the compiler would complain, though since
1085 the expansion of the macro still has the extra comma after the format
1088 To help solve this problem, CPP behaves specially for variable arguments
1089 used with the token paste operator, @samp{##}. If instead you write
1092 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1095 and if the variable arguments are omitted or empty, the @samp{##}
1096 operator causes the preprocessor to remove the comma before it. If you
1097 do provide some variable arguments in your macro invocation, GNU CPP
1098 does not complain about the paste operation and instead places the
1099 variable arguments after the comma. Just like any other pasted macro
1100 argument, these arguments are not macro expanded.
1102 @node Escaped Newlines
1103 @section Slightly Looser Rules for Escaped Newlines
1104 @cindex escaped newlines
1105 @cindex newlines (escaped)
1107 Recently, the preprocessor has relaxed its treatment of escaped
1108 newlines. Previously, the newline had to immediately follow a
1109 backslash. The current implementation allows whitespace in the form
1110 of spaces, horizontal and vertical tabs, and form feeds between the
1111 backslash and the subsequent newline. The preprocessor issues a
1112 warning, but treats it as a valid escaped newline and combines the two
1113 lines to form a single logical line. This works within comments and
1114 tokens, as well as between tokens. Comments are @emph{not} treated as
1115 whitespace for the purposes of this relaxation, since they have not
1116 yet been replaced with spaces.
1119 @section Non-Lvalue Arrays May Have Subscripts
1120 @cindex subscripting
1121 @cindex arrays, non-lvalue
1123 @cindex subscripting and function values
1124 In ISO C99, arrays that are not lvalues still decay to pointers, and
1125 may be subscripted, although they may not be modified or used after
1126 the next sequence point and the unary @samp{&} operator may not be
1127 applied to them. As an extension, GCC allows such arrays to be
1128 subscripted in C89 mode, though otherwise they do not decay to
1129 pointers outside C99 mode. For example,
1130 this is valid in GNU C though not valid in C89:
1134 struct foo @{int a[4];@};
1140 return f().a[index];
1146 @section Arithmetic on @code{void}- and Function-Pointers
1147 @cindex void pointers, arithmetic
1148 @cindex void, size of pointer to
1149 @cindex function pointers, arithmetic
1150 @cindex function, size of pointer to
1152 In GNU C, addition and subtraction operations are supported on pointers to
1153 @code{void} and on pointers to functions. This is done by treating the
1154 size of a @code{void} or of a function as 1.
1156 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1157 and on function types, and returns 1.
1159 @opindex Wpointer-arith
1160 The option @option{-Wpointer-arith} requests a warning if these extensions
1164 @section Non-Constant Initializers
1165 @cindex initializers, non-constant
1166 @cindex non-constant initializers
1168 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1169 automatic variable are not required to be constant expressions in GNU C@.
1170 Here is an example of an initializer with run-time varying elements:
1173 foo (float f, float g)
1175 float beat_freqs[2] = @{ f-g, f+g @};
1180 @node Compound Literals
1181 @section Compound Literals
1182 @cindex constructor expressions
1183 @cindex initializations in expressions
1184 @cindex structures, constructor expression
1185 @cindex expressions, constructor
1186 @cindex compound literals
1187 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1189 ISO C99 supports compound literals. A compound literal looks like
1190 a cast containing an initializer. Its value is an object of the
1191 type specified in the cast, containing the elements specified in
1192 the initializer; it is an lvalue. As an extension, GCC supports
1193 compound literals in C89 mode and in C++.
1195 Usually, the specified type is a structure. Assume that
1196 @code{struct foo} and @code{structure} are declared as shown:
1199 struct foo @{int a; char b[2];@} structure;
1203 Here is an example of constructing a @code{struct foo} with a compound literal:
1206 structure = ((struct foo) @{x + y, 'a', 0@});
1210 This is equivalent to writing the following:
1214 struct foo temp = @{x + y, 'a', 0@};
1219 You can also construct an array. If all the elements of the compound literal
1220 are (made up of) simple constant expressions, suitable for use in
1221 initializers of objects of static storage duration, then the compound
1222 literal can be coerced to a pointer to its first element and used in
1223 such an initializer, as shown here:
1226 char **foo = (char *[]) @{ "x", "y", "z" @};
1229 Compound literals for scalar types and union types are is
1230 also allowed, but then the compound literal is equivalent
1233 As a GNU extension, GCC allows initialization of objects with static storage
1234 duration by compound literals (which is not possible in ISO C99, because
1235 the initializer is not a constant).
1236 It is handled as if the object was initialized only with the bracket
1237 enclosed list if compound literal's and object types match.
1238 The initializer list of the compound literal must be constant.
1239 If the object being initialized has array type of unknown size, the size is
1240 determined by compound literal size.
1243 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1244 static int y[] = (int []) @{1, 2, 3@};
1245 static int z[] = (int [3]) @{1@};
1249 The above lines are equivalent to the following:
1251 static struct foo x = @{1, 'a', 'b'@};
1252 static int y[] = @{1, 2, 3@};
1253 static int z[] = @{1, 0, 0@};
1256 @node Designated Inits
1257 @section Designated Initializers
1258 @cindex initializers with labeled elements
1259 @cindex labeled elements in initializers
1260 @cindex case labels in initializers
1261 @cindex designated initializers
1263 Standard C89 requires the elements of an initializer to appear in a fixed
1264 order, the same as the order of the elements in the array or structure
1267 In ISO C99 you can give the elements in any order, specifying the array
1268 indices or structure field names they apply to, and GNU C allows this as
1269 an extension in C89 mode as well. This extension is not
1270 implemented in GNU C++.
1272 To specify an array index, write
1273 @samp{[@var{index}] =} before the element value. For example,
1276 int a[6] = @{ [4] = 29, [2] = 15 @};
1283 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1287 The index values must be constant expressions, even if the array being
1288 initialized is automatic.
1290 An alternative syntax for this which has been obsolete since GCC 2.5 but
1291 GCC still accepts is to write @samp{[@var{index}]} before the element
1292 value, with no @samp{=}.
1294 To initialize a range of elements to the same value, write
1295 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1296 extension. For example,
1299 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1303 If the value in it has side-effects, the side-effects will happen only once,
1304 not for each initialized field by the range initializer.
1307 Note that the length of the array is the highest value specified
1310 In a structure initializer, specify the name of a field to initialize
1311 with @samp{.@var{fieldname} =} before the element value. For example,
1312 given the following structure,
1315 struct point @{ int x, y; @};
1319 the following initialization
1322 struct point p = @{ .y = yvalue, .x = xvalue @};
1329 struct point p = @{ xvalue, yvalue @};
1332 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1333 @samp{@var{fieldname}:}, as shown here:
1336 struct point p = @{ y: yvalue, x: xvalue @};
1340 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1341 @dfn{designator}. You can also use a designator (or the obsolete colon
1342 syntax) when initializing a union, to specify which element of the union
1343 should be used. For example,
1346 union foo @{ int i; double d; @};
1348 union foo f = @{ .d = 4 @};
1352 will convert 4 to a @code{double} to store it in the union using
1353 the second element. By contrast, casting 4 to type @code{union foo}
1354 would store it into the union as the integer @code{i}, since it is
1355 an integer. (@xref{Cast to Union}.)
1357 You can combine this technique of naming elements with ordinary C
1358 initialization of successive elements. Each initializer element that
1359 does not have a designator applies to the next consecutive element of the
1360 array or structure. For example,
1363 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1370 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1373 Labeling the elements of an array initializer is especially useful
1374 when the indices are characters or belong to an @code{enum} type.
1379 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1380 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1383 @cindex designator lists
1384 You can also write a series of @samp{.@var{fieldname}} and
1385 @samp{[@var{index}]} designators before an @samp{=} to specify a
1386 nested subobject to initialize; the list is taken relative to the
1387 subobject corresponding to the closest surrounding brace pair. For
1388 example, with the @samp{struct point} declaration above:
1391 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1395 If the same field is initialized multiple times, it will have value from
1396 the last initialization. If any such overridden initialization has
1397 side-effect, it is unspecified whether the side-effect happens or not.
1398 Currently, GCC will discard them and issue a warning.
1401 @section Case Ranges
1403 @cindex ranges in case statements
1405 You can specify a range of consecutive values in a single @code{case} label,
1409 case @var{low} ... @var{high}:
1413 This has the same effect as the proper number of individual @code{case}
1414 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1416 This feature is especially useful for ranges of ASCII character codes:
1422 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1423 it may be parsed wrong when you use it with integer values. For example,
1438 @section Cast to a Union Type
1439 @cindex cast to a union
1440 @cindex union, casting to a
1442 A cast to union type is similar to other casts, except that the type
1443 specified is a union type. You can specify the type either with
1444 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1445 a constructor though, not a cast, and hence does not yield an lvalue like
1446 normal casts. (@xref{Compound Literals}.)
1448 The types that may be cast to the union type are those of the members
1449 of the union. Thus, given the following union and variables:
1452 union foo @{ int i; double d; @};
1458 both @code{x} and @code{y} can be cast to type @code{union foo}.
1460 Using the cast as the right-hand side of an assignment to a variable of
1461 union type is equivalent to storing in a member of the union:
1466 u = (union foo) x @equiv{} u.i = x
1467 u = (union foo) y @equiv{} u.d = y
1470 You can also use the union cast as a function argument:
1473 void hack (union foo);
1475 hack ((union foo) x);
1478 @node Mixed Declarations
1479 @section Mixed Declarations and Code
1480 @cindex mixed declarations and code
1481 @cindex declarations, mixed with code
1482 @cindex code, mixed with declarations
1484 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1485 within compound statements. As an extension, GCC also allows this in
1486 C89 mode. For example, you could do:
1495 Each identifier is visible from where it is declared until the end of
1496 the enclosing block.
1498 @node Function Attributes
1499 @section Declaring Attributes of Functions
1500 @cindex function attributes
1501 @cindex declaring attributes of functions
1502 @cindex functions that never return
1503 @cindex functions that return more than once
1504 @cindex functions that have no side effects
1505 @cindex functions in arbitrary sections
1506 @cindex functions that behave like malloc
1507 @cindex @code{volatile} applied to function
1508 @cindex @code{const} applied to function
1509 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1510 @cindex functions with non-null pointer arguments
1511 @cindex functions that are passed arguments in registers on the 386
1512 @cindex functions that pop the argument stack on the 386
1513 @cindex functions that do not pop the argument stack on the 386
1515 In GNU C, you declare certain things about functions called in your program
1516 which help the compiler optimize function calls and check your code more
1519 The keyword @code{__attribute__} allows you to specify special
1520 attributes when making a declaration. This keyword is followed by an
1521 attribute specification inside double parentheses. The following
1522 attributes are currently defined for functions on all targets:
1523 @code{noreturn}, @code{returns_twice}, @code{noinline}, @code{always_inline},
1524 @code{pure}, @code{const}, @code{nothrow}, @code{sentinel},
1525 @code{format}, @code{format_arg}, @code{no_instrument_function},
1526 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1527 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
1528 @code{alias}, @code{warn_unused_result} and @code{nonnull}. Several other
1529 attributes are defined for functions on particular target systems. Other
1530 attributes, including @code{section} are supported for variables declarations
1531 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1533 You may also specify attributes with @samp{__} preceding and following
1534 each keyword. This allows you to use them in header files without
1535 being concerned about a possible macro of the same name. For example,
1536 you may use @code{__noreturn__} instead of @code{noreturn}.
1538 @xref{Attribute Syntax}, for details of the exact syntax for using
1542 @c Keep this table alphabetized by attribute name. Treat _ as space.
1544 @item alias ("@var{target}")
1545 @cindex @code{alias} attribute
1546 The @code{alias} attribute causes the declaration to be emitted as an
1547 alias for another symbol, which must be specified. For instance,
1550 void __f () @{ /* @r{Do something.} */; @}
1551 void f () __attribute__ ((weak, alias ("__f")));
1554 declares @samp{f} to be a weak alias for @samp{__f}. In C++, the
1555 mangled name for the target must be used. It is an error if @samp{__f}
1556 is not defined in the same translation unit.
1558 Not all target machines support this attribute.
1561 @cindex @code{always_inline} function attribute
1562 Generally, functions are not inlined unless optimization is specified.
1563 For functions declared inline, this attribute inlines the function even
1564 if no optimization level was specified.
1567 @cindex functions that do pop the argument stack on the 386
1569 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1570 assume that the calling function will pop off the stack space used to
1571 pass arguments. This is
1572 useful to override the effects of the @option{-mrtd} switch.
1575 @cindex @code{const} function attribute
1576 Many functions do not examine any values except their arguments, and
1577 have no effects except the return value. Basically this is just slightly
1578 more strict class than the @code{pure} attribute below, since function is not
1579 allowed to read global memory.
1581 @cindex pointer arguments
1582 Note that a function that has pointer arguments and examines the data
1583 pointed to must @emph{not} be declared @code{const}. Likewise, a
1584 function that calls a non-@code{const} function usually must not be
1585 @code{const}. It does not make sense for a @code{const} function to
1588 The attribute @code{const} is not implemented in GCC versions earlier
1589 than 2.5. An alternative way to declare that a function has no side
1590 effects, which works in the current version and in some older versions,
1594 typedef int intfn ();
1596 extern const intfn square;
1599 This approach does not work in GNU C++ from 2.6.0 on, since the language
1600 specifies that the @samp{const} must be attached to the return value.
1604 @cindex @code{constructor} function attribute
1605 @cindex @code{destructor} function attribute
1606 The @code{constructor} attribute causes the function to be called
1607 automatically before execution enters @code{main ()}. Similarly, the
1608 @code{destructor} attribute causes the function to be called
1609 automatically after @code{main ()} has completed or @code{exit ()} has
1610 been called. Functions with these attributes are useful for
1611 initializing data that will be used implicitly during the execution of
1614 These attributes are not currently implemented for Objective-C@.
1617 @cindex @code{deprecated} attribute.
1618 The @code{deprecated} attribute results in a warning if the function
1619 is used anywhere in the source file. This is useful when identifying
1620 functions that are expected to be removed in a future version of a
1621 program. The warning also includes the location of the declaration
1622 of the deprecated function, to enable users to easily find further
1623 information about why the function is deprecated, or what they should
1624 do instead. Note that the warnings only occurs for uses:
1627 int old_fn () __attribute__ ((deprecated));
1629 int (*fn_ptr)() = old_fn;
1632 results in a warning on line 3 but not line 2.
1634 The @code{deprecated} attribute can also be used for variables and
1635 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
1638 @cindex @code{__declspec(dllexport)}
1639 On Microsoft Windows targets and Symbian OS targets the
1640 @code{dllexport} attribute causes the compiler to provide a global
1641 pointer to a pointer in a DLL, so that it can be referenced with the
1642 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
1643 name is formed by combining @code{_imp__} and the function or variable
1646 You can use @code{__declspec(dllexport)} as a synonym for
1647 @code{__attribute__ ((dllexport))} for compatibility with other
1650 On systems that support the @code{visibility} attribute, this
1651 attribute also implies ``default'' visibility, unless a
1652 @code{visibility} attribute is explicitly specified. You should avoid
1653 the use of @code{dllexport} with ``hidden'' or ``internal''
1654 visibility; in the future GCC may issue an error for those cases.
1656 Currently, the @code{dllexport} attribute is ignored for inlined
1657 functions, unless the @option{-fkeep-inline-functions} flag has been
1658 used. The attribute is also ignored for undefined symbols.
1660 When applied to C++ classes, the attribute marks defined non-inlined
1661 member functions and static data members as exports. Static consts
1662 initialized in-class are not marked unless they are also defined
1665 For Microsoft Windows targets there are alternative methods for
1666 including the symbol in the DLL's export table such as using a
1667 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
1668 the @option{--export-all} linker flag.
1671 @cindex @code{__declspec(dllimport)}
1672 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
1673 attribute causes the compiler to reference a function or variable via
1674 a global pointer to a pointer that is set up by the DLL exporting the
1675 symbol. The attribute implies @code{extern} storage. On Microsoft
1676 Windows targets, the pointer name is formed by combining @code{_imp__}
1677 and the function or variable name.
1679 You can use @code{__declspec(dllimport)} as a synonym for
1680 @code{__attribute__ ((dllimport))} for compatibility with other
1683 Currently, the attribute is ignored for inlined functions. If the
1684 attribute is applied to a symbol @emph{definition}, an error is reported.
1685 If a symbol previously declared @code{dllimport} is later defined, the
1686 attribute is ignored in subsequent references, and a warning is emitted.
1687 The attribute is also overridden by a subsequent declaration as
1690 When applied to C++ classes, the attribute marks non-inlined
1691 member functions and static data members as imports. However, the
1692 attribute is ignored for virtual methods to allow creation of vtables
1695 On the SH Symbian OS target the @code{dllimport} attribute also has
1696 another affect---it can cause the vtable and run-time type information
1697 for a class to be exported. This happens when the class has a
1698 dllimport'ed constructor or a non-inline, non-pure virtual function
1699 and, for either of those two conditions, the class also has a inline
1700 constructor or destructor and has a key function that is defined in
1701 the current translation unit.
1703 For Microsoft Windows based targets the use of the @code{dllimport}
1704 attribute on functions is not necessary, but provides a small
1705 performance benefit by eliminating a thunk in the DLL@. The use of the
1706 @code{dllimport} attribute on imported variables was required on older
1707 versions of the GNU linker, but can now be avoided by passing the
1708 @option{--enable-auto-import} switch to the GNU linker. As with
1709 functions, using the attribute for a variable eliminates a thunk in
1712 One drawback to using this attribute is that a pointer to a function
1713 or variable marked as @code{dllimport} cannot be used as a constant
1714 address. On Microsoft Windows targets, the attribute can be disabled
1715 for functions by setting the @option{-mnop-fun-dllimport} flag.
1718 @cindex eight bit data on the H8/300, H8/300H, and H8S
1719 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1720 variable should be placed into the eight bit data section.
1721 The compiler will generate more efficient code for certain operations
1722 on data in the eight bit data area. Note the eight bit data area is limited to
1725 You must use GAS and GLD from GNU binutils version 2.7 or later for
1726 this attribute to work correctly.
1728 @item exception_handler
1729 @cindex exception handler functions on the Blackfin processor
1730 Use this attribute on the Blackfin to indicate that the specified function
1731 is an exception handler. The compiler will generate function entry and
1732 exit sequences suitable for use in an exception handler when this
1733 attribute is present.
1736 @cindex functions which handle memory bank switching
1737 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
1738 use a calling convention that takes care of switching memory banks when
1739 entering and leaving a function. This calling convention is also the
1740 default when using the @option{-mlong-calls} option.
1742 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
1743 to call and return from a function.
1745 On 68HC11 the compiler will generate a sequence of instructions
1746 to invoke a board-specific routine to switch the memory bank and call the
1747 real function. The board-specific routine simulates a @code{call}.
1748 At the end of a function, it will jump to a board-specific routine
1749 instead of using @code{rts}. The board-specific return routine simulates
1753 @cindex functions that pop the argument stack on the 386
1754 On the Intel 386, the @code{fastcall} attribute causes the compiler to
1755 pass the first two arguments in the registers ECX and EDX@. Subsequent
1756 arguments are passed on the stack. The called function will pop the
1757 arguments off the stack. If the number of arguments is variable all
1758 arguments are pushed on the stack.
1760 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
1761 @cindex @code{format} function attribute
1763 The @code{format} attribute specifies that a function takes @code{printf},
1764 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
1765 should be type-checked against a format string. For example, the
1770 my_printf (void *my_object, const char *my_format, ...)
1771 __attribute__ ((format (printf, 2, 3)));
1775 causes the compiler to check the arguments in calls to @code{my_printf}
1776 for consistency with the @code{printf} style format string argument
1779 The parameter @var{archetype} determines how the format string is
1780 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
1781 or @code{strfmon}. (You can also use @code{__printf__},
1782 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
1783 parameter @var{string-index} specifies which argument is the format
1784 string argument (starting from 1), while @var{first-to-check} is the
1785 number of the first argument to check against the format string. For
1786 functions where the arguments are not available to be checked (such as
1787 @code{vprintf}), specify the third parameter as zero. In this case the
1788 compiler only checks the format string for consistency. For
1789 @code{strftime} formats, the third parameter is required to be zero.
1790 Since non-static C++ methods have an implicit @code{this} argument, the
1791 arguments of such methods should be counted from two, not one, when
1792 giving values for @var{string-index} and @var{first-to-check}.
1794 In the example above, the format string (@code{my_format}) is the second
1795 argument of the function @code{my_print}, and the arguments to check
1796 start with the third argument, so the correct parameters for the format
1797 attribute are 2 and 3.
1799 @opindex ffreestanding
1800 @opindex fno-builtin
1801 The @code{format} attribute allows you to identify your own functions
1802 which take format strings as arguments, so that GCC can check the
1803 calls to these functions for errors. The compiler always (unless
1804 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
1805 for the standard library functions @code{printf}, @code{fprintf},
1806 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
1807 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
1808 warnings are requested (using @option{-Wformat}), so there is no need to
1809 modify the header file @file{stdio.h}. In C99 mode, the functions
1810 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
1811 @code{vsscanf} are also checked. Except in strictly conforming C
1812 standard modes, the X/Open function @code{strfmon} is also checked as
1813 are @code{printf_unlocked} and @code{fprintf_unlocked}.
1814 @xref{C Dialect Options,,Options Controlling C Dialect}.
1816 The target may provide additional types of format checks.
1817 @xref{Target Format Checks,,Format Checks Specific to Particular
1820 @item format_arg (@var{string-index})
1821 @cindex @code{format_arg} function attribute
1822 @opindex Wformat-nonliteral
1823 The @code{format_arg} attribute specifies that a function takes a format
1824 string for a @code{printf}, @code{scanf}, @code{strftime} or
1825 @code{strfmon} style function and modifies it (for example, to translate
1826 it into another language), so the result can be passed to a
1827 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
1828 function (with the remaining arguments to the format function the same
1829 as they would have been for the unmodified string). For example, the
1834 my_dgettext (char *my_domain, const char *my_format)
1835 __attribute__ ((format_arg (2)));
1839 causes the compiler to check the arguments in calls to a @code{printf},
1840 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
1841 format string argument is a call to the @code{my_dgettext} function, for
1842 consistency with the format string argument @code{my_format}. If the
1843 @code{format_arg} attribute had not been specified, all the compiler
1844 could tell in such calls to format functions would be that the format
1845 string argument is not constant; this would generate a warning when
1846 @option{-Wformat-nonliteral} is used, but the calls could not be checked
1847 without the attribute.
1849 The parameter @var{string-index} specifies which argument is the format
1850 string argument (starting from one). Since non-static C++ methods have
1851 an implicit @code{this} argument, the arguments of such methods should
1852 be counted from two.
1854 The @code{format-arg} attribute allows you to identify your own
1855 functions which modify format strings, so that GCC can check the
1856 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
1857 type function whose operands are a call to one of your own function.
1858 The compiler always treats @code{gettext}, @code{dgettext}, and
1859 @code{dcgettext} in this manner except when strict ISO C support is
1860 requested by @option{-ansi} or an appropriate @option{-std} option, or
1861 @option{-ffreestanding} or @option{-fno-builtin}
1862 is used. @xref{C Dialect Options,,Options
1863 Controlling C Dialect}.
1865 @item function_vector
1866 @cindex calling functions through the function vector on the H8/300 processors
1867 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1868 function should be called through the function vector. Calling a
1869 function through the function vector will reduce code size, however;
1870 the function vector has a limited size (maximum 128 entries on the H8/300
1871 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
1873 You must use GAS and GLD from GNU binutils version 2.7 or later for
1874 this attribute to work correctly.
1877 @cindex interrupt handler functions
1878 Use this attribute on the ARM, AVR, C4x, M32R/D and Xstormy16 ports to indicate
1879 that the specified function is an interrupt handler. The compiler will
1880 generate function entry and exit sequences suitable for use in an
1881 interrupt handler when this attribute is present.
1883 Note, interrupt handlers for the Blackfin, m68k, H8/300, H8/300H, H8S, and
1884 SH processors can be specified via the @code{interrupt_handler} attribute.
1886 Note, on the AVR, interrupts will be enabled inside the function.
1888 Note, for the ARM, you can specify the kind of interrupt to be handled by
1889 adding an optional parameter to the interrupt attribute like this:
1892 void f () __attribute__ ((interrupt ("IRQ")));
1895 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
1897 @item interrupt_handler
1898 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
1899 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
1900 indicate that the specified function is an interrupt handler. The compiler
1901 will generate function entry and exit sequences suitable for use in an
1902 interrupt handler when this attribute is present.
1905 @cindex User stack pointer in interrupts on the Blackfin
1906 When used together with @code{interrupt_handler}, @code{exception_handler}
1907 or @code{nmi_handler}, code will be generated to load the stack pointer
1908 from the USP register in the function prologue.
1910 @item long_call/short_call
1911 @cindex indirect calls on ARM
1912 This attribute specifies how a particular function is called on
1913 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
1914 command line switch and @code{#pragma long_calls} settings. The
1915 @code{long_call} attribute causes the compiler to always call the
1916 function by first loading its address into a register and then using the
1917 contents of that register. The @code{short_call} attribute always places
1918 the offset to the function from the call site into the @samp{BL}
1919 instruction directly.
1921 @item longcall/shortcall
1922 @cindex functions called via pointer on the RS/6000 and PowerPC
1923 On the RS/6000 and PowerPC, the @code{longcall} attribute causes the
1924 compiler to always call this function via a pointer, just as it would if
1925 the @option{-mlongcall} option had been specified. The @code{shortcall}
1926 attribute causes the compiler not to do this. These attributes override
1927 both the @option{-mlongcall} switch and the @code{#pragma longcall}
1930 @xref{RS/6000 and PowerPC Options}, for more information on whether long
1931 calls are necessary.
1934 @cindex @code{malloc} attribute
1935 The @code{malloc} attribute is used to tell the compiler that a function
1936 may be treated as if any non-@code{NULL} pointer it returns cannot
1937 alias any other pointer valid when the function returns.
1938 This will often improve optimization.
1939 Standard functions with this property include @code{malloc} and
1940 @code{calloc}. @code{realloc}-like functions have this property as
1941 long as the old pointer is never referred to (including comparing it
1942 to the new pointer) after the function returns a non-@code{NULL}
1945 @item model (@var{model-name})
1946 @cindex function addressability on the M32R/D
1947 @cindex variable addressability on the IA-64
1949 On the M32R/D, use this attribute to set the addressability of an
1950 object, and of the code generated for a function. The identifier
1951 @var{model-name} is one of @code{small}, @code{medium}, or
1952 @code{large}, representing each of the code models.
1954 Small model objects live in the lower 16MB of memory (so that their
1955 addresses can be loaded with the @code{ld24} instruction), and are
1956 callable with the @code{bl} instruction.
1958 Medium model objects may live anywhere in the 32-bit address space (the
1959 compiler will generate @code{seth/add3} instructions to load their addresses),
1960 and are callable with the @code{bl} instruction.
1962 Large model objects may live anywhere in the 32-bit address space (the
1963 compiler will generate @code{seth/add3} instructions to load their addresses),
1964 and may not be reachable with the @code{bl} instruction (the compiler will
1965 generate the much slower @code{seth/add3/jl} instruction sequence).
1967 On IA-64, use this attribute to set the addressability of an object.
1968 At present, the only supported identifier for @var{model-name} is
1969 @code{small}, indicating addressability via ``small'' (22-bit)
1970 addresses (so that their addresses can be loaded with the @code{addl}
1971 instruction). Caveat: such addressing is by definition not position
1972 independent and hence this attribute must not be used for objects
1973 defined by shared libraries.
1976 @cindex function without a prologue/epilogue code
1977 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
1978 specified function does not need prologue/epilogue sequences generated by
1979 the compiler. It is up to the programmer to provide these sequences.
1982 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
1983 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
1984 use the normal calling convention based on @code{jsr} and @code{rts}.
1985 This attribute can be used to cancel the effect of the @option{-mlong-calls}
1989 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
1990 Use this attribute together with @code{interrupt_handler},
1991 @code{exception_handler} or @code{nmi_handler} to indicate that the function
1992 entry code should enable nested interrupts or exceptions.
1995 @cindex NMI handler functions on the Blackfin processor
1996 Use this attribute on the Blackfin to indicate that the specified function
1997 is an NMI handler. The compiler will generate function entry and
1998 exit sequences suitable for use in an NMI handler when this
1999 attribute is present.
2001 @item no_instrument_function
2002 @cindex @code{no_instrument_function} function attribute
2003 @opindex finstrument-functions
2004 If @option{-finstrument-functions} is given, profiling function calls will
2005 be generated at entry and exit of most user-compiled functions.
2006 Functions with this attribute will not be so instrumented.
2009 @cindex @code{noinline} function attribute
2010 This function attribute prevents a function from being considered for
2013 @item nonnull (@var{arg-index}, @dots{})
2014 @cindex @code{nonnull} function attribute
2015 The @code{nonnull} attribute specifies that some function parameters should
2016 be non-null pointers. For instance, the declaration:
2020 my_memcpy (void *dest, const void *src, size_t len)
2021 __attribute__((nonnull (1, 2)));
2025 causes the compiler to check that, in calls to @code{my_memcpy},
2026 arguments @var{dest} and @var{src} are non-null. If the compiler
2027 determines that a null pointer is passed in an argument slot marked
2028 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2029 is issued. The compiler may also choose to make optimizations based
2030 on the knowledge that certain function arguments will not be null.
2032 If no argument index list is given to the @code{nonnull} attribute,
2033 all pointer arguments are marked as non-null. To illustrate, the
2034 following declaration is equivalent to the previous example:
2038 my_memcpy (void *dest, const void *src, size_t len)
2039 __attribute__((nonnull));
2043 @cindex @code{noreturn} function attribute
2044 A few standard library functions, such as @code{abort} and @code{exit},
2045 cannot return. GCC knows this automatically. Some programs define
2046 their own functions that never return. You can declare them
2047 @code{noreturn} to tell the compiler this fact. For example,
2051 void fatal () __attribute__ ((noreturn));
2054 fatal (/* @r{@dots{}} */)
2056 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2062 The @code{noreturn} keyword tells the compiler to assume that
2063 @code{fatal} cannot return. It can then optimize without regard to what
2064 would happen if @code{fatal} ever did return. This makes slightly
2065 better code. More importantly, it helps avoid spurious warnings of
2066 uninitialized variables.
2068 The @code{noreturn} keyword does not affect the exceptional path when that
2069 applies: a @code{noreturn}-marked function may still return to the caller
2070 by throwing an exception or calling @code{longjmp}.
2072 Do not assume that registers saved by the calling function are
2073 restored before calling the @code{noreturn} function.
2075 It does not make sense for a @code{noreturn} function to have a return
2076 type other than @code{void}.
2078 The attribute @code{noreturn} is not implemented in GCC versions
2079 earlier than 2.5. An alternative way to declare that a function does
2080 not return, which works in the current version and in some older
2081 versions, is as follows:
2084 typedef void voidfn ();
2086 volatile voidfn fatal;
2089 This approach does not work in GNU C++.
2092 @cindex @code{nothrow} function attribute
2093 The @code{nothrow} attribute is used to inform the compiler that a
2094 function cannot throw an exception. For example, most functions in
2095 the standard C library can be guaranteed not to throw an exception
2096 with the notable exceptions of @code{qsort} and @code{bsearch} that
2097 take function pointer arguments. The @code{nothrow} attribute is not
2098 implemented in GCC versions earlier than 3.3.
2101 @cindex @code{pure} function attribute
2102 Many functions have no effects except the return value and their
2103 return value depends only on the parameters and/or global variables.
2104 Such a function can be subject
2105 to common subexpression elimination and loop optimization just as an
2106 arithmetic operator would be. These functions should be declared
2107 with the attribute @code{pure}. For example,
2110 int square (int) __attribute__ ((pure));
2114 says that the hypothetical function @code{square} is safe to call
2115 fewer times than the program says.
2117 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2118 Interesting non-pure functions are functions with infinite loops or those
2119 depending on volatile memory or other system resource, that may change between
2120 two consecutive calls (such as @code{feof} in a multithreading environment).
2122 The attribute @code{pure} is not implemented in GCC versions earlier
2125 @item regparm (@var{number})
2126 @cindex @code{regparm} attribute
2127 @cindex functions that are passed arguments in registers on the 386
2128 On the Intel 386, the @code{regparm} attribute causes the compiler to
2129 pass up to @var{number} integer arguments in registers EAX,
2130 EDX, and ECX instead of on the stack. Functions that take a
2131 variable number of arguments will continue to be passed all of their
2132 arguments on the stack.
2134 Beware that on some ELF systems this attribute is unsuitable for
2135 global functions in shared libraries with lazy binding (which is the
2136 default). Lazy binding will send the first call via resolving code in
2137 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2138 per the standard calling conventions. Solaris 8 is affected by this.
2139 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2140 safe since the loaders there save all registers. (Lazy binding can be
2141 disabled with the linker or the loader if desired, to avoid the
2145 @cindex @code{returns_twice} attribute
2146 The @code{returns_twice} attribute tells the compiler that a function may
2147 return more than one time. The compiler will ensure that all registers
2148 are dead before calling such a function and will emit a warning about
2149 the variables that may be clobbered after the second return from the
2150 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2151 The @code{longjmp}-like counterpart of such function, if any, might need
2152 to be marked with the @code{noreturn} attribute.
2155 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2156 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2157 all registers except the stack pointer should be saved in the prologue
2158 regardless of whether they are used or not.
2160 @item section ("@var{section-name}")
2161 @cindex @code{section} function attribute
2162 Normally, the compiler places the code it generates in the @code{text} section.
2163 Sometimes, however, you need additional sections, or you need certain
2164 particular functions to appear in special sections. The @code{section}
2165 attribute specifies that a function lives in a particular section.
2166 For example, the declaration:
2169 extern void foobar (void) __attribute__ ((section ("bar")));
2173 puts the function @code{foobar} in the @code{bar} section.
2175 Some file formats do not support arbitrary sections so the @code{section}
2176 attribute is not available on all platforms.
2177 If you need to map the entire contents of a module to a particular
2178 section, consider using the facilities of the linker instead.
2181 @cindex @code{sentinel} function attribute
2182 This function attribute ensures that a parameter in a function call is
2183 an explicit @code{NULL}. The attribute is only valid on variadic
2184 functions. By default, the sentinel is located at position zero, the
2185 last parameter of the function call. If an optional integer position
2186 argument P is supplied to the attribute, the sentinel must be located at
2187 position P counting backwards from the end of the argument list.
2190 __attribute__ ((sentinel))
2192 __attribute__ ((sentinel(0)))
2195 The attribute is automatically set with a position of 0 for the built-in
2196 functions @code{execl} and @code{execlp}. The built-in function
2197 @code{execle} has the attribute set with a position of 1.
2199 A valid @code{NULL} in this context is defined as zero with any pointer
2200 type. If your system defines the @code{NULL} macro with an integer type
2201 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2202 with a copy that redefines NULL appropriately.
2204 The warnings for missing or incorrect sentinels are enabled with
2208 See long_call/short_call.
2211 See longcall/shortcall.
2214 @cindex signal handler functions on the AVR processors
2215 Use this attribute on the AVR to indicate that the specified
2216 function is a signal handler. The compiler will generate function
2217 entry and exit sequences suitable for use in a signal handler when this
2218 attribute is present. Interrupts will be disabled inside the function.
2221 Use this attribute on the SH to indicate an @code{interrupt_handler}
2222 function should switch to an alternate stack. It expects a string
2223 argument that names a global variable holding the address of the
2228 void f () __attribute__ ((interrupt_handler,
2229 sp_switch ("alt_stack")));
2233 @cindex functions that pop the argument stack on the 386
2234 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2235 assume that the called function will pop off the stack space used to
2236 pass arguments, unless it takes a variable number of arguments.
2239 @cindex tiny data section on the H8/300H and H8S
2240 Use this attribute on the H8/300H and H8S to indicate that the specified
2241 variable should be placed into the tiny data section.
2242 The compiler will generate more efficient code for loads and stores
2243 on data in the tiny data section. Note the tiny data area is limited to
2244 slightly under 32kbytes of data.
2247 Use this attribute on the SH for an @code{interrupt_handler} to return using
2248 @code{trapa} instead of @code{rte}. This attribute expects an integer
2249 argument specifying the trap number to be used.
2252 @cindex @code{unused} attribute.
2253 This attribute, attached to a function, means that the function is meant
2254 to be possibly unused. GCC will not produce a warning for this
2258 @cindex @code{used} attribute.
2259 This attribute, attached to a function, means that code must be emitted
2260 for the function even if it appears that the function is not referenced.
2261 This is useful, for example, when the function is referenced only in
2264 @item visibility ("@var{visibility_type}")
2265 @cindex @code{visibility} attribute
2266 The @code{visibility} attribute on ELF targets causes the declaration
2267 to be emitted with default, hidden, protected or internal visibility.
2270 void __attribute__ ((visibility ("protected")))
2271 f () @{ /* @r{Do something.} */; @}
2272 int i __attribute__ ((visibility ("hidden")));
2275 See the ELF gABI for complete details, but the short story is:
2278 @c keep this list of visibilities in alphabetical order.
2281 Default visibility is the normal case for ELF@. This value is
2282 available for the visibility attribute to override other options
2283 that may change the assumed visibility of symbols.
2286 Hidden visibility indicates that the symbol will not be placed into
2287 the dynamic symbol table, so no other @dfn{module} (executable or
2288 shared library) can reference it directly.
2291 Internal visibility is like hidden visibility, but with additional
2292 processor specific semantics. Unless otherwise specified by the psABI,
2293 GCC defines internal visibility to mean that the function is @emph{never}
2294 called from another module. Note that hidden symbols, while they cannot
2295 be referenced directly by other modules, can be referenced indirectly via
2296 function pointers. By indicating that a symbol cannot be called from
2297 outside the module, GCC may for instance omit the load of a PIC register
2298 since it is known that the calling function loaded the correct value.
2301 Protected visibility indicates that the symbol will be placed in the
2302 dynamic symbol table, but that references within the defining module
2303 will bind to the local symbol. That is, the symbol cannot be overridden
2308 Not all ELF targets support this attribute.
2310 @item warn_unused_result
2311 @cindex @code{warn_unused_result} attribute
2312 The @code{warn_unused_result} attribute causes a warning to be emitted
2313 if a caller of the function with this attribute does not use its
2314 return value. This is useful for functions where not checking
2315 the result is either a security problem or always a bug, such as
2319 int fn () __attribute__ ((warn_unused_result));
2322 if (fn () < 0) return -1;
2328 results in warning on line 5.
2331 @cindex @code{weak} attribute
2332 The @code{weak} attribute causes the declaration to be emitted as a weak
2333 symbol rather than a global. This is primarily useful in defining
2334 library functions which can be overridden in user code, though it can
2335 also be used with non-function declarations. Weak symbols are supported
2336 for ELF targets, and also for a.out targets when using the GNU assembler
2341 You can specify multiple attributes in a declaration by separating them
2342 by commas within the double parentheses or by immediately following an
2343 attribute declaration with another attribute declaration.
2345 @cindex @code{#pragma}, reason for not using
2346 @cindex pragma, reason for not using
2347 Some people object to the @code{__attribute__} feature, suggesting that
2348 ISO C's @code{#pragma} should be used instead. At the time
2349 @code{__attribute__} was designed, there were two reasons for not doing
2354 It is impossible to generate @code{#pragma} commands from a macro.
2357 There is no telling what the same @code{#pragma} might mean in another
2361 These two reasons applied to almost any application that might have been
2362 proposed for @code{#pragma}. It was basically a mistake to use
2363 @code{#pragma} for @emph{anything}.
2365 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2366 to be generated from macros. In addition, a @code{#pragma GCC}
2367 namespace is now in use for GCC-specific pragmas. However, it has been
2368 found convenient to use @code{__attribute__} to achieve a natural
2369 attachment of attributes to their corresponding declarations, whereas
2370 @code{#pragma GCC} is of use for constructs that do not naturally form
2371 part of the grammar. @xref{Other Directives,,Miscellaneous
2372 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2374 @node Attribute Syntax
2375 @section Attribute Syntax
2376 @cindex attribute syntax
2378 This section describes the syntax with which @code{__attribute__} may be
2379 used, and the constructs to which attribute specifiers bind, for the C
2380 language. Some details may vary for C++ and Objective-C@. Because of
2381 infelicities in the grammar for attributes, some forms described here
2382 may not be successfully parsed in all cases.
2384 There are some problems with the semantics of attributes in C++. For
2385 example, there are no manglings for attributes, although they may affect
2386 code generation, so problems may arise when attributed types are used in
2387 conjunction with templates or overloading. Similarly, @code{typeid}
2388 does not distinguish between types with different attributes. Support
2389 for attributes in C++ may be restricted in future to attributes on
2390 declarations only, but not on nested declarators.
2392 @xref{Function Attributes}, for details of the semantics of attributes
2393 applying to functions. @xref{Variable Attributes}, for details of the
2394 semantics of attributes applying to variables. @xref{Type Attributes},
2395 for details of the semantics of attributes applying to structure, union
2396 and enumerated types.
2398 An @dfn{attribute specifier} is of the form
2399 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2400 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2401 each attribute is one of the following:
2405 Empty. Empty attributes are ignored.
2408 A word (which may be an identifier such as @code{unused}, or a reserved
2409 word such as @code{const}).
2412 A word, followed by, in parentheses, parameters for the attribute.
2413 These parameters take one of the following forms:
2417 An identifier. For example, @code{mode} attributes use this form.
2420 An identifier followed by a comma and a non-empty comma-separated list
2421 of expressions. For example, @code{format} attributes use this form.
2424 A possibly empty comma-separated list of expressions. For example,
2425 @code{format_arg} attributes use this form with the list being a single
2426 integer constant expression, and @code{alias} attributes use this form
2427 with the list being a single string constant.
2431 An @dfn{attribute specifier list} is a sequence of one or more attribute
2432 specifiers, not separated by any other tokens.
2434 In GNU C, an attribute specifier list may appear after the colon following a
2435 label, other than a @code{case} or @code{default} label. The only
2436 attribute it makes sense to use after a label is @code{unused}. This
2437 feature is intended for code generated by programs which contains labels
2438 that may be unused but which is compiled with @option{-Wall}. It would
2439 not normally be appropriate to use in it human-written code, though it
2440 could be useful in cases where the code that jumps to the label is
2441 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2442 such placement of attribute lists, as it is permissible for a
2443 declaration, which could begin with an attribute list, to be labelled in
2444 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2445 does not arise there.
2447 An attribute specifier list may appear as part of a @code{struct},
2448 @code{union} or @code{enum} specifier. It may go either immediately
2449 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2450 the closing brace. It is ignored if the content of the structure, union
2451 or enumerated type is not defined in the specifier in which the
2452 attribute specifier list is used---that is, in usages such as
2453 @code{struct __attribute__((foo)) bar} with no following opening brace.
2454 Where attribute specifiers follow the closing brace, they are considered
2455 to relate to the structure, union or enumerated type defined, not to any
2456 enclosing declaration the type specifier appears in, and the type
2457 defined is not complete until after the attribute specifiers.
2458 @c Otherwise, there would be the following problems: a shift/reduce
2459 @c conflict between attributes binding the struct/union/enum and
2460 @c binding to the list of specifiers/qualifiers; and "aligned"
2461 @c attributes could use sizeof for the structure, but the size could be
2462 @c changed later by "packed" attributes.
2464 Otherwise, an attribute specifier appears as part of a declaration,
2465 counting declarations of unnamed parameters and type names, and relates
2466 to that declaration (which may be nested in another declaration, for
2467 example in the case of a parameter declaration), or to a particular declarator
2468 within a declaration. Where an
2469 attribute specifier is applied to a parameter declared as a function or
2470 an array, it should apply to the function or array rather than the
2471 pointer to which the parameter is implicitly converted, but this is not
2472 yet correctly implemented.
2474 Any list of specifiers and qualifiers at the start of a declaration may
2475 contain attribute specifiers, whether or not such a list may in that
2476 context contain storage class specifiers. (Some attributes, however,
2477 are essentially in the nature of storage class specifiers, and only make
2478 sense where storage class specifiers may be used; for example,
2479 @code{section}.) There is one necessary limitation to this syntax: the
2480 first old-style parameter declaration in a function definition cannot
2481 begin with an attribute specifier, because such an attribute applies to
2482 the function instead by syntax described below (which, however, is not
2483 yet implemented in this case). In some other cases, attribute
2484 specifiers are permitted by this grammar but not yet supported by the
2485 compiler. All attribute specifiers in this place relate to the
2486 declaration as a whole. In the obsolescent usage where a type of
2487 @code{int} is implied by the absence of type specifiers, such a list of
2488 specifiers and qualifiers may be an attribute specifier list with no
2489 other specifiers or qualifiers.
2491 At present, the first parameter in a function prototype must have some
2492 type specifier which is not an attribute specifier; this resolves an
2493 ambiguity in the interpretation of @code{void f(int
2494 (__attribute__((foo)) x))}, but is subject to change. At present, if
2495 the parentheses of a function declarator contain only attributes then
2496 those attributes are ignored, rather than yielding an error or warning
2497 or implying a single parameter of type int, but this is subject to
2500 An attribute specifier list may appear immediately before a declarator
2501 (other than the first) in a comma-separated list of declarators in a
2502 declaration of more than one identifier using a single list of
2503 specifiers and qualifiers. Such attribute specifiers apply
2504 only to the identifier before whose declarator they appear. For
2508 __attribute__((noreturn)) void d0 (void),
2509 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2514 the @code{noreturn} attribute applies to all the functions
2515 declared; the @code{format} attribute only applies to @code{d1}.
2517 An attribute specifier list may appear immediately before the comma,
2518 @code{=} or semicolon terminating the declaration of an identifier other
2519 than a function definition. At present, such attribute specifiers apply
2520 to the declared object or function, but in future they may attach to the
2521 outermost adjacent declarator. In simple cases there is no difference,
2522 but, for example, in
2525 void (****f)(void) __attribute__((noreturn));
2529 at present the @code{noreturn} attribute applies to @code{f}, which
2530 causes a warning since @code{f} is not a function, but in future it may
2531 apply to the function @code{****f}. The precise semantics of what
2532 attributes in such cases will apply to are not yet specified. Where an
2533 assembler name for an object or function is specified (@pxref{Asm
2534 Labels}), at present the attribute must follow the @code{asm}
2535 specification; in future, attributes before the @code{asm} specification
2536 may apply to the adjacent declarator, and those after it to the declared
2539 An attribute specifier list may, in future, be permitted to appear after
2540 the declarator in a function definition (before any old-style parameter
2541 declarations or the function body).
2543 Attribute specifiers may be mixed with type qualifiers appearing inside
2544 the @code{[]} of a parameter array declarator, in the C99 construct by
2545 which such qualifiers are applied to the pointer to which the array is
2546 implicitly converted. Such attribute specifiers apply to the pointer,
2547 not to the array, but at present this is not implemented and they are
2550 An attribute specifier list may appear at the start of a nested
2551 declarator. At present, there are some limitations in this usage: the
2552 attributes correctly apply to the declarator, but for most individual
2553 attributes the semantics this implies are not implemented.
2554 When attribute specifiers follow the @code{*} of a pointer
2555 declarator, they may be mixed with any type qualifiers present.
2556 The following describes the formal semantics of this syntax. It will make the
2557 most sense if you are familiar with the formal specification of
2558 declarators in the ISO C standard.
2560 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2561 D1}, where @code{T} contains declaration specifiers that specify a type
2562 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2563 contains an identifier @var{ident}. The type specified for @var{ident}
2564 for derived declarators whose type does not include an attribute
2565 specifier is as in the ISO C standard.
2567 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2568 and the declaration @code{T D} specifies the type
2569 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2570 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2571 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2573 If @code{D1} has the form @code{*
2574 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2575 declaration @code{T D} specifies the type
2576 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2577 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2578 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2584 void (__attribute__((noreturn)) ****f) (void);
2588 specifies the type ``pointer to pointer to pointer to pointer to
2589 non-returning function returning @code{void}''. As another example,
2592 char *__attribute__((aligned(8))) *f;
2596 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2597 Note again that this does not work with most attributes; for example,
2598 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2599 is not yet supported.
2601 For compatibility with existing code written for compiler versions that
2602 did not implement attributes on nested declarators, some laxity is
2603 allowed in the placing of attributes. If an attribute that only applies
2604 to types is applied to a declaration, it will be treated as applying to
2605 the type of that declaration. If an attribute that only applies to
2606 declarations is applied to the type of a declaration, it will be treated
2607 as applying to that declaration; and, for compatibility with code
2608 placing the attributes immediately before the identifier declared, such
2609 an attribute applied to a function return type will be treated as
2610 applying to the function type, and such an attribute applied to an array
2611 element type will be treated as applying to the array type. If an
2612 attribute that only applies to function types is applied to a
2613 pointer-to-function type, it will be treated as applying to the pointer
2614 target type; if such an attribute is applied to a function return type
2615 that is not a pointer-to-function type, it will be treated as applying
2616 to the function type.
2618 @node Function Prototypes
2619 @section Prototypes and Old-Style Function Definitions
2620 @cindex function prototype declarations
2621 @cindex old-style function definitions
2622 @cindex promotion of formal parameters
2624 GNU C extends ISO C to allow a function prototype to override a later
2625 old-style non-prototype definition. Consider the following example:
2628 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2635 /* @r{Prototype function declaration.} */
2636 int isroot P((uid_t));
2638 /* @r{Old-style function definition.} */
2640 isroot (x) /* @r{??? lossage here ???} */
2647 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2648 not allow this example, because subword arguments in old-style
2649 non-prototype definitions are promoted. Therefore in this example the
2650 function definition's argument is really an @code{int}, which does not
2651 match the prototype argument type of @code{short}.
2653 This restriction of ISO C makes it hard to write code that is portable
2654 to traditional C compilers, because the programmer does not know
2655 whether the @code{uid_t} type is @code{short}, @code{int}, or
2656 @code{long}. Therefore, in cases like these GNU C allows a prototype
2657 to override a later old-style definition. More precisely, in GNU C, a
2658 function prototype argument type overrides the argument type specified
2659 by a later old-style definition if the former type is the same as the
2660 latter type before promotion. Thus in GNU C the above example is
2661 equivalent to the following:
2674 GNU C++ does not support old-style function definitions, so this
2675 extension is irrelevant.
2678 @section C++ Style Comments
2680 @cindex C++ comments
2681 @cindex comments, C++ style
2683 In GNU C, you may use C++ style comments, which start with @samp{//} and
2684 continue until the end of the line. Many other C implementations allow
2685 such comments, and they are included in the 1999 C standard. However,
2686 C++ style comments are not recognized if you specify an @option{-std}
2687 option specifying a version of ISO C before C99, or @option{-ansi}
2688 (equivalent to @option{-std=c89}).
2691 @section Dollar Signs in Identifier Names
2693 @cindex dollar signs in identifier names
2694 @cindex identifier names, dollar signs in
2696 In GNU C, you may normally use dollar signs in identifier names.
2697 This is because many traditional C implementations allow such identifiers.
2698 However, dollar signs in identifiers are not supported on a few target
2699 machines, typically because the target assembler does not allow them.
2701 @node Character Escapes
2702 @section The Character @key{ESC} in Constants
2704 You can use the sequence @samp{\e} in a string or character constant to
2705 stand for the ASCII character @key{ESC}.
2708 @section Inquiring on Alignment of Types or Variables
2710 @cindex type alignment
2711 @cindex variable alignment
2713 The keyword @code{__alignof__} allows you to inquire about how an object
2714 is aligned, or the minimum alignment usually required by a type. Its
2715 syntax is just like @code{sizeof}.
2717 For example, if the target machine requires a @code{double} value to be
2718 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2719 This is true on many RISC machines. On more traditional machine
2720 designs, @code{__alignof__ (double)} is 4 or even 2.
2722 Some machines never actually require alignment; they allow reference to any
2723 data type even at an odd address. For these machines, @code{__alignof__}
2724 reports the @emph{recommended} alignment of a type.
2726 If the operand of @code{__alignof__} is an lvalue rather than a type,
2727 its value is the required alignment for its type, taking into account
2728 any minimum alignment specified with GCC's @code{__attribute__}
2729 extension (@pxref{Variable Attributes}). For example, after this
2733 struct foo @{ int x; char y; @} foo1;
2737 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2738 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2740 It is an error to ask for the alignment of an incomplete type.
2742 @node Variable Attributes
2743 @section Specifying Attributes of Variables
2744 @cindex attribute of variables
2745 @cindex variable attributes
2747 The keyword @code{__attribute__} allows you to specify special
2748 attributes of variables or structure fields. This keyword is followed
2749 by an attribute specification inside double parentheses. Some
2750 attributes are currently defined generically for variables.
2751 Other attributes are defined for variables on particular target
2752 systems. Other attributes are available for functions
2753 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
2754 Other front ends might define more attributes
2755 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2757 You may also specify attributes with @samp{__} preceding and following
2758 each keyword. This allows you to use them in header files without
2759 being concerned about a possible macro of the same name. For example,
2760 you may use @code{__aligned__} instead of @code{aligned}.
2762 @xref{Attribute Syntax}, for details of the exact syntax for using
2766 @cindex @code{aligned} attribute
2767 @item aligned (@var{alignment})
2768 This attribute specifies a minimum alignment for the variable or
2769 structure field, measured in bytes. For example, the declaration:
2772 int x __attribute__ ((aligned (16))) = 0;
2776 causes the compiler to allocate the global variable @code{x} on a
2777 16-byte boundary. On a 68040, this could be used in conjunction with
2778 an @code{asm} expression to access the @code{move16} instruction which
2779 requires 16-byte aligned operands.
2781 You can also specify the alignment of structure fields. For example, to
2782 create a double-word aligned @code{int} pair, you could write:
2785 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2789 This is an alternative to creating a union with a @code{double} member
2790 that forces the union to be double-word aligned.
2792 As in the preceding examples, you can explicitly specify the alignment
2793 (in bytes) that you wish the compiler to use for a given variable or
2794 structure field. Alternatively, you can leave out the alignment factor
2795 and just ask the compiler to align a variable or field to the maximum
2796 useful alignment for the target machine you are compiling for. For
2797 example, you could write:
2800 short array[3] __attribute__ ((aligned));
2803 Whenever you leave out the alignment factor in an @code{aligned} attribute
2804 specification, the compiler automatically sets the alignment for the declared
2805 variable or field to the largest alignment which is ever used for any data
2806 type on the target machine you are compiling for. Doing this can often make
2807 copy operations more efficient, because the compiler can use whatever
2808 instructions copy the biggest chunks of memory when performing copies to
2809 or from the variables or fields that you have aligned this way.
2811 The @code{aligned} attribute can only increase the alignment; but you
2812 can decrease it by specifying @code{packed} as well. See below.
2814 Note that the effectiveness of @code{aligned} attributes may be limited
2815 by inherent limitations in your linker. On many systems, the linker is
2816 only able to arrange for variables to be aligned up to a certain maximum
2817 alignment. (For some linkers, the maximum supported alignment may
2818 be very very small.) If your linker is only able to align variables
2819 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
2820 in an @code{__attribute__} will still only provide you with 8 byte
2821 alignment. See your linker documentation for further information.
2823 @item cleanup (@var{cleanup_function})
2824 @cindex @code{cleanup} attribute
2825 The @code{cleanup} attribute runs a function when the variable goes
2826 out of scope. This attribute can only be applied to auto function
2827 scope variables; it may not be applied to parameters or variables
2828 with static storage duration. The function must take one parameter,
2829 a pointer to a type compatible with the variable. The return value
2830 of the function (if any) is ignored.
2832 If @option{-fexceptions} is enabled, then @var{cleanup_function}
2833 will be run during the stack unwinding that happens during the
2834 processing of the exception. Note that the @code{cleanup} attribute
2835 does not allow the exception to be caught, only to perform an action.
2836 It is undefined what happens if @var{cleanup_function} does not
2841 @cindex @code{common} attribute
2842 @cindex @code{nocommon} attribute
2845 The @code{common} attribute requests GCC to place a variable in
2846 ``common'' storage. The @code{nocommon} attribute requests the
2847 opposite---to allocate space for it directly.
2849 These attributes override the default chosen by the
2850 @option{-fno-common} and @option{-fcommon} flags respectively.
2853 @cindex @code{deprecated} attribute
2854 The @code{deprecated} attribute results in a warning if the variable
2855 is used anywhere in the source file. This is useful when identifying
2856 variables that are expected to be removed in a future version of a
2857 program. The warning also includes the location of the declaration
2858 of the deprecated variable, to enable users to easily find further
2859 information about why the variable is deprecated, or what they should
2860 do instead. Note that the warning only occurs for uses:
2863 extern int old_var __attribute__ ((deprecated));
2865 int new_fn () @{ return old_var; @}
2868 results in a warning on line 3 but not line 2.
2870 The @code{deprecated} attribute can also be used for functions and
2871 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
2873 @item mode (@var{mode})
2874 @cindex @code{mode} attribute
2875 This attribute specifies the data type for the declaration---whichever
2876 type corresponds to the mode @var{mode}. This in effect lets you
2877 request an integer or floating point type according to its width.
2879 You may also specify a mode of @samp{byte} or @samp{__byte__} to
2880 indicate the mode corresponding to a one-byte integer, @samp{word} or
2881 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
2882 or @samp{__pointer__} for the mode used to represent pointers.
2885 @cindex @code{packed} attribute
2886 The @code{packed} attribute specifies that a variable or structure field
2887 should have the smallest possible alignment---one byte for a variable,
2888 and one bit for a field, unless you specify a larger value with the
2889 @code{aligned} attribute.
2891 Here is a structure in which the field @code{x} is packed, so that it
2892 immediately follows @code{a}:
2898 int x[2] __attribute__ ((packed));
2902 @item section ("@var{section-name}")
2903 @cindex @code{section} variable attribute
2904 Normally, the compiler places the objects it generates in sections like
2905 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
2906 or you need certain particular variables to appear in special sections,
2907 for example to map to special hardware. The @code{section}
2908 attribute specifies that a variable (or function) lives in a particular
2909 section. For example, this small program uses several specific section names:
2912 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
2913 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
2914 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
2915 int init_data __attribute__ ((section ("INITDATA"))) = 0;
2919 /* @r{Initialize stack pointer} */
2920 init_sp (stack + sizeof (stack));
2922 /* @r{Initialize initialized data} */
2923 memcpy (&init_data, &data, &edata - &data);
2925 /* @r{Turn on the serial ports} */
2932 Use the @code{section} attribute with an @emph{initialized} definition
2933 of a @emph{global} variable, as shown in the example. GCC issues
2934 a warning and otherwise ignores the @code{section} attribute in
2935 uninitialized variable declarations.
2937 You may only use the @code{section} attribute with a fully initialized
2938 global definition because of the way linkers work. The linker requires
2939 each object be defined once, with the exception that uninitialized
2940 variables tentatively go in the @code{common} (or @code{bss}) section
2941 and can be multiply ``defined''. You can force a variable to be
2942 initialized with the @option{-fno-common} flag or the @code{nocommon}
2945 Some file formats do not support arbitrary sections so the @code{section}
2946 attribute is not available on all platforms.
2947 If you need to map the entire contents of a module to a particular
2948 section, consider using the facilities of the linker instead.
2951 @cindex @code{shared} variable attribute
2952 On Microsoft Windows, in addition to putting variable definitions in a named
2953 section, the section can also be shared among all running copies of an
2954 executable or DLL@. For example, this small program defines shared data
2955 by putting it in a named section @code{shared} and marking the section
2959 int foo __attribute__((section ("shared"), shared)) = 0;
2964 /* @r{Read and write foo. All running
2965 copies see the same value.} */
2971 You may only use the @code{shared} attribute along with @code{section}
2972 attribute with a fully initialized global definition because of the way
2973 linkers work. See @code{section} attribute for more information.
2975 The @code{shared} attribute is only available on Microsoft Windows@.
2977 @item tls_model ("@var{tls_model}")
2978 @cindex @code{tls_model} attribute
2979 The @code{tls_model} attribute sets thread-local storage model
2980 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
2981 overriding @option{-ftls-model=} command line switch on a per-variable
2983 The @var{tls_model} argument should be one of @code{global-dynamic},
2984 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
2986 Not all targets support this attribute.
2988 @item transparent_union
2989 This attribute, attached to a function parameter which is a union, means
2990 that the corresponding argument may have the type of any union member,
2991 but the argument is passed as if its type were that of the first union
2992 member. For more details see @xref{Type Attributes}. You can also use
2993 this attribute on a @code{typedef} for a union data type; then it
2994 applies to all function parameters with that type.
2997 This attribute, attached to a variable, means that the variable is meant
2998 to be possibly unused. GCC will not produce a warning for this
3001 @item vector_size (@var{bytes})
3002 This attribute specifies the vector size for the variable, measured in
3003 bytes. For example, the declaration:
3006 int foo __attribute__ ((vector_size (16)));
3010 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3011 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3012 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3014 This attribute is only applicable to integral and float scalars,
3015 although arrays, pointers, and function return values are allowed in
3016 conjunction with this construct.
3018 Aggregates with this attribute are invalid, even if they are of the same
3019 size as a corresponding scalar. For example, the declaration:
3022 struct S @{ int a; @};
3023 struct S __attribute__ ((vector_size (16))) foo;
3027 is invalid even if the size of the structure is the same as the size of
3031 The @code{selectany} attribute causes an initialized global variable to
3032 have link-once semantics. When multiple definitions of the variable are
3033 encountered by the linker, the first is selected and the remainder are
3034 discarded. Following usage by the Microsoft compiler, the linker is told
3035 @emph{not} to warn about size or content differences of the multiple
3038 Although the primary usage of this attribute is for POD types, the
3039 attribute can also be applied to global C++ objects that are initialized
3040 by a constructor. In this case, the static initialization and destruction
3041 code for the object is emitted in each translation defining the object,
3042 but the calls to the constructor and destructor are protected by a
3043 link-once guard variable.
3045 The @code{selectany} attribute is only available on Microsoft Windows
3046 targets. You can use @code{__declspec (selectany)} as a synonym for
3047 @code{__attribute__ ((selectany))} for compatibility with other
3051 The @code{weak} attribute is described in @xref{Function Attributes}.
3054 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3057 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3061 @subsection M32R/D Variable Attributes
3063 One attribute is currently defined for the M32R/D@.
3066 @item model (@var{model-name})
3067 @cindex variable addressability on the M32R/D
3068 Use this attribute on the M32R/D to set the addressability of an object.
3069 The identifier @var{model-name} is one of @code{small}, @code{medium},
3070 or @code{large}, representing each of the code models.
3072 Small model objects live in the lower 16MB of memory (so that their
3073 addresses can be loaded with the @code{ld24} instruction).
3075 Medium and large model objects may live anywhere in the 32-bit address space
3076 (the compiler will generate @code{seth/add3} instructions to load their
3080 @subsection i386 Variable Attributes
3082 Two attributes are currently defined for i386 configurations:
3083 @code{ms_struct} and @code{gcc_struct}
3088 @cindex @code{ms_struct} attribute
3089 @cindex @code{gcc_struct} attribute
3091 If @code{packed} is used on a structure, or if bit-fields are used
3092 it may be that the Microsoft ABI packs them differently
3093 than GCC would normally pack them. Particularly when moving packed
3094 data between functions compiled with GCC and the native Microsoft compiler
3095 (either via function call or as data in a file), it may be necessary to access
3098 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3099 compilers to match the native Microsoft compiler.
3102 @subsection Xstormy16 Variable Attributes
3104 One attribute is currently defined for xstormy16 configurations:
3109 @cindex @code{below100} attribute
3111 If a variable has the @code{below100} attribute (@code{BELOW100} is
3112 allowed also), GCC will place the variable in the first 0x100 bytes of
3113 memory and use special opcodes to access it. Such variables will be
3114 placed in either the @code{.bss_below100} section or the
3115 @code{.data_below100} section.
3119 @node Type Attributes
3120 @section Specifying Attributes of Types
3121 @cindex attribute of types
3122 @cindex type attributes
3124 The keyword @code{__attribute__} allows you to specify special
3125 attributes of @code{struct} and @code{union} types when you define such
3126 types. This keyword is followed by an attribute specification inside
3127 double parentheses. Six attributes are currently defined for types:
3128 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3129 @code{deprecated} and @code{may_alias}. Other attributes are defined for
3130 functions (@pxref{Function Attributes}) and for variables
3131 (@pxref{Variable Attributes}).
3133 You may also specify any one of these attributes with @samp{__}
3134 preceding and following its keyword. This allows you to use these
3135 attributes in header files without being concerned about a possible
3136 macro of the same name. For example, you may use @code{__aligned__}
3137 instead of @code{aligned}.
3139 You may specify the @code{aligned} and @code{transparent_union}
3140 attributes either in a @code{typedef} declaration or just past the
3141 closing curly brace of a complete enum, struct or union type
3142 @emph{definition} and the @code{packed} attribute only past the closing
3143 brace of a definition.
3145 You may also specify attributes between the enum, struct or union
3146 tag and the name of the type rather than after the closing brace.
3148 @xref{Attribute Syntax}, for details of the exact syntax for using
3152 @cindex @code{aligned} attribute
3153 @item aligned (@var{alignment})
3154 This attribute specifies a minimum alignment (in bytes) for variables
3155 of the specified type. For example, the declarations:
3158 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3159 typedef int more_aligned_int __attribute__ ((aligned (8)));
3163 force the compiler to insure (as far as it can) that each variable whose
3164 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3165 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3166 variables of type @code{struct S} aligned to 8-byte boundaries allows
3167 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3168 store) instructions when copying one variable of type @code{struct S} to
3169 another, thus improving run-time efficiency.
3171 Note that the alignment of any given @code{struct} or @code{union} type
3172 is required by the ISO C standard to be at least a perfect multiple of
3173 the lowest common multiple of the alignments of all of the members of
3174 the @code{struct} or @code{union} in question. This means that you @emph{can}
3175 effectively adjust the alignment of a @code{struct} or @code{union}
3176 type by attaching an @code{aligned} attribute to any one of the members
3177 of such a type, but the notation illustrated in the example above is a
3178 more obvious, intuitive, and readable way to request the compiler to
3179 adjust the alignment of an entire @code{struct} or @code{union} type.
3181 As in the preceding example, you can explicitly specify the alignment
3182 (in bytes) that you wish the compiler to use for a given @code{struct}
3183 or @code{union} type. Alternatively, you can leave out the alignment factor
3184 and just ask the compiler to align a type to the maximum
3185 useful alignment for the target machine you are compiling for. For
3186 example, you could write:
3189 struct S @{ short f[3]; @} __attribute__ ((aligned));
3192 Whenever you leave out the alignment factor in an @code{aligned}
3193 attribute specification, the compiler automatically sets the alignment
3194 for the type to the largest alignment which is ever used for any data
3195 type on the target machine you are compiling for. Doing this can often
3196 make copy operations more efficient, because the compiler can use
3197 whatever instructions copy the biggest chunks of memory when performing
3198 copies to or from the variables which have types that you have aligned
3201 In the example above, if the size of each @code{short} is 2 bytes, then
3202 the size of the entire @code{struct S} type is 6 bytes. The smallest
3203 power of two which is greater than or equal to that is 8, so the
3204 compiler sets the alignment for the entire @code{struct S} type to 8
3207 Note that although you can ask the compiler to select a time-efficient
3208 alignment for a given type and then declare only individual stand-alone
3209 objects of that type, the compiler's ability to select a time-efficient
3210 alignment is primarily useful only when you plan to create arrays of
3211 variables having the relevant (efficiently aligned) type. If you
3212 declare or use arrays of variables of an efficiently-aligned type, then
3213 it is likely that your program will also be doing pointer arithmetic (or
3214 subscripting, which amounts to the same thing) on pointers to the
3215 relevant type, and the code that the compiler generates for these
3216 pointer arithmetic operations will often be more efficient for
3217 efficiently-aligned types than for other types.
3219 The @code{aligned} attribute can only increase the alignment; but you
3220 can decrease it by specifying @code{packed} as well. See below.
3222 Note that the effectiveness of @code{aligned} attributes may be limited
3223 by inherent limitations in your linker. On many systems, the linker is
3224 only able to arrange for variables to be aligned up to a certain maximum
3225 alignment. (For some linkers, the maximum supported alignment may
3226 be very very small.) If your linker is only able to align variables
3227 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3228 in an @code{__attribute__} will still only provide you with 8 byte
3229 alignment. See your linker documentation for further information.
3232 This attribute, attached to @code{struct} or @code{union} type
3233 definition, specifies that each member of the structure or union is
3234 placed to minimize the memory required. When attached to an @code{enum}
3235 definition, it indicates that the smallest integral type should be used.
3237 @opindex fshort-enums
3238 Specifying this attribute for @code{struct} and @code{union} types is
3239 equivalent to specifying the @code{packed} attribute on each of the
3240 structure or union members. Specifying the @option{-fshort-enums}
3241 flag on the line is equivalent to specifying the @code{packed}
3242 attribute on all @code{enum} definitions.
3244 In the following example @code{struct my_packed_struct}'s members are
3245 packed closely together, but the internal layout of its @code{s} member
3246 is not packed---to do that, @code{struct my_unpacked_struct} would need to
3250 struct my_unpacked_struct
3256 struct __attribute__ ((__packed__)) my_packed_struct
3260 struct my_unpacked_struct s;
3264 You may only specify this attribute on the definition of a @code{enum},
3265 @code{struct} or @code{union}, not on a @code{typedef} which does not
3266 also define the enumerated type, structure or union.
3268 @item transparent_union
3269 This attribute, attached to a @code{union} type definition, indicates
3270 that any function parameter having that union type causes calls to that
3271 function to be treated in a special way.
3273 First, the argument corresponding to a transparent union type can be of
3274 any type in the union; no cast is required. Also, if the union contains
3275 a pointer type, the corresponding argument can be a null pointer
3276 constant or a void pointer expression; and if the union contains a void
3277 pointer type, the corresponding argument can be any pointer expression.
3278 If the union member type is a pointer, qualifiers like @code{const} on
3279 the referenced type must be respected, just as with normal pointer
3282 Second, the argument is passed to the function using the calling
3283 conventions of the first member of the transparent union, not the calling
3284 conventions of the union itself. All members of the union must have the
3285 same machine representation; this is necessary for this argument passing
3288 Transparent unions are designed for library functions that have multiple
3289 interfaces for compatibility reasons. For example, suppose the
3290 @code{wait} function must accept either a value of type @code{int *} to
3291 comply with Posix, or a value of type @code{union wait *} to comply with
3292 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3293 @code{wait} would accept both kinds of arguments, but it would also
3294 accept any other pointer type and this would make argument type checking
3295 less useful. Instead, @code{<sys/wait.h>} might define the interface
3303 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3305 pid_t wait (wait_status_ptr_t);
3308 This interface allows either @code{int *} or @code{union wait *}
3309 arguments to be passed, using the @code{int *} calling convention.
3310 The program can call @code{wait} with arguments of either type:
3313 int w1 () @{ int w; return wait (&w); @}
3314 int w2 () @{ union wait w; return wait (&w); @}
3317 With this interface, @code{wait}'s implementation might look like this:
3320 pid_t wait (wait_status_ptr_t p)
3322 return waitpid (-1, p.__ip, 0);
3327 When attached to a type (including a @code{union} or a @code{struct}),
3328 this attribute means that variables of that type are meant to appear
3329 possibly unused. GCC will not produce a warning for any variables of
3330 that type, even if the variable appears to do nothing. This is often
3331 the case with lock or thread classes, which are usually defined and then
3332 not referenced, but contain constructors and destructors that have
3333 nontrivial bookkeeping functions.
3336 The @code{deprecated} attribute results in a warning if the type
3337 is used anywhere in the source file. This is useful when identifying
3338 types that are expected to be removed in a future version of a program.
3339 If possible, the warning also includes the location of the declaration
3340 of the deprecated type, to enable users to easily find further
3341 information about why the type is deprecated, or what they should do
3342 instead. Note that the warnings only occur for uses and then only
3343 if the type is being applied to an identifier that itself is not being
3344 declared as deprecated.
3347 typedef int T1 __attribute__ ((deprecated));
3351 typedef T1 T3 __attribute__ ((deprecated));
3352 T3 z __attribute__ ((deprecated));
3355 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3356 warning is issued for line 4 because T2 is not explicitly
3357 deprecated. Line 5 has no warning because T3 is explicitly
3358 deprecated. Similarly for line 6.
3360 The @code{deprecated} attribute can also be used for functions and
3361 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3364 Accesses to objects with types with this attribute are not subjected to
3365 type-based alias analysis, but are instead assumed to be able to alias
3366 any other type of objects, just like the @code{char} type. See
3367 @option{-fstrict-aliasing} for more information on aliasing issues.
3372 typedef short __attribute__((__may_alias__)) short_a;
3378 short_a *b = (short_a *) &a;
3382 if (a == 0x12345678)
3389 If you replaced @code{short_a} with @code{short} in the variable
3390 declaration, the above program would abort when compiled with
3391 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3392 above in recent GCC versions.
3394 @subsection ARM Type Attributes
3396 On those ARM targets that support @code{dllimport} (such as Symbian
3397 OS), you can use the @code{notshared} attribute to indicate that the
3398 virtual table and other similar data for a class should not be
3399 exported from a DLL@. For example:
3402 class __declspec(notshared) C @{
3404 __declspec(dllimport) C();
3408 __declspec(dllexport)
3412 In this code, @code{C::C} is exported from the current DLL, but the
3413 virtual table for @code{C} is not exported. (You can use
3414 @code{__attribute__} instead of @code{__declspec} if you prefer, but
3415 most Symbian OS code uses @code{__declspec}.)
3417 @subsection i386 Type Attributes
3419 Two attributes are currently defined for i386 configurations:
3420 @code{ms_struct} and @code{gcc_struct}
3424 @cindex @code{ms_struct}
3425 @cindex @code{gcc_struct}
3427 If @code{packed} is used on a structure, or if bit-fields are used
3428 it may be that the Microsoft ABI packs them differently
3429 than GCC would normally pack them. Particularly when moving packed
3430 data between functions compiled with GCC and the native Microsoft compiler
3431 (either via function call or as data in a file), it may be necessary to access
3434 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3435 compilers to match the native Microsoft compiler.
3438 To specify multiple attributes, separate them by commas within the
3439 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3443 @section An Inline Function is As Fast As a Macro
3444 @cindex inline functions
3445 @cindex integrating function code
3447 @cindex macros, inline alternative
3449 By declaring a function @code{inline}, you can direct GCC to
3450 integrate that function's code into the code for its callers. This
3451 makes execution faster by eliminating the function-call overhead; in
3452 addition, if any of the actual argument values are constant, their known
3453 values may permit simplifications at compile time so that not all of the
3454 inline function's code needs to be included. The effect on code size is
3455 less predictable; object code may be larger or smaller with function
3456 inlining, depending on the particular case. Inlining of functions is an
3457 optimization and it really ``works'' only in optimizing compilation. If
3458 you don't use @option{-O}, no function is really inline.
3460 Inline functions are included in the ISO C99 standard, but there are
3461 currently substantial differences between what GCC implements and what
3462 the ISO C99 standard requires.
3464 To declare a function inline, use the @code{inline} keyword in its
3465 declaration, like this:
3475 (If you are writing a header file to be included in ISO C programs, write
3476 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3477 You can also make all ``simple enough'' functions inline with the option
3478 @option{-finline-functions}.
3481 Note that certain usages in a function definition can make it unsuitable
3482 for inline substitution. Among these usages are: use of varargs, use of
3483 alloca, use of variable sized data types (@pxref{Variable Length}),
3484 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3485 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3486 will warn when a function marked @code{inline} could not be substituted,
3487 and will give the reason for the failure.
3489 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3490 does not affect the linkage of the function.
3492 @cindex automatic @code{inline} for C++ member fns
3493 @cindex @code{inline} automatic for C++ member fns
3494 @cindex member fns, automatically @code{inline}
3495 @cindex C++ member fns, automatically @code{inline}
3496 @opindex fno-default-inline
3497 GCC automatically inlines member functions defined within the class
3498 body of C++ programs even if they are not explicitly declared
3499 @code{inline}. (You can override this with @option{-fno-default-inline};
3500 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3502 @cindex inline functions, omission of
3503 @opindex fkeep-inline-functions
3504 When a function is both inline and @code{static}, if all calls to the
3505 function are integrated into the caller, and the function's address is
3506 never used, then the function's own assembler code is never referenced.
3507 In this case, GCC does not actually output assembler code for the
3508 function, unless you specify the option @option{-fkeep-inline-functions}.
3509 Some calls cannot be integrated for various reasons (in particular,
3510 calls that precede the function's definition cannot be integrated, and
3511 neither can recursive calls within the definition). If there is a
3512 nonintegrated call, then the function is compiled to assembler code as
3513 usual. The function must also be compiled as usual if the program
3514 refers to its address, because that can't be inlined.
3516 @cindex non-static inline function
3517 When an inline function is not @code{static}, then the compiler must assume
3518 that there may be calls from other source files; since a global symbol can
3519 be defined only once in any program, the function must not be defined in
3520 the other source files, so the calls therein cannot be integrated.
3521 Therefore, a non-@code{static} inline function is always compiled on its
3522 own in the usual fashion.
3524 If you specify both @code{inline} and @code{extern} in the function
3525 definition, then the definition is used only for inlining. In no case
3526 is the function compiled on its own, not even if you refer to its
3527 address explicitly. Such an address becomes an external reference, as
3528 if you had only declared the function, and had not defined it.
3530 This combination of @code{inline} and @code{extern} has almost the
3531 effect of a macro. The way to use it is to put a function definition in
3532 a header file with these keywords, and put another copy of the
3533 definition (lacking @code{inline} and @code{extern}) in a library file.
3534 The definition in the header file will cause most calls to the function
3535 to be inlined. If any uses of the function remain, they will refer to
3536 the single copy in the library.
3538 Since GCC eventually will implement ISO C99 semantics for
3539 inline functions, it is best to use @code{static inline} only
3540 to guarantee compatibility. (The
3541 existing semantics will remain available when @option{-std=gnu89} is
3542 specified, but eventually the default will be @option{-std=gnu99} and
3543 that will implement the C99 semantics, though it does not do so yet.)
3545 GCC does not inline any functions when not optimizing unless you specify
3546 the @samp{always_inline} attribute for the function, like this:
3549 /* @r{Prototype.} */
3550 inline void foo (const char) __attribute__((always_inline));
3554 @section Assembler Instructions with C Expression Operands
3555 @cindex extended @code{asm}
3556 @cindex @code{asm} expressions
3557 @cindex assembler instructions
3560 In an assembler instruction using @code{asm}, you can specify the
3561 operands of the instruction using C expressions. This means you need not
3562 guess which registers or memory locations will contain the data you want
3565 You must specify an assembler instruction template much like what
3566 appears in a machine description, plus an operand constraint string for
3569 For example, here is how to use the 68881's @code{fsinx} instruction:
3572 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3576 Here @code{angle} is the C expression for the input operand while
3577 @code{result} is that of the output operand. Each has @samp{"f"} as its
3578 operand constraint, saying that a floating point register is required.
3579 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3580 output operands' constraints must use @samp{=}. The constraints use the
3581 same language used in the machine description (@pxref{Constraints}).
3583 Each operand is described by an operand-constraint string followed by
3584 the C expression in parentheses. A colon separates the assembler
3585 template from the first output operand and another separates the last
3586 output operand from the first input, if any. Commas separate the
3587 operands within each group. The total number of operands is currently
3588 limited to 30; this limitation may be lifted in some future version of
3591 If there are no output operands but there are input operands, you must
3592 place two consecutive colons surrounding the place where the output
3595 As of GCC version 3.1, it is also possible to specify input and output
3596 operands using symbolic names which can be referenced within the
3597 assembler code. These names are specified inside square brackets
3598 preceding the constraint string, and can be referenced inside the
3599 assembler code using @code{%[@var{name}]} instead of a percentage sign
3600 followed by the operand number. Using named operands the above example
3604 asm ("fsinx %[angle],%[output]"
3605 : [output] "=f" (result)
3606 : [angle] "f" (angle));
3610 Note that the symbolic operand names have no relation whatsoever to
3611 other C identifiers. You may use any name you like, even those of
3612 existing C symbols, but you must ensure that no two operands within the same
3613 assembler construct use the same symbolic name.
3615 Output operand expressions must be lvalues; the compiler can check this.
3616 The input operands need not be lvalues. The compiler cannot check
3617 whether the operands have data types that are reasonable for the
3618 instruction being executed. It does not parse the assembler instruction
3619 template and does not know what it means or even whether it is valid
3620 assembler input. The extended @code{asm} feature is most often used for
3621 machine instructions the compiler itself does not know exist. If
3622 the output expression cannot be directly addressed (for example, it is a
3623 bit-field), your constraint must allow a register. In that case, GCC
3624 will use the register as the output of the @code{asm}, and then store
3625 that register into the output.
3627 The ordinary output operands must be write-only; GCC will assume that
3628 the values in these operands before the instruction are dead and need
3629 not be generated. Extended asm supports input-output or read-write
3630 operands. Use the constraint character @samp{+} to indicate such an
3631 operand and list it with the output operands. You should only use
3632 read-write operands when the constraints for the operand (or the
3633 operand in which only some of the bits are to be changed) allow a
3636 You may, as an alternative, logically split its function into two
3637 separate operands, one input operand and one write-only output
3638 operand. The connection between them is expressed by constraints
3639 which say they need to be in the same location when the instruction
3640 executes. You can use the same C expression for both operands, or
3641 different expressions. For example, here we write the (fictitious)
3642 @samp{combine} instruction with @code{bar} as its read-only source
3643 operand and @code{foo} as its read-write destination:
3646 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3650 The constraint @samp{"0"} for operand 1 says that it must occupy the
3651 same location as operand 0. A number in constraint is allowed only in
3652 an input operand and it must refer to an output operand.
3654 Only a number in the constraint can guarantee that one operand will be in
3655 the same place as another. The mere fact that @code{foo} is the value
3656 of both operands is not enough to guarantee that they will be in the
3657 same place in the generated assembler code. The following would not
3661 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3664 Various optimizations or reloading could cause operands 0 and 1 to be in
3665 different registers; GCC knows no reason not to do so. For example, the
3666 compiler might find a copy of the value of @code{foo} in one register and
3667 use it for operand 1, but generate the output operand 0 in a different
3668 register (copying it afterward to @code{foo}'s own address). Of course,
3669 since the register for operand 1 is not even mentioned in the assembler
3670 code, the result will not work, but GCC can't tell that.
3672 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3673 the operand number for a matching constraint. For example:
3676 asm ("cmoveq %1,%2,%[result]"
3677 : [result] "=r"(result)
3678 : "r" (test), "r"(new), "[result]"(old));
3681 Sometimes you need to make an @code{asm} operand be a specific register,
3682 but there's no matching constraint letter for that register @emph{by
3683 itself}. To force the operand into that register, use a local variable
3684 for the operand and specify the register in the variable declaration.
3685 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
3686 register constraint letter that matches the register:
3689 register int *p1 asm ("r0") = @dots{};
3690 register int *p2 asm ("r1") = @dots{};
3691 register int *result asm ("r0");
3692 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3695 @anchor{Example of asm with clobbered asm reg}
3696 In the above example, beware that a register that is call-clobbered by
3697 the target ABI will be overwritten by any function call in the
3698 assignment, including library calls for arithmetic operators.
3699 Assuming it is a call-clobbered register, this may happen to @code{r0}
3700 above by the assignment to @code{p2}. If you have to use such a
3701 register, use temporary variables for expressions between the register
3706 register int *p1 asm ("r0") = @dots{};
3707 register int *p2 asm ("r1") = t1;
3708 register int *result asm ("r0");
3709 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3712 Some instructions clobber specific hard registers. To describe this,
3713 write a third colon after the input operands, followed by the names of
3714 the clobbered hard registers (given as strings). Here is a realistic
3715 example for the VAX:
3718 asm volatile ("movc3 %0,%1,%2"
3719 : /* @r{no outputs} */
3720 : "g" (from), "g" (to), "g" (count)
3721 : "r0", "r1", "r2", "r3", "r4", "r5");
3724 You may not write a clobber description in a way that overlaps with an
3725 input or output operand. For example, you may not have an operand
3726 describing a register class with one member if you mention that register
3727 in the clobber list. Variables declared to live in specific registers
3728 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
3729 have no part mentioned in the clobber description.
3730 There is no way for you to specify that an input
3731 operand is modified without also specifying it as an output
3732 operand. Note that if all the output operands you specify are for this
3733 purpose (and hence unused), you will then also need to specify
3734 @code{volatile} for the @code{asm} construct, as described below, to
3735 prevent GCC from deleting the @code{asm} statement as unused.
3737 If you refer to a particular hardware register from the assembler code,
3738 you will probably have to list the register after the third colon to
3739 tell the compiler the register's value is modified. In some assemblers,
3740 the register names begin with @samp{%}; to produce one @samp{%} in the
3741 assembler code, you must write @samp{%%} in the input.
3743 If your assembler instruction can alter the condition code register, add
3744 @samp{cc} to the list of clobbered registers. GCC on some machines
3745 represents the condition codes as a specific hardware register;
3746 @samp{cc} serves to name this register. On other machines, the
3747 condition code is handled differently, and specifying @samp{cc} has no
3748 effect. But it is valid no matter what the machine.
3750 If your assembler instructions access memory in an unpredictable
3751 fashion, add @samp{memory} to the list of clobbered registers. This
3752 will cause GCC to not keep memory values cached in registers across the
3753 assembler instruction and not optimize stores or loads to that memory.
3754 You will also want to add the @code{volatile} keyword if the memory
3755 affected is not listed in the inputs or outputs of the @code{asm}, as
3756 the @samp{memory} clobber does not count as a side-effect of the
3757 @code{asm}. If you know how large the accessed memory is, you can add
3758 it as input or output but if this is not known, you should add
3759 @samp{memory}. As an example, if you access ten bytes of a string, you
3760 can use a memory input like:
3763 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
3766 Note that in the following example the memory input is necessary,
3767 otherwise GCC might optimize the store to @code{x} away:
3774 asm ("magic stuff accessing an 'int' pointed to by '%1'"
3775 "=&d" (r) : "a" (y), "m" (*y));
3780 You can put multiple assembler instructions together in a single
3781 @code{asm} template, separated by the characters normally used in assembly
3782 code for the system. A combination that works in most places is a newline
3783 to break the line, plus a tab character to move to the instruction field
3784 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
3785 assembler allows semicolons as a line-breaking character. Note that some
3786 assembler dialects use semicolons to start a comment.
3787 The input operands are guaranteed not to use any of the clobbered
3788 registers, and neither will the output operands' addresses, so you can
3789 read and write the clobbered registers as many times as you like. Here
3790 is an example of multiple instructions in a template; it assumes the
3791 subroutine @code{_foo} accepts arguments in registers 9 and 10:
3794 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
3796 : "g" (from), "g" (to)
3800 Unless an output operand has the @samp{&} constraint modifier, GCC
3801 may allocate it in the same register as an unrelated input operand, on
3802 the assumption the inputs are consumed before the outputs are produced.
3803 This assumption may be false if the assembler code actually consists of
3804 more than one instruction. In such a case, use @samp{&} for each output
3805 operand that may not overlap an input. @xref{Modifiers}.
3807 If you want to test the condition code produced by an assembler
3808 instruction, you must include a branch and a label in the @code{asm}
3809 construct, as follows:
3812 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
3818 This assumes your assembler supports local labels, as the GNU assembler
3819 and most Unix assemblers do.
3821 Speaking of labels, jumps from one @code{asm} to another are not
3822 supported. The compiler's optimizers do not know about these jumps, and
3823 therefore they cannot take account of them when deciding how to
3826 @cindex macros containing @code{asm}
3827 Usually the most convenient way to use these @code{asm} instructions is to
3828 encapsulate them in macros that look like functions. For example,
3832 (@{ double __value, __arg = (x); \
3833 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
3838 Here the variable @code{__arg} is used to make sure that the instruction
3839 operates on a proper @code{double} value, and to accept only those
3840 arguments @code{x} which can convert automatically to a @code{double}.
3842 Another way to make sure the instruction operates on the correct data
3843 type is to use a cast in the @code{asm}. This is different from using a
3844 variable @code{__arg} in that it converts more different types. For
3845 example, if the desired type were @code{int}, casting the argument to
3846 @code{int} would accept a pointer with no complaint, while assigning the
3847 argument to an @code{int} variable named @code{__arg} would warn about
3848 using a pointer unless the caller explicitly casts it.
3850 If an @code{asm} has output operands, GCC assumes for optimization
3851 purposes the instruction has no side effects except to change the output
3852 operands. This does not mean instructions with a side effect cannot be
3853 used, but you must be careful, because the compiler may eliminate them
3854 if the output operands aren't used, or move them out of loops, or
3855 replace two with one if they constitute a common subexpression. Also,
3856 if your instruction does have a side effect on a variable that otherwise
3857 appears not to change, the old value of the variable may be reused later
3858 if it happens to be found in a register.
3860 You can prevent an @code{asm} instruction from being deleted
3861 by writing the keyword @code{volatile} after
3862 the @code{asm}. For example:
3865 #define get_and_set_priority(new) \
3867 asm volatile ("get_and_set_priority %0, %1" \
3868 : "=g" (__old) : "g" (new)); \
3873 The @code{volatile} keyword indicates that the instruction has
3874 important side-effects. GCC will not delete a volatile @code{asm} if
3875 it is reachable. (The instruction can still be deleted if GCC can
3876 prove that control-flow will never reach the location of the
3877 instruction.) Note that even a volatile @code{asm} instruction
3878 can be moved relative to other code, including across jump
3879 instructions. For example, on many targets there is a system
3880 register which can be set to control the rounding mode of
3881 floating point operations. You might try
3882 setting it with a volatile @code{asm}, like this PowerPC example:
3885 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
3890 This will not work reliably, as the compiler may move the addition back
3891 before the volatile @code{asm}. To make it work you need to add an
3892 artificial dependency to the @code{asm} referencing a variable in the code
3893 you don't want moved, for example:
3896 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
3900 Similarly, you can't expect a
3901 sequence of volatile @code{asm} instructions to remain perfectly
3902 consecutive. If you want consecutive output, use a single @code{asm}.
3903 Also, GCC will perform some optimizations across a volatile @code{asm}
3904 instruction; GCC does not ``forget everything'' when it encounters
3905 a volatile @code{asm} instruction the way some other compilers do.
3907 An @code{asm} instruction without any output operands will be treated
3908 identically to a volatile @code{asm} instruction.
3910 It is a natural idea to look for a way to give access to the condition
3911 code left by the assembler instruction. However, when we attempted to
3912 implement this, we found no way to make it work reliably. The problem
3913 is that output operands might need reloading, which would result in
3914 additional following ``store'' instructions. On most machines, these
3915 instructions would alter the condition code before there was time to
3916 test it. This problem doesn't arise for ordinary ``test'' and
3917 ``compare'' instructions because they don't have any output operands.
3919 For reasons similar to those described above, it is not possible to give
3920 an assembler instruction access to the condition code left by previous
3923 If you are writing a header file that should be includable in ISO C
3924 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
3927 @subsection Size of an @code{asm}
3929 Some targets require that GCC track the size of each instruction used in
3930 order to generate correct code. Because the final length of an
3931 @code{asm} is only known by the assembler, GCC must make an estimate as
3932 to how big it will be. The estimate is formed by counting the number of
3933 statements in the pattern of the @code{asm} and multiplying that by the
3934 length of the longest instruction on that processor. Statements in the
3935 @code{asm} are identified by newline characters and whatever statement
3936 separator characters are supported by the assembler; on most processors
3937 this is the `@code{;}' character.
3939 Normally, GCC's estimate is perfectly adequate to ensure that correct
3940 code is generated, but it is possible to confuse the compiler if you use
3941 pseudo instructions or assembler macros that expand into multiple real
3942 instructions or if you use assembler directives that expand to more
3943 space in the object file than would be needed for a single instruction.
3944 If this happens then the assembler will produce a diagnostic saying that
3945 a label is unreachable.
3947 @subsection i386 floating point asm operands
3949 There are several rules on the usage of stack-like regs in
3950 asm_operands insns. These rules apply only to the operands that are
3955 Given a set of input regs that die in an asm_operands, it is
3956 necessary to know which are implicitly popped by the asm, and
3957 which must be explicitly popped by gcc.
3959 An input reg that is implicitly popped by the asm must be
3960 explicitly clobbered, unless it is constrained to match an
3964 For any input reg that is implicitly popped by an asm, it is
3965 necessary to know how to adjust the stack to compensate for the pop.
3966 If any non-popped input is closer to the top of the reg-stack than
3967 the implicitly popped reg, it would not be possible to know what the
3968 stack looked like---it's not clear how the rest of the stack ``slides
3971 All implicitly popped input regs must be closer to the top of
3972 the reg-stack than any input that is not implicitly popped.
3974 It is possible that if an input dies in an insn, reload might
3975 use the input reg for an output reload. Consider this example:
3978 asm ("foo" : "=t" (a) : "f" (b));
3981 This asm says that input B is not popped by the asm, and that
3982 the asm pushes a result onto the reg-stack, i.e., the stack is one
3983 deeper after the asm than it was before. But, it is possible that
3984 reload will think that it can use the same reg for both the input and
3985 the output, if input B dies in this insn.
3987 If any input operand uses the @code{f} constraint, all output reg
3988 constraints must use the @code{&} earlyclobber.
3990 The asm above would be written as
3993 asm ("foo" : "=&t" (a) : "f" (b));
3997 Some operands need to be in particular places on the stack. All
3998 output operands fall in this category---there is no other way to
3999 know which regs the outputs appear in unless the user indicates
4000 this in the constraints.
4002 Output operands must specifically indicate which reg an output
4003 appears in after an asm. @code{=f} is not allowed: the operand
4004 constraints must select a class with a single reg.
4007 Output operands may not be ``inserted'' between existing stack regs.
4008 Since no 387 opcode uses a read/write operand, all output operands
4009 are dead before the asm_operands, and are pushed by the asm_operands.
4010 It makes no sense to push anywhere but the top of the reg-stack.
4012 Output operands must start at the top of the reg-stack: output
4013 operands may not ``skip'' a reg.
4016 Some asm statements may need extra stack space for internal
4017 calculations. This can be guaranteed by clobbering stack registers
4018 unrelated to the inputs and outputs.
4022 Here are a couple of reasonable asms to want to write. This asm
4023 takes one input, which is internally popped, and produces two outputs.
4026 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4029 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4030 and replaces them with one output. The user must code the @code{st(1)}
4031 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4034 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4040 @section Controlling Names Used in Assembler Code
4041 @cindex assembler names for identifiers
4042 @cindex names used in assembler code
4043 @cindex identifiers, names in assembler code
4045 You can specify the name to be used in the assembler code for a C
4046 function or variable by writing the @code{asm} (or @code{__asm__})
4047 keyword after the declarator as follows:
4050 int foo asm ("myfoo") = 2;
4054 This specifies that the name to be used for the variable @code{foo} in
4055 the assembler code should be @samp{myfoo} rather than the usual
4058 On systems where an underscore is normally prepended to the name of a C
4059 function or variable, this feature allows you to define names for the
4060 linker that do not start with an underscore.
4062 It does not make sense to use this feature with a non-static local
4063 variable since such variables do not have assembler names. If you are
4064 trying to put the variable in a particular register, see @ref{Explicit
4065 Reg Vars}. GCC presently accepts such code with a warning, but will
4066 probably be changed to issue an error, rather than a warning, in the
4069 You cannot use @code{asm} in this way in a function @emph{definition}; but
4070 you can get the same effect by writing a declaration for the function
4071 before its definition and putting @code{asm} there, like this:
4074 extern func () asm ("FUNC");
4081 It is up to you to make sure that the assembler names you choose do not
4082 conflict with any other assembler symbols. Also, you must not use a
4083 register name; that would produce completely invalid assembler code. GCC
4084 does not as yet have the ability to store static variables in registers.
4085 Perhaps that will be added.
4087 @node Explicit Reg Vars
4088 @section Variables in Specified Registers
4089 @cindex explicit register variables
4090 @cindex variables in specified registers
4091 @cindex specified registers
4092 @cindex registers, global allocation
4094 GNU C allows you to put a few global variables into specified hardware
4095 registers. You can also specify the register in which an ordinary
4096 register variable should be allocated.
4100 Global register variables reserve registers throughout the program.
4101 This may be useful in programs such as programming language
4102 interpreters which have a couple of global variables that are accessed
4106 Local register variables in specific registers do not reserve the
4107 registers, except at the point where they are used as input or output
4108 operands in an @code{asm} statement and the @code{asm} statement itself is
4109 not deleted. The compiler's data flow analysis is capable of determining
4110 where the specified registers contain live values, and where they are
4111 available for other uses. Stores into local register variables may be deleted
4112 when they appear to be dead according to dataflow analysis. References
4113 to local register variables may be deleted or moved or simplified.
4115 These local variables are sometimes convenient for use with the extended
4116 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4117 output of the assembler instruction directly into a particular register.
4118 (This will work provided the register you specify fits the constraints
4119 specified for that operand in the @code{asm}.)
4127 @node Global Reg Vars
4128 @subsection Defining Global Register Variables
4129 @cindex global register variables
4130 @cindex registers, global variables in
4132 You can define a global register variable in GNU C like this:
4135 register int *foo asm ("a5");
4139 Here @code{a5} is the name of the register which should be used. Choose a
4140 register which is normally saved and restored by function calls on your
4141 machine, so that library routines will not clobber it.
4143 Naturally the register name is cpu-dependent, so you would need to
4144 conditionalize your program according to cpu type. The register
4145 @code{a5} would be a good choice on a 68000 for a variable of pointer
4146 type. On machines with register windows, be sure to choose a ``global''
4147 register that is not affected magically by the function call mechanism.
4149 In addition, operating systems on one type of cpu may differ in how they
4150 name the registers; then you would need additional conditionals. For
4151 example, some 68000 operating systems call this register @code{%a5}.
4153 Eventually there may be a way of asking the compiler to choose a register
4154 automatically, but first we need to figure out how it should choose and
4155 how to enable you to guide the choice. No solution is evident.
4157 Defining a global register variable in a certain register reserves that
4158 register entirely for this use, at least within the current compilation.
4159 The register will not be allocated for any other purpose in the functions
4160 in the current compilation. The register will not be saved and restored by
4161 these functions. Stores into this register are never deleted even if they
4162 would appear to be dead, but references may be deleted or moved or
4165 It is not safe to access the global register variables from signal
4166 handlers, or from more than one thread of control, because the system
4167 library routines may temporarily use the register for other things (unless
4168 you recompile them specially for the task at hand).
4170 @cindex @code{qsort}, and global register variables
4171 It is not safe for one function that uses a global register variable to
4172 call another such function @code{foo} by way of a third function
4173 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4174 different source file in which the variable wasn't declared). This is
4175 because @code{lose} might save the register and put some other value there.
4176 For example, you can't expect a global register variable to be available in
4177 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4178 might have put something else in that register. (If you are prepared to
4179 recompile @code{qsort} with the same global register variable, you can
4180 solve this problem.)
4182 If you want to recompile @code{qsort} or other source files which do not
4183 actually use your global register variable, so that they will not use that
4184 register for any other purpose, then it suffices to specify the compiler
4185 option @option{-ffixed-@var{reg}}. You need not actually add a global
4186 register declaration to their source code.
4188 A function which can alter the value of a global register variable cannot
4189 safely be called from a function compiled without this variable, because it
4190 could clobber the value the caller expects to find there on return.
4191 Therefore, the function which is the entry point into the part of the
4192 program that uses the global register variable must explicitly save and
4193 restore the value which belongs to its caller.
4195 @cindex register variable after @code{longjmp}
4196 @cindex global register after @code{longjmp}
4197 @cindex value after @code{longjmp}
4200 On most machines, @code{longjmp} will restore to each global register
4201 variable the value it had at the time of the @code{setjmp}. On some
4202 machines, however, @code{longjmp} will not change the value of global
4203 register variables. To be portable, the function that called @code{setjmp}
4204 should make other arrangements to save the values of the global register
4205 variables, and to restore them in a @code{longjmp}. This way, the same
4206 thing will happen regardless of what @code{longjmp} does.
4208 All global register variable declarations must precede all function
4209 definitions. If such a declaration could appear after function
4210 definitions, the declaration would be too late to prevent the register from
4211 being used for other purposes in the preceding functions.
4213 Global register variables may not have initial values, because an
4214 executable file has no means to supply initial contents for a register.
4216 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4217 registers, but certain library functions, such as @code{getwd}, as well
4218 as the subroutines for division and remainder, modify g3 and g4. g1 and
4219 g2 are local temporaries.
4221 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4222 Of course, it will not do to use more than a few of those.
4224 @node Local Reg Vars
4225 @subsection Specifying Registers for Local Variables
4226 @cindex local variables, specifying registers
4227 @cindex specifying registers for local variables
4228 @cindex registers for local variables
4230 You can define a local register variable with a specified register
4234 register int *foo asm ("a5");
4238 Here @code{a5} is the name of the register which should be used. Note
4239 that this is the same syntax used for defining global register
4240 variables, but for a local variable it would appear within a function.
4242 Naturally the register name is cpu-dependent, but this is not a
4243 problem, since specific registers are most often useful with explicit
4244 assembler instructions (@pxref{Extended Asm}). Both of these things
4245 generally require that you conditionalize your program according to
4248 In addition, operating systems on one type of cpu may differ in how they
4249 name the registers; then you would need additional conditionals. For
4250 example, some 68000 operating systems call this register @code{%a5}.
4252 Defining such a register variable does not reserve the register; it
4253 remains available for other uses in places where flow control determines
4254 the variable's value is not live.
4256 This option does not guarantee that GCC will generate code that has
4257 this variable in the register you specify at all times. You may not
4258 code an explicit reference to this register in the @emph{assembler
4259 instruction template} part of an @code{asm} statement and assume it will
4260 always refer to this variable. However, using the variable as an
4261 @code{asm} @emph{operand} guarantees that the specified register is used
4264 Stores into local register variables may be deleted when they appear to be dead
4265 according to dataflow analysis. References to local register variables may
4266 be deleted or moved or simplified.
4268 As for global register variables, it's recommended that you choose a
4269 register which is normally saved and restored by function calls on
4270 your machine, so that library routines will not clobber it. A common
4271 pitfall is to initialize multiple call-clobbered registers with
4272 arbitrary expressions, where a function call or library call for an
4273 arithmetic operator will overwrite a register value from a previous
4274 assignment, for example @code{r0} below:
4276 register int *p1 asm ("r0") = @dots{};
4277 register int *p2 asm ("r1") = @dots{};
4279 In those cases, a solution is to use a temporary variable for
4280 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
4282 @node Alternate Keywords
4283 @section Alternate Keywords
4284 @cindex alternate keywords
4285 @cindex keywords, alternate
4287 @option{-ansi} and the various @option{-std} options disable certain
4288 keywords. This causes trouble when you want to use GNU C extensions, or
4289 a general-purpose header file that should be usable by all programs,
4290 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4291 @code{inline} are not available in programs compiled with
4292 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4293 program compiled with @option{-std=c99}). The ISO C99 keyword
4294 @code{restrict} is only available when @option{-std=gnu99} (which will
4295 eventually be the default) or @option{-std=c99} (or the equivalent
4296 @option{-std=iso9899:1999}) is used.
4298 The way to solve these problems is to put @samp{__} at the beginning and
4299 end of each problematical keyword. For example, use @code{__asm__}
4300 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4302 Other C compilers won't accept these alternative keywords; if you want to
4303 compile with another compiler, you can define the alternate keywords as
4304 macros to replace them with the customary keywords. It looks like this:
4312 @findex __extension__
4314 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4316 prevent such warnings within one expression by writing
4317 @code{__extension__} before the expression. @code{__extension__} has no
4318 effect aside from this.
4320 @node Incomplete Enums
4321 @section Incomplete @code{enum} Types
4323 You can define an @code{enum} tag without specifying its possible values.
4324 This results in an incomplete type, much like what you get if you write
4325 @code{struct foo} without describing the elements. A later declaration
4326 which does specify the possible values completes the type.
4328 You can't allocate variables or storage using the type while it is
4329 incomplete. However, you can work with pointers to that type.
4331 This extension may not be very useful, but it makes the handling of
4332 @code{enum} more consistent with the way @code{struct} and @code{union}
4335 This extension is not supported by GNU C++.
4337 @node Function Names
4338 @section Function Names as Strings
4339 @cindex @code{__func__} identifier
4340 @cindex @code{__FUNCTION__} identifier
4341 @cindex @code{__PRETTY_FUNCTION__} identifier
4343 GCC provides three magic variables which hold the name of the current
4344 function, as a string. The first of these is @code{__func__}, which
4345 is part of the C99 standard:
4348 The identifier @code{__func__} is implicitly declared by the translator
4349 as if, immediately following the opening brace of each function
4350 definition, the declaration
4353 static const char __func__[] = "function-name";
4356 appeared, where function-name is the name of the lexically-enclosing
4357 function. This name is the unadorned name of the function.
4360 @code{__FUNCTION__} is another name for @code{__func__}. Older
4361 versions of GCC recognize only this name. However, it is not
4362 standardized. For maximum portability, we recommend you use
4363 @code{__func__}, but provide a fallback definition with the
4367 #if __STDC_VERSION__ < 199901L
4369 # define __func__ __FUNCTION__
4371 # define __func__ "<unknown>"
4376 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4377 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4378 the type signature of the function as well as its bare name. For
4379 example, this program:
4383 extern int printf (char *, ...);
4390 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4391 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4409 __PRETTY_FUNCTION__ = void a::sub(int)
4412 These identifiers are not preprocessor macros. In GCC 3.3 and
4413 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4414 were treated as string literals; they could be used to initialize
4415 @code{char} arrays, and they could be concatenated with other string
4416 literals. GCC 3.4 and later treat them as variables, like
4417 @code{__func__}. In C++, @code{__FUNCTION__} and
4418 @code{__PRETTY_FUNCTION__} have always been variables.
4420 @node Return Address
4421 @section Getting the Return or Frame Address of a Function
4423 These functions may be used to get information about the callers of a
4426 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4427 This function returns the return address of the current function, or of
4428 one of its callers. The @var{level} argument is number of frames to
4429 scan up the call stack. A value of @code{0} yields the return address
4430 of the current function, a value of @code{1} yields the return address
4431 of the caller of the current function, and so forth. When inlining
4432 the expected behavior is that the function will return the address of
4433 the function that will be returned to. To work around this behavior use
4434 the @code{noinline} function attribute.
4436 The @var{level} argument must be a constant integer.
4438 On some machines it may be impossible to determine the return address of
4439 any function other than the current one; in such cases, or when the top
4440 of the stack has been reached, this function will return @code{0} or a
4441 random value. In addition, @code{__builtin_frame_address} may be used
4442 to determine if the top of the stack has been reached.
4444 This function should only be used with a nonzero argument for debugging
4448 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4449 This function is similar to @code{__builtin_return_address}, but it
4450 returns the address of the function frame rather than the return address
4451 of the function. Calling @code{__builtin_frame_address} with a value of
4452 @code{0} yields the frame address of the current function, a value of
4453 @code{1} yields the frame address of the caller of the current function,
4456 The frame is the area on the stack which holds local variables and saved
4457 registers. The frame address is normally the address of the first word
4458 pushed on to the stack by the function. However, the exact definition
4459 depends upon the processor and the calling convention. If the processor
4460 has a dedicated frame pointer register, and the function has a frame,
4461 then @code{__builtin_frame_address} will return the value of the frame
4464 On some machines it may be impossible to determine the frame address of
4465 any function other than the current one; in such cases, or when the top
4466 of the stack has been reached, this function will return @code{0} if
4467 the first frame pointer is properly initialized by the startup code.
4469 This function should only be used with a nonzero argument for debugging
4473 @node Vector Extensions
4474 @section Using vector instructions through built-in functions
4476 On some targets, the instruction set contains SIMD vector instructions that
4477 operate on multiple values contained in one large register at the same time.
4478 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4481 The first step in using these extensions is to provide the necessary data
4482 types. This should be done using an appropriate @code{typedef}:
4485 typedef int v4si __attribute__ ((vector_size (16)));
4488 The @code{int} type specifies the base type, while the attribute specifies
4489 the vector size for the variable, measured in bytes. For example, the
4490 declaration above causes the compiler to set the mode for the @code{v4si}
4491 type to be 16 bytes wide and divided into @code{int} sized units. For
4492 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4493 corresponding mode of @code{foo} will be @acronym{V4SI}.
4495 The @code{vector_size} attribute is only applicable to integral and
4496 float scalars, although arrays, pointers, and function return values
4497 are allowed in conjunction with this construct.
4499 All the basic integer types can be used as base types, both as signed
4500 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4501 @code{long long}. In addition, @code{float} and @code{double} can be
4502 used to build floating-point vector types.
4504 Specifying a combination that is not valid for the current architecture
4505 will cause GCC to synthesize the instructions using a narrower mode.
4506 For example, if you specify a variable of type @code{V4SI} and your
4507 architecture does not allow for this specific SIMD type, GCC will
4508 produce code that uses 4 @code{SIs}.
4510 The types defined in this manner can be used with a subset of normal C
4511 operations. Currently, GCC will allow using the following operators
4512 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4514 The operations behave like C++ @code{valarrays}. Addition is defined as
4515 the addition of the corresponding elements of the operands. For
4516 example, in the code below, each of the 4 elements in @var{a} will be
4517 added to the corresponding 4 elements in @var{b} and the resulting
4518 vector will be stored in @var{c}.
4521 typedef int v4si __attribute__ ((vector_size (16)));
4528 Subtraction, multiplication, division, and the logical operations
4529 operate in a similar manner. Likewise, the result of using the unary
4530 minus or complement operators on a vector type is a vector whose
4531 elements are the negative or complemented values of the corresponding
4532 elements in the operand.
4534 You can declare variables and use them in function calls and returns, as
4535 well as in assignments and some casts. You can specify a vector type as
4536 a return type for a function. Vector types can also be used as function
4537 arguments. It is possible to cast from one vector type to another,
4538 provided they are of the same size (in fact, you can also cast vectors
4539 to and from other datatypes of the same size).
4541 You cannot operate between vectors of different lengths or different
4542 signedness without a cast.
4544 A port that supports hardware vector operations, usually provides a set
4545 of built-in functions that can be used to operate on vectors. For
4546 example, a function to add two vectors and multiply the result by a
4547 third could look like this:
4550 v4si f (v4si a, v4si b, v4si c)
4552 v4si tmp = __builtin_addv4si (a, b);
4553 return __builtin_mulv4si (tmp, c);
4560 @findex __builtin_offsetof
4562 GCC implements for both C and C++ a syntactic extension to implement
4563 the @code{offsetof} macro.
4567 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
4569 offsetof_member_designator:
4571 | offsetof_member_designator "." @code{identifier}
4572 | offsetof_member_designator "[" @code{expr} "]"
4575 This extension is sufficient such that
4578 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
4581 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
4582 may be dependent. In either case, @var{member} may consist of a single
4583 identifier, or a sequence of member accesses and array references.
4585 @node Atomic Builtins
4586 @section Built-in functions for atomic memory access
4588 The following builtins are intended to be compatible with those described
4589 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
4590 section 7.4. As such, they depart from the normal GCC practice of using
4591 the ``__builtin_'' prefix, and further that they are overloaded such that
4592 they work on multiple types.
4594 The definition given in the Intel documentation allows only for the use of
4595 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
4596 counterparts. GCC will allow any integral scalar or pointer type that is
4597 1, 2, 4 or 8 bytes in length.
4599 Not all operations are supported by all target processors. If a particular
4600 operation cannot be implemented on the target processor, a warning will be
4601 generated and a call an external function will be generated. The external
4602 function will carry the same name as the builtin, with an additional suffix
4603 @samp{_@var{n}} where @var{n} is the size of the data type.
4605 @c ??? Should we have a mechanism to suppress this warning? This is almost
4606 @c useful for implementing the operation under the control of an external
4609 In most cases, these builtins are considered a @dfn{full barrier}. That is,
4610 no memory operand will be moved across the operation, either forward or
4611 backward. Further, instructions will be issued as necessary to prevent the
4612 processor from speculating loads across the operation and from queuing stores
4613 after the operation.
4615 All of the routines are are described in the Intel documentation to take
4616 ``an optional list of variables protected by the memory barrier''. It's
4617 not clear what is meant by that; it could mean that @emph{only} the
4618 following variables are protected, or it could mean that these variables
4619 should in addition be protected. At present GCC ignores this list and
4620 protects all variables which are globally accessible. If in the future
4621 we make some use of this list, an empty list will continue to mean all
4622 globally accessible variables.
4625 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
4626 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
4627 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
4628 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
4629 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
4630 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
4631 @findex __sync_fetch_and_add
4632 @findex __sync_fetch_and_sub
4633 @findex __sync_fetch_and_or
4634 @findex __sync_fetch_and_and
4635 @findex __sync_fetch_and_xor
4636 @findex __sync_fetch_and_nand
4637 These builtins perform the operation suggested by the name, and
4638 returns the value that had previously been in memory. That is,
4641 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
4642 @{ tmp = *ptr; *ptr = ~tmp & value; return tmp; @} // nand
4645 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
4646 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
4647 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
4648 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
4649 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
4650 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
4651 @findex __sync_add_and_fetch
4652 @findex __sync_sub_and_fetch
4653 @findex __sync_or_and_fetch
4654 @findex __sync_and_and_fetch
4655 @findex __sync_xor_and_fetch
4656 @findex __sync_nand_and_fetch
4657 These builtins perform the operation suggested by the name, and
4658 return the new value. That is,
4661 @{ *ptr @var{op}= value; return *ptr; @}
4662 @{ *ptr = ~*ptr & value; return *ptr; @} // nand
4665 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
4666 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
4667 @findex __sync_bool_compare_and_swap
4668 @findex __sync_val_compare_and_swap
4669 These builtins perform an atomic compare and swap. That is, if the current
4670 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
4673 The ``bool'' version returns true if the comparison is successful and
4674 @var{newval} was written. The ``val'' version returns the contents
4675 of @code{*@var{ptr}} before the operation.
4677 @item __sync_synchronize (...)
4678 @findex __sync_synchronize
4679 This builtin issues a full memory barrier.
4681 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
4682 @findex __sync_lock_test_and_set
4683 This builtin, as described by Intel, is not a traditional test-and-set
4684 operation, but rather an atomic exchange operation. It writes @var{value}
4685 into @code{*@var{ptr}}, and returns the previous contents of
4688 Many targets have only minimal support for such locks, and do not support
4689 a full exchange operation. In this case, a target may support reduced
4690 functionality here by which the @emph{only} valid value to store is the
4691 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
4692 is implementation defined.
4694 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
4695 This means that references after the builtin cannot move to (or be
4696 speculated to) before the builtin, but previous memory stores may not
4697 be globally visible yet, and previous memory loads may not yet be
4700 @item void __sync_lock_release (@var{type} *ptr, ...)
4701 @findex __sync_lock_release
4702 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
4703 Normally this means writing the constant 0 to @code{*@var{ptr}}.
4705 This builtin is not a full barrier, but rather a @dfn{release barrier}.
4706 This means that all previous memory stores are globally visible, and all
4707 previous memory loads have been satisfied, but following memory reads
4708 are not prevented from being speculated to before the barrier.
4711 @node Other Builtins
4712 @section Other built-in functions provided by GCC
4713 @cindex built-in functions
4714 @findex __builtin_isgreater
4715 @findex __builtin_isgreaterequal
4716 @findex __builtin_isless
4717 @findex __builtin_islessequal
4718 @findex __builtin_islessgreater
4719 @findex __builtin_isunordered
4720 @findex __builtin_powi
4721 @findex __builtin_powif
4722 @findex __builtin_powil
4880 @findex fprintf_unlocked
4882 @findex fputs_unlocked
4992 @findex printf_unlocked
5021 @findex significandf
5022 @findex significandl
5093 GCC provides a large number of built-in functions other than the ones
5094 mentioned above. Some of these are for internal use in the processing
5095 of exceptions or variable-length argument lists and will not be
5096 documented here because they may change from time to time; we do not
5097 recommend general use of these functions.
5099 The remaining functions are provided for optimization purposes.
5101 @opindex fno-builtin
5102 GCC includes built-in versions of many of the functions in the standard
5103 C library. The versions prefixed with @code{__builtin_} will always be
5104 treated as having the same meaning as the C library function even if you
5105 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
5106 Many of these functions are only optimized in certain cases; if they are
5107 not optimized in a particular case, a call to the library function will
5112 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
5113 @option{-std=c99}), the functions
5114 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
5115 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
5116 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
5117 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
5118 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
5119 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
5120 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
5121 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
5122 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
5123 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
5124 @code{significandf}, @code{significandl}, @code{significand},
5125 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
5126 @code{stpncpy}, @code{strcasecmp}, @code{strdup}, @code{strfmon},
5127 @code{strncasecmp}, @code{strndup}, @code{toascii}, @code{y0f},
5128 @code{y0l}, @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf},
5129 @code{ynl} and @code{yn}
5130 may be handled as built-in functions.
5131 All these functions have corresponding versions
5132 prefixed with @code{__builtin_}, which may be used even in strict C89
5135 The ISO C99 functions
5136 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
5137 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
5138 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
5139 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
5140 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
5141 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
5142 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
5143 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
5144 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
5145 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
5146 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
5147 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
5148 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
5149 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
5150 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
5151 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
5152 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
5153 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
5154 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
5155 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
5156 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
5157 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
5158 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
5159 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
5160 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
5161 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
5162 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
5163 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
5164 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
5165 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
5166 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
5167 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
5168 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
5169 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
5170 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
5171 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
5172 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
5173 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
5174 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
5175 are handled as built-in functions
5176 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5178 There are also built-in versions of the ISO C99 functions
5179 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5180 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5181 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5182 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5183 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5184 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5185 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5186 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5187 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5188 that are recognized in any mode since ISO C90 reserves these names for
5189 the purpose to which ISO C99 puts them. All these functions have
5190 corresponding versions prefixed with @code{__builtin_}.
5192 The ISO C94 functions
5193 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
5194 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
5195 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
5197 are handled as built-in functions
5198 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5200 The ISO C90 functions
5201 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5202 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5203 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5204 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
5205 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
5206 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
5207 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
5208 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
5209 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
5210 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
5211 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
5212 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
5213 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
5214 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
5215 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
5216 @code{vprintf} and @code{vsprintf}
5217 are all recognized as built-in functions unless
5218 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5219 is specified for an individual function). All of these functions have
5220 corresponding versions prefixed with @code{__builtin_}.
5222 GCC provides built-in versions of the ISO C99 floating point comparison
5223 macros that avoid raising exceptions for unordered operands. They have
5224 the same names as the standard macros ( @code{isgreater},
5225 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5226 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5227 prefixed. We intend for a library implementor to be able to simply
5228 @code{#define} each standard macro to its built-in equivalent.
5230 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
5232 You can use the built-in function @code{__builtin_types_compatible_p} to
5233 determine whether two types are the same.
5235 This built-in function returns 1 if the unqualified versions of the
5236 types @var{type1} and @var{type2} (which are types, not expressions) are
5237 compatible, 0 otherwise. The result of this built-in function can be
5238 used in integer constant expressions.
5240 This built-in function ignores top level qualifiers (e.g., @code{const},
5241 @code{volatile}). For example, @code{int} is equivalent to @code{const
5244 The type @code{int[]} and @code{int[5]} are compatible. On the other
5245 hand, @code{int} and @code{char *} are not compatible, even if the size
5246 of their types, on the particular architecture are the same. Also, the
5247 amount of pointer indirection is taken into account when determining
5248 similarity. Consequently, @code{short *} is not similar to
5249 @code{short **}. Furthermore, two types that are typedefed are
5250 considered compatible if their underlying types are compatible.
5252 An @code{enum} type is not considered to be compatible with another
5253 @code{enum} type even if both are compatible with the same integer
5254 type; this is what the C standard specifies.
5255 For example, @code{enum @{foo, bar@}} is not similar to
5256 @code{enum @{hot, dog@}}.
5258 You would typically use this function in code whose execution varies
5259 depending on the arguments' types. For example:
5265 if (__builtin_types_compatible_p (typeof (x), long double)) \
5266 tmp = foo_long_double (tmp); \
5267 else if (__builtin_types_compatible_p (typeof (x), double)) \
5268 tmp = foo_double (tmp); \
5269 else if (__builtin_types_compatible_p (typeof (x), float)) \
5270 tmp = foo_float (tmp); \
5277 @emph{Note:} This construct is only available for C@.
5281 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
5283 You can use the built-in function @code{__builtin_choose_expr} to
5284 evaluate code depending on the value of a constant expression. This
5285 built-in function returns @var{exp1} if @var{const_exp}, which is a
5286 constant expression that must be able to be determined at compile time,
5287 is nonzero. Otherwise it returns 0.
5289 This built-in function is analogous to the @samp{? :} operator in C,
5290 except that the expression returned has its type unaltered by promotion
5291 rules. Also, the built-in function does not evaluate the expression
5292 that was not chosen. For example, if @var{const_exp} evaluates to true,
5293 @var{exp2} is not evaluated even if it has side-effects.
5295 This built-in function can return an lvalue if the chosen argument is an
5298 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5299 type. Similarly, if @var{exp2} is returned, its return type is the same
5306 __builtin_choose_expr ( \
5307 __builtin_types_compatible_p (typeof (x), double), \
5309 __builtin_choose_expr ( \
5310 __builtin_types_compatible_p (typeof (x), float), \
5312 /* @r{The void expression results in a compile-time error} \
5313 @r{when assigning the result to something.} */ \
5317 @emph{Note:} This construct is only available for C@. Furthermore, the
5318 unused expression (@var{exp1} or @var{exp2} depending on the value of
5319 @var{const_exp}) may still generate syntax errors. This may change in
5324 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5325 You can use the built-in function @code{__builtin_constant_p} to
5326 determine if a value is known to be constant at compile-time and hence
5327 that GCC can perform constant-folding on expressions involving that
5328 value. The argument of the function is the value to test. The function
5329 returns the integer 1 if the argument is known to be a compile-time
5330 constant and 0 if it is not known to be a compile-time constant. A
5331 return of 0 does not indicate that the value is @emph{not} a constant,
5332 but merely that GCC cannot prove it is a constant with the specified
5333 value of the @option{-O} option.
5335 You would typically use this function in an embedded application where
5336 memory was a critical resource. If you have some complex calculation,
5337 you may want it to be folded if it involves constants, but need to call
5338 a function if it does not. For example:
5341 #define Scale_Value(X) \
5342 (__builtin_constant_p (X) \
5343 ? ((X) * SCALE + OFFSET) : Scale (X))
5346 You may use this built-in function in either a macro or an inline
5347 function. However, if you use it in an inlined function and pass an
5348 argument of the function as the argument to the built-in, GCC will
5349 never return 1 when you call the inline function with a string constant
5350 or compound literal (@pxref{Compound Literals}) and will not return 1
5351 when you pass a constant numeric value to the inline function unless you
5352 specify the @option{-O} option.
5354 You may also use @code{__builtin_constant_p} in initializers for static
5355 data. For instance, you can write
5358 static const int table[] = @{
5359 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5365 This is an acceptable initializer even if @var{EXPRESSION} is not a
5366 constant expression. GCC must be more conservative about evaluating the
5367 built-in in this case, because it has no opportunity to perform
5370 Previous versions of GCC did not accept this built-in in data
5371 initializers. The earliest version where it is completely safe is
5375 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5376 @opindex fprofile-arcs
5377 You may use @code{__builtin_expect} to provide the compiler with
5378 branch prediction information. In general, you should prefer to
5379 use actual profile feedback for this (@option{-fprofile-arcs}), as
5380 programmers are notoriously bad at predicting how their programs
5381 actually perform. However, there are applications in which this
5382 data is hard to collect.
5384 The return value is the value of @var{exp}, which should be an
5385 integral expression. The value of @var{c} must be a compile-time
5386 constant. The semantics of the built-in are that it is expected
5387 that @var{exp} == @var{c}. For example:
5390 if (__builtin_expect (x, 0))
5395 would indicate that we do not expect to call @code{foo}, since
5396 we expect @code{x} to be zero. Since you are limited to integral
5397 expressions for @var{exp}, you should use constructions such as
5400 if (__builtin_expect (ptr != NULL, 1))
5405 when testing pointer or floating-point values.
5408 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5409 This function is used to minimize cache-miss latency by moving data into
5410 a cache before it is accessed.
5411 You can insert calls to @code{__builtin_prefetch} into code for which
5412 you know addresses of data in memory that is likely to be accessed soon.
5413 If the target supports them, data prefetch instructions will be generated.
5414 If the prefetch is done early enough before the access then the data will
5415 be in the cache by the time it is accessed.
5417 The value of @var{addr} is the address of the memory to prefetch.
5418 There are two optional arguments, @var{rw} and @var{locality}.
5419 The value of @var{rw} is a compile-time constant one or zero; one
5420 means that the prefetch is preparing for a write to the memory address
5421 and zero, the default, means that the prefetch is preparing for a read.
5422 The value @var{locality} must be a compile-time constant integer between
5423 zero and three. A value of zero means that the data has no temporal
5424 locality, so it need not be left in the cache after the access. A value
5425 of three means that the data has a high degree of temporal locality and
5426 should be left in all levels of cache possible. Values of one and two
5427 mean, respectively, a low or moderate degree of temporal locality. The
5431 for (i = 0; i < n; i++)
5434 __builtin_prefetch (&a[i+j], 1, 1);
5435 __builtin_prefetch (&b[i+j], 0, 1);
5440 Data prefetch does not generate faults if @var{addr} is invalid, but
5441 the address expression itself must be valid. For example, a prefetch
5442 of @code{p->next} will not fault if @code{p->next} is not a valid
5443 address, but evaluation will fault if @code{p} is not a valid address.
5445 If the target does not support data prefetch, the address expression
5446 is evaluated if it includes side effects but no other code is generated
5447 and GCC does not issue a warning.
5450 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5451 Returns a positive infinity, if supported by the floating-point format,
5452 else @code{DBL_MAX}. This function is suitable for implementing the
5453 ISO C macro @code{HUGE_VAL}.
5456 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5457 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5460 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5461 Similar to @code{__builtin_huge_val}, except the return
5462 type is @code{long double}.
5465 @deftypefn {Built-in Function} double __builtin_inf (void)
5466 Similar to @code{__builtin_huge_val}, except a warning is generated
5467 if the target floating-point format does not support infinities.
5470 @deftypefn {Built-in Function} float __builtin_inff (void)
5471 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5472 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5475 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5476 Similar to @code{__builtin_inf}, except the return
5477 type is @code{long double}.
5480 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5481 This is an implementation of the ISO C99 function @code{nan}.
5483 Since ISO C99 defines this function in terms of @code{strtod}, which we
5484 do not implement, a description of the parsing is in order. The string
5485 is parsed as by @code{strtol}; that is, the base is recognized by
5486 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5487 in the significand such that the least significant bit of the number
5488 is at the least significant bit of the significand. The number is
5489 truncated to fit the significand field provided. The significand is
5490 forced to be a quiet NaN@.
5492 This function, if given a string literal, is evaluated early enough
5493 that it is considered a compile-time constant.
5496 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
5497 Similar to @code{__builtin_nan}, except the return type is @code{float}.
5500 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
5501 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
5504 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
5505 Similar to @code{__builtin_nan}, except the significand is forced
5506 to be a signaling NaN@. The @code{nans} function is proposed by
5507 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
5510 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
5511 Similar to @code{__builtin_nans}, except the return type is @code{float}.
5514 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
5515 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
5518 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
5519 Returns one plus the index of the least significant 1-bit of @var{x}, or
5520 if @var{x} is zero, returns zero.
5523 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
5524 Returns the number of leading 0-bits in @var{x}, starting at the most
5525 significant bit position. If @var{x} is 0, the result is undefined.
5528 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
5529 Returns the number of trailing 0-bits in @var{x}, starting at the least
5530 significant bit position. If @var{x} is 0, the result is undefined.
5533 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
5534 Returns the number of 1-bits in @var{x}.
5537 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
5538 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
5542 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
5543 Similar to @code{__builtin_ffs}, except the argument type is
5544 @code{unsigned long}.
5547 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
5548 Similar to @code{__builtin_clz}, except the argument type is
5549 @code{unsigned long}.
5552 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
5553 Similar to @code{__builtin_ctz}, except the argument type is
5554 @code{unsigned long}.
5557 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
5558 Similar to @code{__builtin_popcount}, except the argument type is
5559 @code{unsigned long}.
5562 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
5563 Similar to @code{__builtin_parity}, except the argument type is
5564 @code{unsigned long}.
5567 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
5568 Similar to @code{__builtin_ffs}, except the argument type is
5569 @code{unsigned long long}.
5572 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
5573 Similar to @code{__builtin_clz}, except the argument type is
5574 @code{unsigned long long}.
5577 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
5578 Similar to @code{__builtin_ctz}, except the argument type is
5579 @code{unsigned long long}.
5582 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
5583 Similar to @code{__builtin_popcount}, except the argument type is
5584 @code{unsigned long long}.
5587 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
5588 Similar to @code{__builtin_parity}, except the argument type is
5589 @code{unsigned long long}.
5592 @deftypefn {Built-in Function} double __builtin_powi (double, int)
5593 Returns the first argument raised to the power of the second. Unlike the
5594 @code{pow} function no guarantees about precision and rounding are made.
5597 @deftypefn {Built-in Function} float __builtin_powif (float, int)
5598 Similar to @code{__builtin_powi}, except the argument and return types
5602 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
5603 Similar to @code{__builtin_powi}, except the argument and return types
5604 are @code{long double}.
5608 @node Target Builtins
5609 @section Built-in Functions Specific to Particular Target Machines
5611 On some target machines, GCC supports many built-in functions specific
5612 to those machines. Generally these generate calls to specific machine
5613 instructions, but allow the compiler to schedule those calls.
5616 * Alpha Built-in Functions::
5617 * ARM Built-in Functions::
5618 * FR-V Built-in Functions::
5619 * X86 Built-in Functions::
5620 * MIPS Paired-Single Support::
5621 * PowerPC AltiVec Built-in Functions::
5622 * SPARC VIS Built-in Functions::
5625 @node Alpha Built-in Functions
5626 @subsection Alpha Built-in Functions
5628 These built-in functions are available for the Alpha family of
5629 processors, depending on the command-line switches used.
5631 The following built-in functions are always available. They
5632 all generate the machine instruction that is part of the name.
5635 long __builtin_alpha_implver (void)
5636 long __builtin_alpha_rpcc (void)
5637 long __builtin_alpha_amask (long)
5638 long __builtin_alpha_cmpbge (long, long)
5639 long __builtin_alpha_extbl (long, long)
5640 long __builtin_alpha_extwl (long, long)
5641 long __builtin_alpha_extll (long, long)
5642 long __builtin_alpha_extql (long, long)
5643 long __builtin_alpha_extwh (long, long)
5644 long __builtin_alpha_extlh (long, long)
5645 long __builtin_alpha_extqh (long, long)
5646 long __builtin_alpha_insbl (long, long)
5647 long __builtin_alpha_inswl (long, long)
5648 long __builtin_alpha_insll (long, long)
5649 long __builtin_alpha_insql (long, long)
5650 long __builtin_alpha_inswh (long, long)
5651 long __builtin_alpha_inslh (long, long)
5652 long __builtin_alpha_insqh (long, long)
5653 long __builtin_alpha_mskbl (long, long)
5654 long __builtin_alpha_mskwl (long, long)
5655 long __builtin_alpha_mskll (long, long)
5656 long __builtin_alpha_mskql (long, long)
5657 long __builtin_alpha_mskwh (long, long)
5658 long __builtin_alpha_msklh (long, long)
5659 long __builtin_alpha_mskqh (long, long)
5660 long __builtin_alpha_umulh (long, long)
5661 long __builtin_alpha_zap (long, long)
5662 long __builtin_alpha_zapnot (long, long)
5665 The following built-in functions are always with @option{-mmax}
5666 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
5667 later. They all generate the machine instruction that is part
5671 long __builtin_alpha_pklb (long)
5672 long __builtin_alpha_pkwb (long)
5673 long __builtin_alpha_unpkbl (long)
5674 long __builtin_alpha_unpkbw (long)
5675 long __builtin_alpha_minub8 (long, long)
5676 long __builtin_alpha_minsb8 (long, long)
5677 long __builtin_alpha_minuw4 (long, long)
5678 long __builtin_alpha_minsw4 (long, long)
5679 long __builtin_alpha_maxub8 (long, long)
5680 long __builtin_alpha_maxsb8 (long, long)
5681 long __builtin_alpha_maxuw4 (long, long)
5682 long __builtin_alpha_maxsw4 (long, long)
5683 long __builtin_alpha_perr (long, long)
5686 The following built-in functions are always with @option{-mcix}
5687 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
5688 later. They all generate the machine instruction that is part
5692 long __builtin_alpha_cttz (long)
5693 long __builtin_alpha_ctlz (long)
5694 long __builtin_alpha_ctpop (long)
5697 The following builtins are available on systems that use the OSF/1
5698 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
5699 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
5700 @code{rdval} and @code{wrval}.
5703 void *__builtin_thread_pointer (void)
5704 void __builtin_set_thread_pointer (void *)
5707 @node ARM Built-in Functions
5708 @subsection ARM Built-in Functions
5710 These built-in functions are available for the ARM family of
5711 processors, when the @option{-mcpu=iwmmxt} switch is used:
5714 typedef int v2si __attribute__ ((vector_size (8)));
5715 typedef short v4hi __attribute__ ((vector_size (8)));
5716 typedef char v8qi __attribute__ ((vector_size (8)));
5718 int __builtin_arm_getwcx (int)
5719 void __builtin_arm_setwcx (int, int)
5720 int __builtin_arm_textrmsb (v8qi, int)
5721 int __builtin_arm_textrmsh (v4hi, int)
5722 int __builtin_arm_textrmsw (v2si, int)
5723 int __builtin_arm_textrmub (v8qi, int)
5724 int __builtin_arm_textrmuh (v4hi, int)
5725 int __builtin_arm_textrmuw (v2si, int)
5726 v8qi __builtin_arm_tinsrb (v8qi, int)
5727 v4hi __builtin_arm_tinsrh (v4hi, int)
5728 v2si __builtin_arm_tinsrw (v2si, int)
5729 long long __builtin_arm_tmia (long long, int, int)
5730 long long __builtin_arm_tmiabb (long long, int, int)
5731 long long __builtin_arm_tmiabt (long long, int, int)
5732 long long __builtin_arm_tmiaph (long long, int, int)
5733 long long __builtin_arm_tmiatb (long long, int, int)
5734 long long __builtin_arm_tmiatt (long long, int, int)
5735 int __builtin_arm_tmovmskb (v8qi)
5736 int __builtin_arm_tmovmskh (v4hi)
5737 int __builtin_arm_tmovmskw (v2si)
5738 long long __builtin_arm_waccb (v8qi)
5739 long long __builtin_arm_wacch (v4hi)
5740 long long __builtin_arm_waccw (v2si)
5741 v8qi __builtin_arm_waddb (v8qi, v8qi)
5742 v8qi __builtin_arm_waddbss (v8qi, v8qi)
5743 v8qi __builtin_arm_waddbus (v8qi, v8qi)
5744 v4hi __builtin_arm_waddh (v4hi, v4hi)
5745 v4hi __builtin_arm_waddhss (v4hi, v4hi)
5746 v4hi __builtin_arm_waddhus (v4hi, v4hi)
5747 v2si __builtin_arm_waddw (v2si, v2si)
5748 v2si __builtin_arm_waddwss (v2si, v2si)
5749 v2si __builtin_arm_waddwus (v2si, v2si)
5750 v8qi __builtin_arm_walign (v8qi, v8qi, int)
5751 long long __builtin_arm_wand(long long, long long)
5752 long long __builtin_arm_wandn (long long, long long)
5753 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
5754 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
5755 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
5756 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
5757 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
5758 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
5759 v2si __builtin_arm_wcmpeqw (v2si, v2si)
5760 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
5761 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
5762 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
5763 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
5764 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
5765 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
5766 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
5767 long long __builtin_arm_wmacsz (v4hi, v4hi)
5768 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
5769 long long __builtin_arm_wmacuz (v4hi, v4hi)
5770 v4hi __builtin_arm_wmadds (v4hi, v4hi)
5771 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
5772 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
5773 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
5774 v2si __builtin_arm_wmaxsw (v2si, v2si)
5775 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
5776 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
5777 v2si __builtin_arm_wmaxuw (v2si, v2si)
5778 v8qi __builtin_arm_wminsb (v8qi, v8qi)
5779 v4hi __builtin_arm_wminsh (v4hi, v4hi)
5780 v2si __builtin_arm_wminsw (v2si, v2si)
5781 v8qi __builtin_arm_wminub (v8qi, v8qi)
5782 v4hi __builtin_arm_wminuh (v4hi, v4hi)
5783 v2si __builtin_arm_wminuw (v2si, v2si)
5784 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
5785 v4hi __builtin_arm_wmulul (v4hi, v4hi)
5786 v4hi __builtin_arm_wmulum (v4hi, v4hi)
5787 long long __builtin_arm_wor (long long, long long)
5788 v2si __builtin_arm_wpackdss (long long, long long)
5789 v2si __builtin_arm_wpackdus (long long, long long)
5790 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
5791 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
5792 v4hi __builtin_arm_wpackwss (v2si, v2si)
5793 v4hi __builtin_arm_wpackwus (v2si, v2si)
5794 long long __builtin_arm_wrord (long long, long long)
5795 long long __builtin_arm_wrordi (long long, int)
5796 v4hi __builtin_arm_wrorh (v4hi, long long)
5797 v4hi __builtin_arm_wrorhi (v4hi, int)
5798 v2si __builtin_arm_wrorw (v2si, long long)
5799 v2si __builtin_arm_wrorwi (v2si, int)
5800 v2si __builtin_arm_wsadb (v8qi, v8qi)
5801 v2si __builtin_arm_wsadbz (v8qi, v8qi)
5802 v2si __builtin_arm_wsadh (v4hi, v4hi)
5803 v2si __builtin_arm_wsadhz (v4hi, v4hi)
5804 v4hi __builtin_arm_wshufh (v4hi, int)
5805 long long __builtin_arm_wslld (long long, long long)
5806 long long __builtin_arm_wslldi (long long, int)
5807 v4hi __builtin_arm_wsllh (v4hi, long long)
5808 v4hi __builtin_arm_wsllhi (v4hi, int)
5809 v2si __builtin_arm_wsllw (v2si, long long)
5810 v2si __builtin_arm_wsllwi (v2si, int)
5811 long long __builtin_arm_wsrad (long long, long long)
5812 long long __builtin_arm_wsradi (long long, int)
5813 v4hi __builtin_arm_wsrah (v4hi, long long)
5814 v4hi __builtin_arm_wsrahi (v4hi, int)
5815 v2si __builtin_arm_wsraw (v2si, long long)
5816 v2si __builtin_arm_wsrawi (v2si, int)
5817 long long __builtin_arm_wsrld (long long, long long)
5818 long long __builtin_arm_wsrldi (long long, int)
5819 v4hi __builtin_arm_wsrlh (v4hi, long long)
5820 v4hi __builtin_arm_wsrlhi (v4hi, int)
5821 v2si __builtin_arm_wsrlw (v2si, long long)
5822 v2si __builtin_arm_wsrlwi (v2si, int)
5823 v8qi __builtin_arm_wsubb (v8qi, v8qi)
5824 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
5825 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
5826 v4hi __builtin_arm_wsubh (v4hi, v4hi)
5827 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
5828 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
5829 v2si __builtin_arm_wsubw (v2si, v2si)
5830 v2si __builtin_arm_wsubwss (v2si, v2si)
5831 v2si __builtin_arm_wsubwus (v2si, v2si)
5832 v4hi __builtin_arm_wunpckehsb (v8qi)
5833 v2si __builtin_arm_wunpckehsh (v4hi)
5834 long long __builtin_arm_wunpckehsw (v2si)
5835 v4hi __builtin_arm_wunpckehub (v8qi)
5836 v2si __builtin_arm_wunpckehuh (v4hi)
5837 long long __builtin_arm_wunpckehuw (v2si)
5838 v4hi __builtin_arm_wunpckelsb (v8qi)
5839 v2si __builtin_arm_wunpckelsh (v4hi)
5840 long long __builtin_arm_wunpckelsw (v2si)
5841 v4hi __builtin_arm_wunpckelub (v8qi)
5842 v2si __builtin_arm_wunpckeluh (v4hi)
5843 long long __builtin_arm_wunpckeluw (v2si)
5844 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
5845 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
5846 v2si __builtin_arm_wunpckihw (v2si, v2si)
5847 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
5848 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
5849 v2si __builtin_arm_wunpckilw (v2si, v2si)
5850 long long __builtin_arm_wxor (long long, long long)
5851 long long __builtin_arm_wzero ()
5854 @node FR-V Built-in Functions
5855 @subsection FR-V Built-in Functions
5857 GCC provides many FR-V-specific built-in functions. In general,
5858 these functions are intended to be compatible with those described
5859 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
5860 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
5861 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
5862 pointer rather than by value.
5864 Most of the functions are named after specific FR-V instructions.
5865 Such functions are said to be ``directly mapped'' and are summarized
5866 here in tabular form.
5870 * Directly-mapped Integer Functions::
5871 * Directly-mapped Media Functions::
5872 * Other Built-in Functions::
5875 @node Argument Types
5876 @subsubsection Argument Types
5878 The arguments to the built-in functions can be divided into three groups:
5879 register numbers, compile-time constants and run-time values. In order
5880 to make this classification clear at a glance, the arguments and return
5881 values are given the following pseudo types:
5883 @multitable @columnfractions .20 .30 .15 .35
5884 @item Pseudo type @tab Real C type @tab Constant? @tab Description
5885 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
5886 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
5887 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
5888 @item @code{uw2} @tab @code{unsigned long long} @tab No
5889 @tab an unsigned doubleword
5890 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
5891 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
5892 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
5893 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
5896 These pseudo types are not defined by GCC, they are simply a notational
5897 convenience used in this manual.
5899 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
5900 and @code{sw2} are evaluated at run time. They correspond to
5901 register operands in the underlying FR-V instructions.
5903 @code{const} arguments represent immediate operands in the underlying
5904 FR-V instructions. They must be compile-time constants.
5906 @code{acc} arguments are evaluated at compile time and specify the number
5907 of an accumulator register. For example, an @code{acc} argument of 2
5908 will select the ACC2 register.
5910 @code{iacc} arguments are similar to @code{acc} arguments but specify the
5911 number of an IACC register. See @pxref{Other Built-in Functions}
5914 @node Directly-mapped Integer Functions
5915 @subsubsection Directly-mapped Integer Functions
5917 The functions listed below map directly to FR-V I-type instructions.
5919 @multitable @columnfractions .45 .32 .23
5920 @item Function prototype @tab Example usage @tab Assembly output
5921 @item @code{sw1 __ADDSS (sw1, sw1)}
5922 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
5923 @tab @code{ADDSS @var{a},@var{b},@var{c}}
5924 @item @code{sw1 __SCAN (sw1, sw1)}
5925 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
5926 @tab @code{SCAN @var{a},@var{b},@var{c}}
5927 @item @code{sw1 __SCUTSS (sw1)}
5928 @tab @code{@var{b} = __SCUTSS (@var{a})}
5929 @tab @code{SCUTSS @var{a},@var{b}}
5930 @item @code{sw1 __SLASS (sw1, sw1)}
5931 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
5932 @tab @code{SLASS @var{a},@var{b},@var{c}}
5933 @item @code{void __SMASS (sw1, sw1)}
5934 @tab @code{__SMASS (@var{a}, @var{b})}
5935 @tab @code{SMASS @var{a},@var{b}}
5936 @item @code{void __SMSSS (sw1, sw1)}
5937 @tab @code{__SMSSS (@var{a}, @var{b})}
5938 @tab @code{SMSSS @var{a},@var{b}}
5939 @item @code{void __SMU (sw1, sw1)}
5940 @tab @code{__SMU (@var{a}, @var{b})}
5941 @tab @code{SMU @var{a},@var{b}}
5942 @item @code{sw2 __SMUL (sw1, sw1)}
5943 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
5944 @tab @code{SMUL @var{a},@var{b},@var{c}}
5945 @item @code{sw1 __SUBSS (sw1, sw1)}
5946 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
5947 @tab @code{SUBSS @var{a},@var{b},@var{c}}
5948 @item @code{uw2 __UMUL (uw1, uw1)}
5949 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
5950 @tab @code{UMUL @var{a},@var{b},@var{c}}
5953 @node Directly-mapped Media Functions
5954 @subsubsection Directly-mapped Media Functions
5956 The functions listed below map directly to FR-V M-type instructions.
5958 @multitable @columnfractions .45 .32 .23
5959 @item Function prototype @tab Example usage @tab Assembly output
5960 @item @code{uw1 __MABSHS (sw1)}
5961 @tab @code{@var{b} = __MABSHS (@var{a})}
5962 @tab @code{MABSHS @var{a},@var{b}}
5963 @item @code{void __MADDACCS (acc, acc)}
5964 @tab @code{__MADDACCS (@var{b}, @var{a})}
5965 @tab @code{MADDACCS @var{a},@var{b}}
5966 @item @code{sw1 __MADDHSS (sw1, sw1)}
5967 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
5968 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
5969 @item @code{uw1 __MADDHUS (uw1, uw1)}
5970 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
5971 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
5972 @item @code{uw1 __MAND (uw1, uw1)}
5973 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
5974 @tab @code{MAND @var{a},@var{b},@var{c}}
5975 @item @code{void __MASACCS (acc, acc)}
5976 @tab @code{__MASACCS (@var{b}, @var{a})}
5977 @tab @code{MASACCS @var{a},@var{b}}
5978 @item @code{uw1 __MAVEH (uw1, uw1)}
5979 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
5980 @tab @code{MAVEH @var{a},@var{b},@var{c}}
5981 @item @code{uw2 __MBTOH (uw1)}
5982 @tab @code{@var{b} = __MBTOH (@var{a})}
5983 @tab @code{MBTOH @var{a},@var{b}}
5984 @item @code{void __MBTOHE (uw1 *, uw1)}
5985 @tab @code{__MBTOHE (&@var{b}, @var{a})}
5986 @tab @code{MBTOHE @var{a},@var{b}}
5987 @item @code{void __MCLRACC (acc)}
5988 @tab @code{__MCLRACC (@var{a})}
5989 @tab @code{MCLRACC @var{a}}
5990 @item @code{void __MCLRACCA (void)}
5991 @tab @code{__MCLRACCA ()}
5992 @tab @code{MCLRACCA}
5993 @item @code{uw1 __Mcop1 (uw1, uw1)}
5994 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
5995 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
5996 @item @code{uw1 __Mcop2 (uw1, uw1)}
5997 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
5998 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
5999 @item @code{uw1 __MCPLHI (uw2, const)}
6000 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
6001 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
6002 @item @code{uw1 __MCPLI (uw2, const)}
6003 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
6004 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
6005 @item @code{void __MCPXIS (acc, sw1, sw1)}
6006 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
6007 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
6008 @item @code{void __MCPXIU (acc, uw1, uw1)}
6009 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
6010 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
6011 @item @code{void __MCPXRS (acc, sw1, sw1)}
6012 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
6013 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
6014 @item @code{void __MCPXRU (acc, uw1, uw1)}
6015 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
6016 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
6017 @item @code{uw1 __MCUT (acc, uw1)}
6018 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
6019 @tab @code{MCUT @var{a},@var{b},@var{c}}
6020 @item @code{uw1 __MCUTSS (acc, sw1)}
6021 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
6022 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
6023 @item @code{void __MDADDACCS (acc, acc)}
6024 @tab @code{__MDADDACCS (@var{b}, @var{a})}
6025 @tab @code{MDADDACCS @var{a},@var{b}}
6026 @item @code{void __MDASACCS (acc, acc)}
6027 @tab @code{__MDASACCS (@var{b}, @var{a})}
6028 @tab @code{MDASACCS @var{a},@var{b}}
6029 @item @code{uw2 __MDCUTSSI (acc, const)}
6030 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
6031 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
6032 @item @code{uw2 __MDPACKH (uw2, uw2)}
6033 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
6034 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
6035 @item @code{uw2 __MDROTLI (uw2, const)}
6036 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
6037 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
6038 @item @code{void __MDSUBACCS (acc, acc)}
6039 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
6040 @tab @code{MDSUBACCS @var{a},@var{b}}
6041 @item @code{void __MDUNPACKH (uw1 *, uw2)}
6042 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
6043 @tab @code{MDUNPACKH @var{a},@var{b}}
6044 @item @code{uw2 __MEXPDHD (uw1, const)}
6045 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
6046 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
6047 @item @code{uw1 __MEXPDHW (uw1, const)}
6048 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
6049 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
6050 @item @code{uw1 __MHDSETH (uw1, const)}
6051 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
6052 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
6053 @item @code{sw1 __MHDSETS (const)}
6054 @tab @code{@var{b} = __MHDSETS (@var{a})}
6055 @tab @code{MHDSETS #@var{a},@var{b}}
6056 @item @code{uw1 __MHSETHIH (uw1, const)}
6057 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
6058 @tab @code{MHSETHIH #@var{a},@var{b}}
6059 @item @code{sw1 __MHSETHIS (sw1, const)}
6060 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
6061 @tab @code{MHSETHIS #@var{a},@var{b}}
6062 @item @code{uw1 __MHSETLOH (uw1, const)}
6063 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
6064 @tab @code{MHSETLOH #@var{a},@var{b}}
6065 @item @code{sw1 __MHSETLOS (sw1, const)}
6066 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
6067 @tab @code{MHSETLOS #@var{a},@var{b}}
6068 @item @code{uw1 __MHTOB (uw2)}
6069 @tab @code{@var{b} = __MHTOB (@var{a})}
6070 @tab @code{MHTOB @var{a},@var{b}}
6071 @item @code{void __MMACHS (acc, sw1, sw1)}
6072 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
6073 @tab @code{MMACHS @var{a},@var{b},@var{c}}
6074 @item @code{void __MMACHU (acc, uw1, uw1)}
6075 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
6076 @tab @code{MMACHU @var{a},@var{b},@var{c}}
6077 @item @code{void __MMRDHS (acc, sw1, sw1)}
6078 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
6079 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
6080 @item @code{void __MMRDHU (acc, uw1, uw1)}
6081 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
6082 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
6083 @item @code{void __MMULHS (acc, sw1, sw1)}
6084 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
6085 @tab @code{MMULHS @var{a},@var{b},@var{c}}
6086 @item @code{void __MMULHU (acc, uw1, uw1)}
6087 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
6088 @tab @code{MMULHU @var{a},@var{b},@var{c}}
6089 @item @code{void __MMULXHS (acc, sw1, sw1)}
6090 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
6091 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
6092 @item @code{void __MMULXHU (acc, uw1, uw1)}
6093 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
6094 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
6095 @item @code{uw1 __MNOT (uw1)}
6096 @tab @code{@var{b} = __MNOT (@var{a})}
6097 @tab @code{MNOT @var{a},@var{b}}
6098 @item @code{uw1 __MOR (uw1, uw1)}
6099 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
6100 @tab @code{MOR @var{a},@var{b},@var{c}}
6101 @item @code{uw1 __MPACKH (uh, uh)}
6102 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
6103 @tab @code{MPACKH @var{a},@var{b},@var{c}}
6104 @item @code{sw2 __MQADDHSS (sw2, sw2)}
6105 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
6106 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
6107 @item @code{uw2 __MQADDHUS (uw2, uw2)}
6108 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
6109 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
6110 @item @code{void __MQCPXIS (acc, sw2, sw2)}
6111 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
6112 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
6113 @item @code{void __MQCPXIU (acc, uw2, uw2)}
6114 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
6115 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
6116 @item @code{void __MQCPXRS (acc, sw2, sw2)}
6117 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
6118 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
6119 @item @code{void __MQCPXRU (acc, uw2, uw2)}
6120 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
6121 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
6122 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
6123 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
6124 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
6125 @item @code{sw2 __MQLMTHS (sw2, sw2)}
6126 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
6127 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
6128 @item @code{void __MQMACHS (acc, sw2, sw2)}
6129 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
6130 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
6131 @item @code{void __MQMACHU (acc, uw2, uw2)}
6132 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
6133 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
6134 @item @code{void __MQMACXHS (acc, sw2, sw2)}
6135 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
6136 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
6137 @item @code{void __MQMULHS (acc, sw2, sw2)}
6138 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
6139 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
6140 @item @code{void __MQMULHU (acc, uw2, uw2)}
6141 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
6142 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
6143 @item @code{void __MQMULXHS (acc, sw2, sw2)}
6144 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
6145 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
6146 @item @code{void __MQMULXHU (acc, uw2, uw2)}
6147 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
6148 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
6149 @item @code{sw2 __MQSATHS (sw2, sw2)}
6150 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
6151 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
6152 @item @code{uw2 __MQSLLHI (uw2, int)}
6153 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
6154 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
6155 @item @code{sw2 __MQSRAHI (sw2, int)}
6156 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
6157 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
6158 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
6159 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
6160 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
6161 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
6162 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
6163 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
6164 @item @code{void __MQXMACHS (acc, sw2, sw2)}
6165 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
6166 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
6167 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
6168 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
6169 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
6170 @item @code{uw1 __MRDACC (acc)}
6171 @tab @code{@var{b} = __MRDACC (@var{a})}
6172 @tab @code{MRDACC @var{a},@var{b}}
6173 @item @code{uw1 __MRDACCG (acc)}
6174 @tab @code{@var{b} = __MRDACCG (@var{a})}
6175 @tab @code{MRDACCG @var{a},@var{b}}
6176 @item @code{uw1 __MROTLI (uw1, const)}
6177 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
6178 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
6179 @item @code{uw1 __MROTRI (uw1, const)}
6180 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
6181 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
6182 @item @code{sw1 __MSATHS (sw1, sw1)}
6183 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
6184 @tab @code{MSATHS @var{a},@var{b},@var{c}}
6185 @item @code{uw1 __MSATHU (uw1, uw1)}
6186 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
6187 @tab @code{MSATHU @var{a},@var{b},@var{c}}
6188 @item @code{uw1 __MSLLHI (uw1, const)}
6189 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
6190 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
6191 @item @code{sw1 __MSRAHI (sw1, const)}
6192 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
6193 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
6194 @item @code{uw1 __MSRLHI (uw1, const)}
6195 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
6196 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
6197 @item @code{void __MSUBACCS (acc, acc)}
6198 @tab @code{__MSUBACCS (@var{b}, @var{a})}
6199 @tab @code{MSUBACCS @var{a},@var{b}}
6200 @item @code{sw1 __MSUBHSS (sw1, sw1)}
6201 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
6202 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
6203 @item @code{uw1 __MSUBHUS (uw1, uw1)}
6204 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
6205 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
6206 @item @code{void __MTRAP (void)}
6207 @tab @code{__MTRAP ()}
6209 @item @code{uw2 __MUNPACKH (uw1)}
6210 @tab @code{@var{b} = __MUNPACKH (@var{a})}
6211 @tab @code{MUNPACKH @var{a},@var{b}}
6212 @item @code{uw1 __MWCUT (uw2, uw1)}
6213 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
6214 @tab @code{MWCUT @var{a},@var{b},@var{c}}
6215 @item @code{void __MWTACC (acc, uw1)}
6216 @tab @code{__MWTACC (@var{b}, @var{a})}
6217 @tab @code{MWTACC @var{a},@var{b}}
6218 @item @code{void __MWTACCG (acc, uw1)}
6219 @tab @code{__MWTACCG (@var{b}, @var{a})}
6220 @tab @code{MWTACCG @var{a},@var{b}}
6221 @item @code{uw1 __MXOR (uw1, uw1)}
6222 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
6223 @tab @code{MXOR @var{a},@var{b},@var{c}}
6226 @node Other Built-in Functions
6227 @subsubsection Other Built-in Functions
6229 This section describes built-in functions that are not named after
6230 a specific FR-V instruction.
6233 @item sw2 __IACCreadll (iacc @var{reg})
6234 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
6235 for future expansion and must be 0.
6237 @item sw1 __IACCreadl (iacc @var{reg})
6238 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
6239 Other values of @var{reg} are rejected as invalid.
6241 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
6242 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
6243 is reserved for future expansion and must be 0.
6245 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
6246 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
6247 is 1. Other values of @var{reg} are rejected as invalid.
6249 @item void __data_prefetch0 (const void *@var{x})
6250 Use the @code{dcpl} instruction to load the contents of address @var{x}
6251 into the data cache.
6253 @item void __data_prefetch (const void *@var{x})
6254 Use the @code{nldub} instruction to load the contents of address @var{x}
6255 into the data cache. The instruction will be issued in slot I1@.
6258 @node X86 Built-in Functions
6259 @subsection X86 Built-in Functions
6261 These built-in functions are available for the i386 and x86-64 family
6262 of computers, depending on the command-line switches used.
6264 The following machine modes are available for use with MMX built-in functions
6265 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
6266 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
6267 vector of eight 8-bit integers. Some of the built-in functions operate on
6268 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
6270 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
6271 of two 32-bit floating point values.
6273 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
6274 floating point values. Some instructions use a vector of four 32-bit
6275 integers, these use @code{V4SI}. Finally, some instructions operate on an
6276 entire vector register, interpreting it as a 128-bit integer, these use mode
6279 The following built-in functions are made available by @option{-mmmx}.
6280 All of them generate the machine instruction that is part of the name.
6283 v8qi __builtin_ia32_paddb (v8qi, v8qi)
6284 v4hi __builtin_ia32_paddw (v4hi, v4hi)
6285 v2si __builtin_ia32_paddd (v2si, v2si)
6286 v8qi __builtin_ia32_psubb (v8qi, v8qi)
6287 v4hi __builtin_ia32_psubw (v4hi, v4hi)
6288 v2si __builtin_ia32_psubd (v2si, v2si)
6289 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
6290 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
6291 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
6292 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
6293 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
6294 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
6295 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
6296 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
6297 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
6298 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
6299 di __builtin_ia32_pand (di, di)
6300 di __builtin_ia32_pandn (di,di)
6301 di __builtin_ia32_por (di, di)
6302 di __builtin_ia32_pxor (di, di)
6303 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
6304 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
6305 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
6306 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
6307 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
6308 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
6309 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
6310 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
6311 v2si __builtin_ia32_punpckhdq (v2si, v2si)
6312 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
6313 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
6314 v2si __builtin_ia32_punpckldq (v2si, v2si)
6315 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
6316 v4hi __builtin_ia32_packssdw (v2si, v2si)
6317 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
6320 The following built-in functions are made available either with
6321 @option{-msse}, or with a combination of @option{-m3dnow} and
6322 @option{-march=athlon}. All of them generate the machine
6323 instruction that is part of the name.
6326 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
6327 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
6328 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
6329 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
6330 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
6331 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
6332 v8qi __builtin_ia32_pminub (v8qi, v8qi)
6333 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
6334 int __builtin_ia32_pextrw (v4hi, int)
6335 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
6336 int __builtin_ia32_pmovmskb (v8qi)
6337 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
6338 void __builtin_ia32_movntq (di *, di)
6339 void __builtin_ia32_sfence (void)
6342 The following built-in functions are available when @option{-msse} is used.
6343 All of them generate the machine instruction that is part of the name.
6346 int __builtin_ia32_comieq (v4sf, v4sf)
6347 int __builtin_ia32_comineq (v4sf, v4sf)
6348 int __builtin_ia32_comilt (v4sf, v4sf)
6349 int __builtin_ia32_comile (v4sf, v4sf)
6350 int __builtin_ia32_comigt (v4sf, v4sf)
6351 int __builtin_ia32_comige (v4sf, v4sf)
6352 int __builtin_ia32_ucomieq (v4sf, v4sf)
6353 int __builtin_ia32_ucomineq (v4sf, v4sf)
6354 int __builtin_ia32_ucomilt (v4sf, v4sf)
6355 int __builtin_ia32_ucomile (v4sf, v4sf)
6356 int __builtin_ia32_ucomigt (v4sf, v4sf)
6357 int __builtin_ia32_ucomige (v4sf, v4sf)
6358 v4sf __builtin_ia32_addps (v4sf, v4sf)
6359 v4sf __builtin_ia32_subps (v4sf, v4sf)
6360 v4sf __builtin_ia32_mulps (v4sf, v4sf)
6361 v4sf __builtin_ia32_divps (v4sf, v4sf)
6362 v4sf __builtin_ia32_addss (v4sf, v4sf)
6363 v4sf __builtin_ia32_subss (v4sf, v4sf)
6364 v4sf __builtin_ia32_mulss (v4sf, v4sf)
6365 v4sf __builtin_ia32_divss (v4sf, v4sf)
6366 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
6367 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
6368 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
6369 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
6370 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
6371 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
6372 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
6373 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
6374 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
6375 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
6376 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
6377 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
6378 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
6379 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
6380 v4si __builtin_ia32_cmpless (v4sf, v4sf)
6381 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
6382 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
6383 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
6384 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
6385 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
6386 v4sf __builtin_ia32_maxps (v4sf, v4sf)
6387 v4sf __builtin_ia32_maxss (v4sf, v4sf)
6388 v4sf __builtin_ia32_minps (v4sf, v4sf)
6389 v4sf __builtin_ia32_minss (v4sf, v4sf)
6390 v4sf __builtin_ia32_andps (v4sf, v4sf)
6391 v4sf __builtin_ia32_andnps (v4sf, v4sf)
6392 v4sf __builtin_ia32_orps (v4sf, v4sf)
6393 v4sf __builtin_ia32_xorps (v4sf, v4sf)
6394 v4sf __builtin_ia32_movss (v4sf, v4sf)
6395 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
6396 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
6397 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
6398 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
6399 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
6400 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
6401 v2si __builtin_ia32_cvtps2pi (v4sf)
6402 int __builtin_ia32_cvtss2si (v4sf)
6403 v2si __builtin_ia32_cvttps2pi (v4sf)
6404 int __builtin_ia32_cvttss2si (v4sf)
6405 v4sf __builtin_ia32_rcpps (v4sf)
6406 v4sf __builtin_ia32_rsqrtps (v4sf)
6407 v4sf __builtin_ia32_sqrtps (v4sf)
6408 v4sf __builtin_ia32_rcpss (v4sf)
6409 v4sf __builtin_ia32_rsqrtss (v4sf)
6410 v4sf __builtin_ia32_sqrtss (v4sf)
6411 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
6412 void __builtin_ia32_movntps (float *, v4sf)
6413 int __builtin_ia32_movmskps (v4sf)
6416 The following built-in functions are available when @option{-msse} is used.
6419 @item v4sf __builtin_ia32_loadaps (float *)
6420 Generates the @code{movaps} machine instruction as a load from memory.
6421 @item void __builtin_ia32_storeaps (float *, v4sf)
6422 Generates the @code{movaps} machine instruction as a store to memory.
6423 @item v4sf __builtin_ia32_loadups (float *)
6424 Generates the @code{movups} machine instruction as a load from memory.
6425 @item void __builtin_ia32_storeups (float *, v4sf)
6426 Generates the @code{movups} machine instruction as a store to memory.
6427 @item v4sf __builtin_ia32_loadsss (float *)
6428 Generates the @code{movss} machine instruction as a load from memory.
6429 @item void __builtin_ia32_storess (float *, v4sf)
6430 Generates the @code{movss} machine instruction as a store to memory.
6431 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
6432 Generates the @code{movhps} machine instruction as a load from memory.
6433 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
6434 Generates the @code{movlps} machine instruction as a load from memory
6435 @item void __builtin_ia32_storehps (v4sf, v2si *)
6436 Generates the @code{movhps} machine instruction as a store to memory.
6437 @item void __builtin_ia32_storelps (v4sf, v2si *)
6438 Generates the @code{movlps} machine instruction as a store to memory.
6441 The following built-in functions are available when @option{-msse3} is used.
6442 All of them generate the machine instruction that is part of the name.
6445 v2df __builtin_ia32_addsubpd (v2df, v2df)
6446 v2df __builtin_ia32_addsubps (v2df, v2df)
6447 v2df __builtin_ia32_haddpd (v2df, v2df)
6448 v2df __builtin_ia32_haddps (v2df, v2df)
6449 v2df __builtin_ia32_hsubpd (v2df, v2df)
6450 v2df __builtin_ia32_hsubps (v2df, v2df)
6451 v16qi __builtin_ia32_lddqu (char const *)
6452 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
6453 v2df __builtin_ia32_movddup (v2df)
6454 v4sf __builtin_ia32_movshdup (v4sf)
6455 v4sf __builtin_ia32_movsldup (v4sf)
6456 void __builtin_ia32_mwait (unsigned int, unsigned int)
6459 The following built-in functions are available when @option{-msse3} is used.
6462 @item v2df __builtin_ia32_loadddup (double const *)
6463 Generates the @code{movddup} machine instruction as a load from memory.
6466 The following built-in functions are available when @option{-m3dnow} is used.
6467 All of them generate the machine instruction that is part of the name.
6470 void __builtin_ia32_femms (void)
6471 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
6472 v2si __builtin_ia32_pf2id (v2sf)
6473 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
6474 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
6475 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
6476 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
6477 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
6478 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
6479 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
6480 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
6481 v2sf __builtin_ia32_pfrcp (v2sf)
6482 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
6483 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
6484 v2sf __builtin_ia32_pfrsqrt (v2sf)
6485 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
6486 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
6487 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
6488 v2sf __builtin_ia32_pi2fd (v2si)
6489 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
6492 The following built-in functions are available when both @option{-m3dnow}
6493 and @option{-march=athlon} are used. All of them generate the machine
6494 instruction that is part of the name.
6497 v2si __builtin_ia32_pf2iw (v2sf)
6498 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
6499 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
6500 v2sf __builtin_ia32_pi2fw (v2si)
6501 v2sf __builtin_ia32_pswapdsf (v2sf)
6502 v2si __builtin_ia32_pswapdsi (v2si)
6505 @node MIPS Paired-Single Support
6506 @subsection MIPS Paired-Single Support
6508 The MIPS64 architecture includes a number of instructions that
6509 operate on pairs of single-precision floating-point values.
6510 Each pair is packed into a 64-bit floating-point register,
6511 with one element being designated the ``upper half'' and
6512 the other being designated the ``lower half''.
6514 GCC supports paired-single operations using both the generic
6515 vector extensions (@pxref{Vector Extensions}) and a collection of
6516 MIPS-specific built-in functions. Both kinds of support are
6517 enabled by the @option{-mpaired-single} command-line option.
6519 The vector type associated with paired-single values is usually
6520 called @code{v2sf}. It can be defined in C as follows:
6523 typedef float v2sf __attribute__ ((vector_size (8)));
6526 @code{v2sf} values are initialized in the same way as aggregates.
6530 v2sf a = @{1.5, 9.1@};
6533 b = (v2sf) @{e, f@};
6536 @emph{Note:} The CPU's endianness determines which value is stored in
6537 the upper half of a register and which value is stored in the lower half.
6538 On little-endian targets, the first value is the lower one and the second
6539 value is the upper one. The opposite order applies to big-endian targets.
6540 For example, the code above will set the lower half of @code{a} to
6541 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
6544 * Paired-Single Arithmetic::
6545 * Paired-Single Built-in Functions::
6546 * MIPS-3D Built-in Functions::
6549 @node Paired-Single Arithmetic
6550 @subsubsection Paired-Single Arithmetic
6552 The table below lists the @code{v2sf} operations for which hardware
6553 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
6554 values and @code{x} is an integral value.
6556 @multitable @columnfractions .50 .50
6557 @item C code @tab MIPS instruction
6558 @item @code{a + b} @tab @code{add.ps}
6559 @item @code{a - b} @tab @code{sub.ps}
6560 @item @code{-a} @tab @code{neg.ps}
6561 @item @code{a * b} @tab @code{mul.ps}
6562 @item @code{a * b + c} @tab @code{madd.ps}
6563 @item @code{a * b - c} @tab @code{msub.ps}
6564 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
6565 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
6566 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
6569 Note that the multiply-accumulate instructions can be disabled
6570 using the command-line option @code{-mno-fused-madd}.
6572 @node Paired-Single Built-in Functions
6573 @subsubsection Paired-Single Built-in Functions
6575 The following paired-single functions map directly to a particular
6576 MIPS instruction. Please refer to the architecture specification
6577 for details on what each instruction does.
6580 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
6581 Pair lower lower (@code{pll.ps}).
6583 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
6584 Pair upper lower (@code{pul.ps}).
6586 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
6587 Pair lower upper (@code{plu.ps}).
6589 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
6590 Pair upper upper (@code{puu.ps}).
6592 @item v2sf __builtin_mips_cvt_ps_s (float, float)
6593 Convert pair to paired single (@code{cvt.ps.s}).
6595 @item float __builtin_mips_cvt_s_pl (v2sf)
6596 Convert pair lower to single (@code{cvt.s.pl}).
6598 @item float __builtin_mips_cvt_s_pu (v2sf)
6599 Convert pair upper to single (@code{cvt.s.pu}).
6601 @item v2sf __builtin_mips_abs_ps (v2sf)
6602 Absolute value (@code{abs.ps}).
6604 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
6605 Align variable (@code{alnv.ps}).
6607 @emph{Note:} The value of the third parameter must be 0 or 4
6608 modulo 8, otherwise the result will be unpredictable. Please read the
6609 instruction description for details.
6612 The following multi-instruction functions are also available.
6613 In each case, @var{cond} can be any of the 16 floating-point conditions:
6614 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
6615 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
6616 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
6619 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6620 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6621 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
6622 @code{movt.ps}/@code{movf.ps}).
6624 The @code{movt} functions return the value @var{x} computed by:
6627 c.@var{cond}.ps @var{cc},@var{a},@var{b}
6628 mov.ps @var{x},@var{c}
6629 movt.ps @var{x},@var{d},@var{cc}
6632 The @code{movf} functions are similar but use @code{movf.ps} instead
6635 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6636 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6637 Comparison of two paired-single values (@code{c.@var{cond}.ps},
6638 @code{bc1t}/@code{bc1f}).
6640 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
6641 and return either the upper or lower half of the result. For example:
6645 if (__builtin_mips_upper_c_eq_ps (a, b))
6646 upper_halves_are_equal ();
6648 upper_halves_are_unequal ();
6650 if (__builtin_mips_lower_c_eq_ps (a, b))
6651 lower_halves_are_equal ();
6653 lower_halves_are_unequal ();
6657 @node MIPS-3D Built-in Functions
6658 @subsubsection MIPS-3D Built-in Functions
6660 The MIPS-3D Application-Specific Extension (ASE) includes additional
6661 paired-single instructions that are designed to improve the performance
6662 of 3D graphics operations. Support for these instructions is controlled
6663 by the @option{-mips3d} command-line option.
6665 The functions listed below map directly to a particular MIPS-3D
6666 instruction. Please refer to the architecture specification for
6667 more details on what each instruction does.
6670 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
6671 Reduction add (@code{addr.ps}).
6673 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
6674 Reduction multiply (@code{mulr.ps}).
6676 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
6677 Convert paired single to paired word (@code{cvt.pw.ps}).
6679 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
6680 Convert paired word to paired single (@code{cvt.ps.pw}).
6682 @item float __builtin_mips_recip1_s (float)
6683 @itemx double __builtin_mips_recip1_d (double)
6684 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
6685 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
6687 @item float __builtin_mips_recip2_s (float, float)
6688 @itemx double __builtin_mips_recip2_d (double, double)
6689 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
6690 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
6692 @item float __builtin_mips_rsqrt1_s (float)
6693 @itemx double __builtin_mips_rsqrt1_d (double)
6694 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
6695 Reduced precision reciprocal square root (sequence step 1)
6696 (@code{rsqrt1.@var{fmt}}).
6698 @item float __builtin_mips_rsqrt2_s (float, float)
6699 @itemx double __builtin_mips_rsqrt2_d (double, double)
6700 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
6701 Reduced precision reciprocal square root (sequence step 2)
6702 (@code{rsqrt2.@var{fmt}}).
6705 The following multi-instruction functions are also available.
6706 In each case, @var{cond} can be any of the 16 floating-point conditions:
6707 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
6708 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
6709 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
6712 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
6713 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
6714 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
6715 @code{bc1t}/@code{bc1f}).
6717 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
6718 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
6723 if (__builtin_mips_cabs_eq_s (a, b))
6729 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6730 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6731 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
6732 @code{bc1t}/@code{bc1f}).
6734 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
6735 and return either the upper or lower half of the result. For example:
6739 if (__builtin_mips_upper_cabs_eq_ps (a, b))
6740 upper_halves_are_equal ();
6742 upper_halves_are_unequal ();
6744 if (__builtin_mips_lower_cabs_eq_ps (a, b))
6745 lower_halves_are_equal ();
6747 lower_halves_are_unequal ();
6750 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6751 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6752 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
6753 @code{movt.ps}/@code{movf.ps}).
6755 The @code{movt} functions return the value @var{x} computed by:
6758 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
6759 mov.ps @var{x},@var{c}
6760 movt.ps @var{x},@var{d},@var{cc}
6763 The @code{movf} functions are similar but use @code{movf.ps} instead
6766 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6767 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6768 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6769 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6770 Comparison of two paired-single values
6771 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
6772 @code{bc1any2t}/@code{bc1any2f}).
6774 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
6775 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
6776 result is true and the @code{all} forms return true if both results are true.
6781 if (__builtin_mips_any_c_eq_ps (a, b))
6786 if (__builtin_mips_all_c_eq_ps (a, b))
6792 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6793 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6794 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6795 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6796 Comparison of four paired-single values
6797 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
6798 @code{bc1any4t}/@code{bc1any4f}).
6800 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
6801 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
6802 The @code{any} forms return true if any of the four results are true
6803 and the @code{all} forms return true if all four results are true.
6808 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
6813 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
6820 @node PowerPC AltiVec Built-in Functions
6821 @subsection PowerPC AltiVec Built-in Functions
6823 GCC provides an interface for the PowerPC family of processors to access
6824 the AltiVec operations described in Motorola's AltiVec Programming
6825 Interface Manual. The interface is made available by including
6826 @code{<altivec.h>} and using @option{-maltivec} and
6827 @option{-mabi=altivec}. The interface supports the following vector
6831 vector unsigned char
6835 vector unsigned short
6846 GCC's implementation of the high-level language interface available from
6847 C and C++ code differs from Motorola's documentation in several ways.
6852 A vector constant is a list of constant expressions within curly braces.
6855 A vector initializer requires no cast if the vector constant is of the
6856 same type as the variable it is initializing.
6859 If @code{signed} or @code{unsigned} is omitted, the signedness of the
6860 vector type is the default signedness of the base type. The default
6861 varies depending on the operating system, so a portable program should
6862 always specify the signedness.
6865 Compiling with @option{-maltivec} adds keywords @code{__vector},
6866 @code{__pixel}, and @code{__bool}. Macros @option{vector},
6867 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
6871 GCC allows using a @code{typedef} name as the type specifier for a
6875 For C, overloaded functions are implemented with macros so the following
6879 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
6882 Since @code{vec_add} is a macro, the vector constant in the example
6883 is treated as four separate arguments. Wrap the entire argument in
6884 parentheses for this to work.
6887 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
6888 Internally, GCC uses built-in functions to achieve the functionality in
6889 the aforementioned header file, but they are not supported and are
6890 subject to change without notice.
6892 The following interfaces are supported for the generic and specific
6893 AltiVec operations and the AltiVec predicates. In cases where there
6894 is a direct mapping between generic and specific operations, only the
6895 generic names are shown here, although the specific operations can also
6898 Arguments that are documented as @code{const int} require literal
6899 integral values within the range required for that operation.
6902 vector signed char vec_abs (vector signed char);
6903 vector signed short vec_abs (vector signed short);
6904 vector signed int vec_abs (vector signed int);
6905 vector float vec_abs (vector float);
6907 vector signed char vec_abss (vector signed char);
6908 vector signed short vec_abss (vector signed short);
6909 vector signed int vec_abss (vector signed int);
6911 vector signed char vec_add (vector bool char, vector signed char);
6912 vector signed char vec_add (vector signed char, vector bool char);
6913 vector signed char vec_add (vector signed char, vector signed char);
6914 vector unsigned char vec_add (vector bool char, vector unsigned char);
6915 vector unsigned char vec_add (vector unsigned char, vector bool char);
6916 vector unsigned char vec_add (vector unsigned char,
6917 vector unsigned char);
6918 vector signed short vec_add (vector bool short, vector signed short);
6919 vector signed short vec_add (vector signed short, vector bool short);
6920 vector signed short vec_add (vector signed short, vector signed short);
6921 vector unsigned short vec_add (vector bool short,
6922 vector unsigned short);
6923 vector unsigned short vec_add (vector unsigned short,
6925 vector unsigned short vec_add (vector unsigned short,
6926 vector unsigned short);
6927 vector signed int vec_add (vector bool int, vector signed int);
6928 vector signed int vec_add (vector signed int, vector bool int);
6929 vector signed int vec_add (vector signed int, vector signed int);
6930 vector unsigned int vec_add (vector bool int, vector unsigned int);
6931 vector unsigned int vec_add (vector unsigned int, vector bool int);
6932 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
6933 vector float vec_add (vector float, vector float);
6935 vector float vec_vaddfp (vector float, vector float);
6937 vector signed int vec_vadduwm (vector bool int, vector signed int);
6938 vector signed int vec_vadduwm (vector signed int, vector bool int);
6939 vector signed int vec_vadduwm (vector signed int, vector signed int);
6940 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
6941 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
6942 vector unsigned int vec_vadduwm (vector unsigned int,
6943 vector unsigned int);
6945 vector signed short vec_vadduhm (vector bool short,
6946 vector signed short);
6947 vector signed short vec_vadduhm (vector signed short,
6949 vector signed short vec_vadduhm (vector signed short,
6950 vector signed short);
6951 vector unsigned short vec_vadduhm (vector bool short,
6952 vector unsigned short);
6953 vector unsigned short vec_vadduhm (vector unsigned short,
6955 vector unsigned short vec_vadduhm (vector unsigned short,
6956 vector unsigned short);
6958 vector signed char vec_vaddubm (vector bool char, vector signed char);
6959 vector signed char vec_vaddubm (vector signed char, vector bool char);
6960 vector signed char vec_vaddubm (vector signed char, vector signed char);
6961 vector unsigned char vec_vaddubm (vector bool char,
6962 vector unsigned char);
6963 vector unsigned char vec_vaddubm (vector unsigned char,
6965 vector unsigned char vec_vaddubm (vector unsigned char,
6966 vector unsigned char);
6968 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
6970 vector unsigned char vec_adds (vector bool char, vector unsigned char);
6971 vector unsigned char vec_adds (vector unsigned char, vector bool char);
6972 vector unsigned char vec_adds (vector unsigned char,
6973 vector unsigned char);
6974 vector signed char vec_adds (vector bool char, vector signed char);
6975 vector signed char vec_adds (vector signed char, vector bool char);
6976 vector signed char vec_adds (vector signed char, vector signed char);
6977 vector unsigned short vec_adds (vector bool short,
6978 vector unsigned short);
6979 vector unsigned short vec_adds (vector unsigned short,
6981 vector unsigned short vec_adds (vector unsigned short,
6982 vector unsigned short);
6983 vector signed short vec_adds (vector bool short, vector signed short);
6984 vector signed short vec_adds (vector signed short, vector bool short);
6985 vector signed short vec_adds (vector signed short, vector signed short);
6986 vector unsigned int vec_adds (vector bool int, vector unsigned int);
6987 vector unsigned int vec_adds (vector unsigned int, vector bool int);
6988 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
6989 vector signed int vec_adds (vector bool int, vector signed int);
6990 vector signed int vec_adds (vector signed int, vector bool int);
6991 vector signed int vec_adds (vector signed int, vector signed int);
6993 vector signed int vec_vaddsws (vector bool int, vector signed int);
6994 vector signed int vec_vaddsws (vector signed int, vector bool int);
6995 vector signed int vec_vaddsws (vector signed int, vector signed int);
6997 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
6998 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
6999 vector unsigned int vec_vadduws (vector unsigned int,
7000 vector unsigned int);
7002 vector signed short vec_vaddshs (vector bool short,
7003 vector signed short);
7004 vector signed short vec_vaddshs (vector signed short,
7006 vector signed short vec_vaddshs (vector signed short,
7007 vector signed short);
7009 vector unsigned short vec_vadduhs (vector bool short,
7010 vector unsigned short);
7011 vector unsigned short vec_vadduhs (vector unsigned short,
7013 vector unsigned short vec_vadduhs (vector unsigned short,
7014 vector unsigned short);
7016 vector signed char vec_vaddsbs (vector bool char, vector signed char);
7017 vector signed char vec_vaddsbs (vector signed char, vector bool char);
7018 vector signed char vec_vaddsbs (vector signed char, vector signed char);
7020 vector unsigned char vec_vaddubs (vector bool char,
7021 vector unsigned char);
7022 vector unsigned char vec_vaddubs (vector unsigned char,
7024 vector unsigned char vec_vaddubs (vector unsigned char,
7025 vector unsigned char);
7027 vector float vec_and (vector float, vector float);
7028 vector float vec_and (vector float, vector bool int);
7029 vector float vec_and (vector bool int, vector float);
7030 vector bool int vec_and (vector bool int, vector bool int);
7031 vector signed int vec_and (vector bool int, vector signed int);
7032 vector signed int vec_and (vector signed int, vector bool int);
7033 vector signed int vec_and (vector signed int, vector signed int);
7034 vector unsigned int vec_and (vector bool int, vector unsigned int);
7035 vector unsigned int vec_and (vector unsigned int, vector bool int);
7036 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
7037 vector bool short vec_and (vector bool short, vector bool short);
7038 vector signed short vec_and (vector bool short, vector signed short);
7039 vector signed short vec_and (vector signed short, vector bool short);
7040 vector signed short vec_and (vector signed short, vector signed short);
7041 vector unsigned short vec_and (vector bool short,
7042 vector unsigned short);
7043 vector unsigned short vec_and (vector unsigned short,
7045 vector unsigned short vec_and (vector unsigned short,
7046 vector unsigned short);
7047 vector signed char vec_and (vector bool char, vector signed char);
7048 vector bool char vec_and (vector bool char, vector bool char);
7049 vector signed char vec_and (vector signed char, vector bool char);
7050 vector signed char vec_and (vector signed char, vector signed char);
7051 vector unsigned char vec_and (vector bool char, vector unsigned char);
7052 vector unsigned char vec_and (vector unsigned char, vector bool char);
7053 vector unsigned char vec_and (vector unsigned char,
7054 vector unsigned char);
7056 vector float vec_andc (vector float, vector float);
7057 vector float vec_andc (vector float, vector bool int);
7058 vector float vec_andc (vector bool int, vector float);
7059 vector bool int vec_andc (vector bool int, vector bool int);
7060 vector signed int vec_andc (vector bool int, vector signed int);
7061 vector signed int vec_andc (vector signed int, vector bool int);
7062 vector signed int vec_andc (vector signed int, vector signed int);
7063 vector unsigned int vec_andc (vector bool int, vector unsigned int);
7064 vector unsigned int vec_andc (vector unsigned int, vector bool int);
7065 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
7066 vector bool short vec_andc (vector bool short, vector bool short);
7067 vector signed short vec_andc (vector bool short, vector signed short);
7068 vector signed short vec_andc (vector signed short, vector bool short);
7069 vector signed short vec_andc (vector signed short, vector signed short);
7070 vector unsigned short vec_andc (vector bool short,
7071 vector unsigned short);
7072 vector unsigned short vec_andc (vector unsigned short,
7074 vector unsigned short vec_andc (vector unsigned short,
7075 vector unsigned short);
7076 vector signed char vec_andc (vector bool char, vector signed char);
7077 vector bool char vec_andc (vector bool char, vector bool char);
7078 vector signed char vec_andc (vector signed char, vector bool char);
7079 vector signed char vec_andc (vector signed char, vector signed char);
7080 vector unsigned char vec_andc (vector bool char, vector unsigned char);
7081 vector unsigned char vec_andc (vector unsigned char, vector bool char);
7082 vector unsigned char vec_andc (vector unsigned char,
7083 vector unsigned char);
7085 vector unsigned char vec_avg (vector unsigned char,
7086 vector unsigned char);
7087 vector signed char vec_avg (vector signed char, vector signed char);
7088 vector unsigned short vec_avg (vector unsigned short,
7089 vector unsigned short);
7090 vector signed short vec_avg (vector signed short, vector signed short);
7091 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
7092 vector signed int vec_avg (vector signed int, vector signed int);
7094 vector signed int vec_vavgsw (vector signed int, vector signed int);
7096 vector unsigned int vec_vavguw (vector unsigned int,
7097 vector unsigned int);
7099 vector signed short vec_vavgsh (vector signed short,
7100 vector signed short);
7102 vector unsigned short vec_vavguh (vector unsigned short,
7103 vector unsigned short);
7105 vector signed char vec_vavgsb (vector signed char, vector signed char);
7107 vector unsigned char vec_vavgub (vector unsigned char,
7108 vector unsigned char);
7110 vector float vec_ceil (vector float);
7112 vector signed int vec_cmpb (vector float, vector float);
7114 vector bool char vec_cmpeq (vector signed char, vector signed char);
7115 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
7116 vector bool short vec_cmpeq (vector signed short, vector signed short);
7117 vector bool short vec_cmpeq (vector unsigned short,
7118 vector unsigned short);
7119 vector bool int vec_cmpeq (vector signed int, vector signed int);
7120 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
7121 vector bool int vec_cmpeq (vector float, vector float);
7123 vector bool int vec_vcmpeqfp (vector float, vector float);
7125 vector bool int vec_vcmpequw (vector signed int, vector signed int);
7126 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
7128 vector bool short vec_vcmpequh (vector signed short,
7129 vector signed short);
7130 vector bool short vec_vcmpequh (vector unsigned short,
7131 vector unsigned short);
7133 vector bool char vec_vcmpequb (vector signed char, vector signed char);
7134 vector bool char vec_vcmpequb (vector unsigned char,
7135 vector unsigned char);
7137 vector bool int vec_cmpge (vector float, vector float);
7139 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
7140 vector bool char vec_cmpgt (vector signed char, vector signed char);
7141 vector bool short vec_cmpgt (vector unsigned short,
7142 vector unsigned short);
7143 vector bool short vec_cmpgt (vector signed short, vector signed short);
7144 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
7145 vector bool int vec_cmpgt (vector signed int, vector signed int);
7146 vector bool int vec_cmpgt (vector float, vector float);
7148 vector bool int vec_vcmpgtfp (vector float, vector float);
7150 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
7152 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
7154 vector bool short vec_vcmpgtsh (vector signed short,
7155 vector signed short);
7157 vector bool short vec_vcmpgtuh (vector unsigned short,
7158 vector unsigned short);
7160 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
7162 vector bool char vec_vcmpgtub (vector unsigned char,
7163 vector unsigned char);
7165 vector bool int vec_cmple (vector float, vector float);
7167 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
7168 vector bool char vec_cmplt (vector signed char, vector signed char);
7169 vector bool short vec_cmplt (vector unsigned short,
7170 vector unsigned short);
7171 vector bool short vec_cmplt (vector signed short, vector signed short);
7172 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
7173 vector bool int vec_cmplt (vector signed int, vector signed int);
7174 vector bool int vec_cmplt (vector float, vector float);
7176 vector float vec_ctf (vector unsigned int, const int);
7177 vector float vec_ctf (vector signed int, const int);
7179 vector float vec_vcfsx (vector signed int, const int);
7181 vector float vec_vcfux (vector unsigned int, const int);
7183 vector signed int vec_cts (vector float, const int);
7185 vector unsigned int vec_ctu (vector float, const int);
7187 void vec_dss (const int);
7189 void vec_dssall (void);
7191 void vec_dst (const vector unsigned char *, int, const int);
7192 void vec_dst (const vector signed char *, int, const int);
7193 void vec_dst (const vector bool char *, int, const int);
7194 void vec_dst (const vector unsigned short *, int, const int);
7195 void vec_dst (const vector signed short *, int, const int);
7196 void vec_dst (const vector bool short *, int, const int);
7197 void vec_dst (const vector pixel *, int, const int);
7198 void vec_dst (const vector unsigned int *, int, const int);
7199 void vec_dst (const vector signed int *, int, const int);
7200 void vec_dst (const vector bool int *, int, const int);
7201 void vec_dst (const vector float *, int, const int);
7202 void vec_dst (const unsigned char *, int, const int);
7203 void vec_dst (const signed char *, int, const int);
7204 void vec_dst (const unsigned short *, int, const int);
7205 void vec_dst (const short *, int, const int);
7206 void vec_dst (const unsigned int *, int, const int);
7207 void vec_dst (const int *, int, const int);
7208 void vec_dst (const unsigned long *, int, const int);
7209 void vec_dst (const long *, int, const int);
7210 void vec_dst (const float *, int, const int);
7212 void vec_dstst (const vector unsigned char *, int, const int);
7213 void vec_dstst (const vector signed char *, int, const int);
7214 void vec_dstst (const vector bool char *, int, const int);
7215 void vec_dstst (const vector unsigned short *, int, const int);
7216 void vec_dstst (const vector signed short *, int, const int);
7217 void vec_dstst (const vector bool short *, int, const int);
7218 void vec_dstst (const vector pixel *, int, const int);
7219 void vec_dstst (const vector unsigned int *, int, const int);
7220 void vec_dstst (const vector signed int *, int, const int);
7221 void vec_dstst (const vector bool int *, int, const int);
7222 void vec_dstst (const vector float *, int, const int);
7223 void vec_dstst (const unsigned char *, int, const int);
7224 void vec_dstst (const signed char *, int, const int);
7225 void vec_dstst (const unsigned short *, int, const int);
7226 void vec_dstst (const short *, int, const int);
7227 void vec_dstst (const unsigned int *, int, const int);
7228 void vec_dstst (const int *, int, const int);
7229 void vec_dstst (const unsigned long *, int, const int);
7230 void vec_dstst (const long *, int, const int);
7231 void vec_dstst (const float *, int, const int);
7233 void vec_dststt (const vector unsigned char *, int, const int);
7234 void vec_dststt (const vector signed char *, int, const int);
7235 void vec_dststt (const vector bool char *, int, const int);
7236 void vec_dststt (const vector unsigned short *, int, const int);
7237 void vec_dststt (const vector signed short *, int, const int);
7238 void vec_dststt (const vector bool short *, int, const int);
7239 void vec_dststt (const vector pixel *, int, const int);
7240 void vec_dststt (const vector unsigned int *, int, const int);
7241 void vec_dststt (const vector signed int *, int, const int);
7242 void vec_dststt (const vector bool int *, int, const int);
7243 void vec_dststt (const vector float *, int, const int);
7244 void vec_dststt (const unsigned char *, int, const int);
7245 void vec_dststt (const signed char *, int, const int);
7246 void vec_dststt (const unsigned short *, int, const int);
7247 void vec_dststt (const short *, int, const int);
7248 void vec_dststt (const unsigned int *, int, const int);
7249 void vec_dststt (const int *, int, const int);
7250 void vec_dststt (const unsigned long *, int, const int);
7251 void vec_dststt (const long *, int, const int);
7252 void vec_dststt (const float *, int, const int);
7254 void vec_dstt (const vector unsigned char *, int, const int);
7255 void vec_dstt (const vector signed char *, int, const int);
7256 void vec_dstt (const vector bool char *, int, const int);
7257 void vec_dstt (const vector unsigned short *, int, const int);
7258 void vec_dstt (const vector signed short *, int, const int);
7259 void vec_dstt (const vector bool short *, int, const int);
7260 void vec_dstt (const vector pixel *, int, const int);
7261 void vec_dstt (const vector unsigned int *, int, const int);
7262 void vec_dstt (const vector signed int *, int, const int);
7263 void vec_dstt (const vector bool int *, int, const int);
7264 void vec_dstt (const vector float *, int, const int);
7265 void vec_dstt (const unsigned char *, int, const int);
7266 void vec_dstt (const signed char *, int, const int);
7267 void vec_dstt (const unsigned short *, int, const int);
7268 void vec_dstt (const short *, int, const int);
7269 void vec_dstt (const unsigned int *, int, const int);
7270 void vec_dstt (const int *, int, const int);
7271 void vec_dstt (const unsigned long *, int, const int);
7272 void vec_dstt (const long *, int, const int);
7273 void vec_dstt (const float *, int, const int);
7275 vector float vec_expte (vector float);
7277 vector float vec_floor (vector float);
7279 vector float vec_ld (int, const vector float *);
7280 vector float vec_ld (int, const float *);
7281 vector bool int vec_ld (int, const vector bool int *);
7282 vector signed int vec_ld (int, const vector signed int *);
7283 vector signed int vec_ld (int, const int *);
7284 vector signed int vec_ld (int, const long *);
7285 vector unsigned int vec_ld (int, const vector unsigned int *);
7286 vector unsigned int vec_ld (int, const unsigned int *);
7287 vector unsigned int vec_ld (int, const unsigned long *);
7288 vector bool short vec_ld (int, const vector bool short *);
7289 vector pixel vec_ld (int, const vector pixel *);
7290 vector signed short vec_ld (int, const vector signed short *);
7291 vector signed short vec_ld (int, const short *);
7292 vector unsigned short vec_ld (int, const vector unsigned short *);
7293 vector unsigned short vec_ld (int, const unsigned short *);
7294 vector bool char vec_ld (int, const vector bool char *);
7295 vector signed char vec_ld (int, const vector signed char *);
7296 vector signed char vec_ld (int, const signed char *);
7297 vector unsigned char vec_ld (int, const vector unsigned char *);
7298 vector unsigned char vec_ld (int, const unsigned char *);
7300 vector signed char vec_lde (int, const signed char *);
7301 vector unsigned char vec_lde (int, const unsigned char *);
7302 vector signed short vec_lde (int, const short *);
7303 vector unsigned short vec_lde (int, const unsigned short *);
7304 vector float vec_lde (int, const float *);
7305 vector signed int vec_lde (int, const int *);
7306 vector unsigned int vec_lde (int, const unsigned int *);
7307 vector signed int vec_lde (int, const long *);
7308 vector unsigned int vec_lde (int, const unsigned long *);
7310 vector float vec_lvewx (int, float *);
7311 vector signed int vec_lvewx (int, int *);
7312 vector unsigned int vec_lvewx (int, unsigned int *);
7313 vector signed int vec_lvewx (int, long *);
7314 vector unsigned int vec_lvewx (int, unsigned long *);
7316 vector signed short vec_lvehx (int, short *);
7317 vector unsigned short vec_lvehx (int, unsigned short *);
7319 vector signed char vec_lvebx (int, char *);
7320 vector unsigned char vec_lvebx (int, unsigned char *);
7322 vector float vec_ldl (int, const vector float *);
7323 vector float vec_ldl (int, const float *);
7324 vector bool int vec_ldl (int, const vector bool int *);
7325 vector signed int vec_ldl (int, const vector signed int *);
7326 vector signed int vec_ldl (int, const int *);
7327 vector signed int vec_ldl (int, const long *);
7328 vector unsigned int vec_ldl (int, const vector unsigned int *);
7329 vector unsigned int vec_ldl (int, const unsigned int *);
7330 vector unsigned int vec_ldl (int, const unsigned long *);
7331 vector bool short vec_ldl (int, const vector bool short *);
7332 vector pixel vec_ldl (int, const vector pixel *);
7333 vector signed short vec_ldl (int, const vector signed short *);
7334 vector signed short vec_ldl (int, const short *);
7335 vector unsigned short vec_ldl (int, const vector unsigned short *);
7336 vector unsigned short vec_ldl (int, const unsigned short *);
7337 vector bool char vec_ldl (int, const vector bool char *);
7338 vector signed char vec_ldl (int, const vector signed char *);
7339 vector signed char vec_ldl (int, const signed char *);
7340 vector unsigned char vec_ldl (int, const vector unsigned char *);
7341 vector unsigned char vec_ldl (int, const unsigned char *);
7343 vector float vec_loge (vector float);
7345 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
7346 vector unsigned char vec_lvsl (int, const volatile signed char *);
7347 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
7348 vector unsigned char vec_lvsl (int, const volatile short *);
7349 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
7350 vector unsigned char vec_lvsl (int, const volatile int *);
7351 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
7352 vector unsigned char vec_lvsl (int, const volatile long *);
7353 vector unsigned char vec_lvsl (int, const volatile float *);
7355 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
7356 vector unsigned char vec_lvsr (int, const volatile signed char *);
7357 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
7358 vector unsigned char vec_lvsr (int, const volatile short *);
7359 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
7360 vector unsigned char vec_lvsr (int, const volatile int *);
7361 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
7362 vector unsigned char vec_lvsr (int, const volatile long *);
7363 vector unsigned char vec_lvsr (int, const volatile float *);
7365 vector float vec_madd (vector float, vector float, vector float);
7367 vector signed short vec_madds (vector signed short,
7368 vector signed short,
7369 vector signed short);
7371 vector unsigned char vec_max (vector bool char, vector unsigned char);
7372 vector unsigned char vec_max (vector unsigned char, vector bool char);
7373 vector unsigned char vec_max (vector unsigned char,
7374 vector unsigned char);
7375 vector signed char vec_max (vector bool char, vector signed char);
7376 vector signed char vec_max (vector signed char, vector bool char);
7377 vector signed char vec_max (vector signed char, vector signed char);
7378 vector unsigned short vec_max (vector bool short,
7379 vector unsigned short);
7380 vector unsigned short vec_max (vector unsigned short,
7382 vector unsigned short vec_max (vector unsigned short,
7383 vector unsigned short);
7384 vector signed short vec_max (vector bool short, vector signed short);
7385 vector signed short vec_max (vector signed short, vector bool short);
7386 vector signed short vec_max (vector signed short, vector signed short);
7387 vector unsigned int vec_max (vector bool int, vector unsigned int);
7388 vector unsigned int vec_max (vector unsigned int, vector bool int);
7389 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
7390 vector signed int vec_max (vector bool int, vector signed int);
7391 vector signed int vec_max (vector signed int, vector bool int);
7392 vector signed int vec_max (vector signed int, vector signed int);
7393 vector float vec_max (vector float, vector float);
7395 vector float vec_vmaxfp (vector float, vector float);
7397 vector signed int vec_vmaxsw (vector bool int, vector signed int);
7398 vector signed int vec_vmaxsw (vector signed int, vector bool int);
7399 vector signed int vec_vmaxsw (vector signed int, vector signed int);
7401 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
7402 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
7403 vector unsigned int vec_vmaxuw (vector unsigned int,
7404 vector unsigned int);
7406 vector signed short vec_vmaxsh (vector bool short, vector signed short);
7407 vector signed short vec_vmaxsh (vector signed short, vector bool short);
7408 vector signed short vec_vmaxsh (vector signed short,
7409 vector signed short);
7411 vector unsigned short vec_vmaxuh (vector bool short,
7412 vector unsigned short);
7413 vector unsigned short vec_vmaxuh (vector unsigned short,
7415 vector unsigned short vec_vmaxuh (vector unsigned short,
7416 vector unsigned short);
7418 vector signed char vec_vmaxsb (vector bool char, vector signed char);
7419 vector signed char vec_vmaxsb (vector signed char, vector bool char);
7420 vector signed char vec_vmaxsb (vector signed char, vector signed char);
7422 vector unsigned char vec_vmaxub (vector bool char,
7423 vector unsigned char);
7424 vector unsigned char vec_vmaxub (vector unsigned char,
7426 vector unsigned char vec_vmaxub (vector unsigned char,
7427 vector unsigned char);
7429 vector bool char vec_mergeh (vector bool char, vector bool char);
7430 vector signed char vec_mergeh (vector signed char, vector signed char);
7431 vector unsigned char vec_mergeh (vector unsigned char,
7432 vector unsigned char);
7433 vector bool short vec_mergeh (vector bool short, vector bool short);
7434 vector pixel vec_mergeh (vector pixel, vector pixel);
7435 vector signed short vec_mergeh (vector signed short,
7436 vector signed short);
7437 vector unsigned short vec_mergeh (vector unsigned short,
7438 vector unsigned short);
7439 vector float vec_mergeh (vector float, vector float);
7440 vector bool int vec_mergeh (vector bool int, vector bool int);
7441 vector signed int vec_mergeh (vector signed int, vector signed int);
7442 vector unsigned int vec_mergeh (vector unsigned int,
7443 vector unsigned int);
7445 vector float vec_vmrghw (vector float, vector float);
7446 vector bool int vec_vmrghw (vector bool int, vector bool int);
7447 vector signed int vec_vmrghw (vector signed int, vector signed int);
7448 vector unsigned int vec_vmrghw (vector unsigned int,
7449 vector unsigned int);
7451 vector bool short vec_vmrghh (vector bool short, vector bool short);
7452 vector signed short vec_vmrghh (vector signed short,
7453 vector signed short);
7454 vector unsigned short vec_vmrghh (vector unsigned short,
7455 vector unsigned short);
7456 vector pixel vec_vmrghh (vector pixel, vector pixel);
7458 vector bool char vec_vmrghb (vector bool char, vector bool char);
7459 vector signed char vec_vmrghb (vector signed char, vector signed char);
7460 vector unsigned char vec_vmrghb (vector unsigned char,
7461 vector unsigned char);
7463 vector bool char vec_mergel (vector bool char, vector bool char);
7464 vector signed char vec_mergel (vector signed char, vector signed char);
7465 vector unsigned char vec_mergel (vector unsigned char,
7466 vector unsigned char);
7467 vector bool short vec_mergel (vector bool short, vector bool short);
7468 vector pixel vec_mergel (vector pixel, vector pixel);
7469 vector signed short vec_mergel (vector signed short,
7470 vector signed short);
7471 vector unsigned short vec_mergel (vector unsigned short,
7472 vector unsigned short);
7473 vector float vec_mergel (vector float, vector float);
7474 vector bool int vec_mergel (vector bool int, vector bool int);
7475 vector signed int vec_mergel (vector signed int, vector signed int);
7476 vector unsigned int vec_mergel (vector unsigned int,
7477 vector unsigned int);
7479 vector float vec_vmrglw (vector float, vector float);
7480 vector signed int vec_vmrglw (vector signed int, vector signed int);
7481 vector unsigned int vec_vmrglw (vector unsigned int,
7482 vector unsigned int);
7483 vector bool int vec_vmrglw (vector bool int, vector bool int);
7485 vector bool short vec_vmrglh (vector bool short, vector bool short);
7486 vector signed short vec_vmrglh (vector signed short,
7487 vector signed short);
7488 vector unsigned short vec_vmrglh (vector unsigned short,
7489 vector unsigned short);
7490 vector pixel vec_vmrglh (vector pixel, vector pixel);
7492 vector bool char vec_vmrglb (vector bool char, vector bool char);
7493 vector signed char vec_vmrglb (vector signed char, vector signed char);
7494 vector unsigned char vec_vmrglb (vector unsigned char,
7495 vector unsigned char);
7497 vector unsigned short vec_mfvscr (void);
7499 vector unsigned char vec_min (vector bool char, vector unsigned char);
7500 vector unsigned char vec_min (vector unsigned char, vector bool char);
7501 vector unsigned char vec_min (vector unsigned char,
7502 vector unsigned char);
7503 vector signed char vec_min (vector bool char, vector signed char);
7504 vector signed char vec_min (vector signed char, vector bool char);
7505 vector signed char vec_min (vector signed char, vector signed char);
7506 vector unsigned short vec_min (vector bool short,
7507 vector unsigned short);
7508 vector unsigned short vec_min (vector unsigned short,
7510 vector unsigned short vec_min (vector unsigned short,
7511 vector unsigned short);
7512 vector signed short vec_min (vector bool short, vector signed short);
7513 vector signed short vec_min (vector signed short, vector bool short);
7514 vector signed short vec_min (vector signed short, vector signed short);
7515 vector unsigned int vec_min (vector bool int, vector unsigned int);
7516 vector unsigned int vec_min (vector unsigned int, vector bool int);
7517 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
7518 vector signed int vec_min (vector bool int, vector signed int);
7519 vector signed int vec_min (vector signed int, vector bool int);
7520 vector signed int vec_min (vector signed int, vector signed int);
7521 vector float vec_min (vector float, vector float);
7523 vector float vec_vminfp (vector float, vector float);
7525 vector signed int vec_vminsw (vector bool int, vector signed int);
7526 vector signed int vec_vminsw (vector signed int, vector bool int);
7527 vector signed int vec_vminsw (vector signed int, vector signed int);
7529 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
7530 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
7531 vector unsigned int vec_vminuw (vector unsigned int,
7532 vector unsigned int);
7534 vector signed short vec_vminsh (vector bool short, vector signed short);
7535 vector signed short vec_vminsh (vector signed short, vector bool short);
7536 vector signed short vec_vminsh (vector signed short,
7537 vector signed short);
7539 vector unsigned short vec_vminuh (vector bool short,
7540 vector unsigned short);
7541 vector unsigned short vec_vminuh (vector unsigned short,
7543 vector unsigned short vec_vminuh (vector unsigned short,
7544 vector unsigned short);
7546 vector signed char vec_vminsb (vector bool char, vector signed char);
7547 vector signed char vec_vminsb (vector signed char, vector bool char);
7548 vector signed char vec_vminsb (vector signed char, vector signed char);
7550 vector unsigned char vec_vminub (vector bool char,
7551 vector unsigned char);
7552 vector unsigned char vec_vminub (vector unsigned char,
7554 vector unsigned char vec_vminub (vector unsigned char,
7555 vector unsigned char);
7557 vector signed short vec_mladd (vector signed short,
7558 vector signed short,
7559 vector signed short);
7560 vector signed short vec_mladd (vector signed short,
7561 vector unsigned short,
7562 vector unsigned short);
7563 vector signed short vec_mladd (vector unsigned short,
7564 vector signed short,
7565 vector signed short);
7566 vector unsigned short vec_mladd (vector unsigned short,
7567 vector unsigned short,
7568 vector unsigned short);
7570 vector signed short vec_mradds (vector signed short,
7571 vector signed short,
7572 vector signed short);
7574 vector unsigned int vec_msum (vector unsigned char,
7575 vector unsigned char,
7576 vector unsigned int);
7577 vector signed int vec_msum (vector signed char,
7578 vector unsigned char,
7580 vector unsigned int vec_msum (vector unsigned short,
7581 vector unsigned short,
7582 vector unsigned int);
7583 vector signed int vec_msum (vector signed short,
7584 vector signed short,
7587 vector signed int vec_vmsumshm (vector signed short,
7588 vector signed short,
7591 vector unsigned int vec_vmsumuhm (vector unsigned short,
7592 vector unsigned short,
7593 vector unsigned int);
7595 vector signed int vec_vmsummbm (vector signed char,
7596 vector unsigned char,
7599 vector unsigned int vec_vmsumubm (vector unsigned char,
7600 vector unsigned char,
7601 vector unsigned int);
7603 vector unsigned int vec_msums (vector unsigned short,
7604 vector unsigned short,
7605 vector unsigned int);
7606 vector signed int vec_msums (vector signed short,
7607 vector signed short,
7610 vector signed int vec_vmsumshs (vector signed short,
7611 vector signed short,
7614 vector unsigned int vec_vmsumuhs (vector unsigned short,
7615 vector unsigned short,
7616 vector unsigned int);
7618 void vec_mtvscr (vector signed int);
7619 void vec_mtvscr (vector unsigned int);
7620 void vec_mtvscr (vector bool int);
7621 void vec_mtvscr (vector signed short);
7622 void vec_mtvscr (vector unsigned short);
7623 void vec_mtvscr (vector bool short);
7624 void vec_mtvscr (vector pixel);
7625 void vec_mtvscr (vector signed char);
7626 void vec_mtvscr (vector unsigned char);
7627 void vec_mtvscr (vector bool char);
7629 vector unsigned short vec_mule (vector unsigned char,
7630 vector unsigned char);
7631 vector signed short vec_mule (vector signed char,
7632 vector signed char);
7633 vector unsigned int vec_mule (vector unsigned short,
7634 vector unsigned short);
7635 vector signed int vec_mule (vector signed short, vector signed short);
7637 vector signed int vec_vmulesh (vector signed short,
7638 vector signed short);
7640 vector unsigned int vec_vmuleuh (vector unsigned short,
7641 vector unsigned short);
7643 vector signed short vec_vmulesb (vector signed char,
7644 vector signed char);
7646 vector unsigned short vec_vmuleub (vector unsigned char,
7647 vector unsigned char);
7649 vector unsigned short vec_mulo (vector unsigned char,
7650 vector unsigned char);
7651 vector signed short vec_mulo (vector signed char, vector signed char);
7652 vector unsigned int vec_mulo (vector unsigned short,
7653 vector unsigned short);
7654 vector signed int vec_mulo (vector signed short, vector signed short);
7656 vector signed int vec_vmulosh (vector signed short,
7657 vector signed short);
7659 vector unsigned int vec_vmulouh (vector unsigned short,
7660 vector unsigned short);
7662 vector signed short vec_vmulosb (vector signed char,
7663 vector signed char);
7665 vector unsigned short vec_vmuloub (vector unsigned char,
7666 vector unsigned char);
7668 vector float vec_nmsub (vector float, vector float, vector float);
7670 vector float vec_nor (vector float, vector float);
7671 vector signed int vec_nor (vector signed int, vector signed int);
7672 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
7673 vector bool int vec_nor (vector bool int, vector bool int);
7674 vector signed short vec_nor (vector signed short, vector signed short);
7675 vector unsigned short vec_nor (vector unsigned short,
7676 vector unsigned short);
7677 vector bool short vec_nor (vector bool short, vector bool short);
7678 vector signed char vec_nor (vector signed char, vector signed char);
7679 vector unsigned char vec_nor (vector unsigned char,
7680 vector unsigned char);
7681 vector bool char vec_nor (vector bool char, vector bool char);
7683 vector float vec_or (vector float, vector float);
7684 vector float vec_or (vector float, vector bool int);
7685 vector float vec_or (vector bool int, vector float);
7686 vector bool int vec_or (vector bool int, vector bool int);
7687 vector signed int vec_or (vector bool int, vector signed int);
7688 vector signed int vec_or (vector signed int, vector bool int);
7689 vector signed int vec_or (vector signed int, vector signed int);
7690 vector unsigned int vec_or (vector bool int, vector unsigned int);
7691 vector unsigned int vec_or (vector unsigned int, vector bool int);
7692 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
7693 vector bool short vec_or (vector bool short, vector bool short);
7694 vector signed short vec_or (vector bool short, vector signed short);
7695 vector signed short vec_or (vector signed short, vector bool short);
7696 vector signed short vec_or (vector signed short, vector signed short);
7697 vector unsigned short vec_or (vector bool short, vector unsigned short);
7698 vector unsigned short vec_or (vector unsigned short, vector bool short);
7699 vector unsigned short vec_or (vector unsigned short,
7700 vector unsigned short);
7701 vector signed char vec_or (vector bool char, vector signed char);
7702 vector bool char vec_or (vector bool char, vector bool char);
7703 vector signed char vec_or (vector signed char, vector bool char);
7704 vector signed char vec_or (vector signed char, vector signed char);
7705 vector unsigned char vec_or (vector bool char, vector unsigned char);
7706 vector unsigned char vec_or (vector unsigned char, vector bool char);
7707 vector unsigned char vec_or (vector unsigned char,
7708 vector unsigned char);
7710 vector signed char vec_pack (vector signed short, vector signed short);
7711 vector unsigned char vec_pack (vector unsigned short,
7712 vector unsigned short);
7713 vector bool char vec_pack (vector bool short, vector bool short);
7714 vector signed short vec_pack (vector signed int, vector signed int);
7715 vector unsigned short vec_pack (vector unsigned int,
7716 vector unsigned int);
7717 vector bool short vec_pack (vector bool int, vector bool int);
7719 vector bool short vec_vpkuwum (vector bool int, vector bool int);
7720 vector signed short vec_vpkuwum (vector signed int, vector signed int);
7721 vector unsigned short vec_vpkuwum (vector unsigned int,
7722 vector unsigned int);
7724 vector bool char vec_vpkuhum (vector bool short, vector bool short);
7725 vector signed char vec_vpkuhum (vector signed short,
7726 vector signed short);
7727 vector unsigned char vec_vpkuhum (vector unsigned short,
7728 vector unsigned short);
7730 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
7732 vector unsigned char vec_packs (vector unsigned short,
7733 vector unsigned short);
7734 vector signed char vec_packs (vector signed short, vector signed short);
7735 vector unsigned short vec_packs (vector unsigned int,
7736 vector unsigned int);
7737 vector signed short vec_packs (vector signed int, vector signed int);
7739 vector signed short vec_vpkswss (vector signed int, vector signed int);
7741 vector unsigned short vec_vpkuwus (vector unsigned int,
7742 vector unsigned int);
7744 vector signed char vec_vpkshss (vector signed short,
7745 vector signed short);
7747 vector unsigned char vec_vpkuhus (vector unsigned short,
7748 vector unsigned short);
7750 vector unsigned char vec_packsu (vector unsigned short,
7751 vector unsigned short);
7752 vector unsigned char vec_packsu (vector signed short,
7753 vector signed short);
7754 vector unsigned short vec_packsu (vector unsigned int,
7755 vector unsigned int);
7756 vector unsigned short vec_packsu (vector signed int, vector signed int);
7758 vector unsigned short vec_vpkswus (vector signed int,
7761 vector unsigned char vec_vpkshus (vector signed short,
7762 vector signed short);
7764 vector float vec_perm (vector float,
7766 vector unsigned char);
7767 vector signed int vec_perm (vector signed int,
7769 vector unsigned char);
7770 vector unsigned int vec_perm (vector unsigned int,
7771 vector unsigned int,
7772 vector unsigned char);
7773 vector bool int vec_perm (vector bool int,
7775 vector unsigned char);
7776 vector signed short vec_perm (vector signed short,
7777 vector signed short,
7778 vector unsigned char);
7779 vector unsigned short vec_perm (vector unsigned short,
7780 vector unsigned short,
7781 vector unsigned char);
7782 vector bool short vec_perm (vector bool short,
7784 vector unsigned char);
7785 vector pixel vec_perm (vector pixel,
7787 vector unsigned char);
7788 vector signed char vec_perm (vector signed char,
7790 vector unsigned char);
7791 vector unsigned char vec_perm (vector unsigned char,
7792 vector unsigned char,
7793 vector unsigned char);
7794 vector bool char vec_perm (vector bool char,
7796 vector unsigned char);
7798 vector float vec_re (vector float);
7800 vector signed char vec_rl (vector signed char,
7801 vector unsigned char);
7802 vector unsigned char vec_rl (vector unsigned char,
7803 vector unsigned char);
7804 vector signed short vec_rl (vector signed short, vector unsigned short);
7805 vector unsigned short vec_rl (vector unsigned short,
7806 vector unsigned short);
7807 vector signed int vec_rl (vector signed int, vector unsigned int);
7808 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
7810 vector signed int vec_vrlw (vector signed int, vector unsigned int);
7811 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
7813 vector signed short vec_vrlh (vector signed short,
7814 vector unsigned short);
7815 vector unsigned short vec_vrlh (vector unsigned short,
7816 vector unsigned short);
7818 vector signed char vec_vrlb (vector signed char, vector unsigned char);
7819 vector unsigned char vec_vrlb (vector unsigned char,
7820 vector unsigned char);
7822 vector float vec_round (vector float);
7824 vector float vec_rsqrte (vector float);
7826 vector float vec_sel (vector float, vector float, vector bool int);
7827 vector float vec_sel (vector float, vector float, vector unsigned int);
7828 vector signed int vec_sel (vector signed int,
7831 vector signed int vec_sel (vector signed int,
7833 vector unsigned int);
7834 vector unsigned int vec_sel (vector unsigned int,
7835 vector unsigned int,
7837 vector unsigned int vec_sel (vector unsigned int,
7838 vector unsigned int,
7839 vector unsigned int);
7840 vector bool int vec_sel (vector bool int,
7843 vector bool int vec_sel (vector bool int,
7845 vector unsigned int);
7846 vector signed short vec_sel (vector signed short,
7847 vector signed short,
7849 vector signed short vec_sel (vector signed short,
7850 vector signed short,
7851 vector unsigned short);
7852 vector unsigned short vec_sel (vector unsigned short,
7853 vector unsigned short,
7855 vector unsigned short vec_sel (vector unsigned short,
7856 vector unsigned short,
7857 vector unsigned short);
7858 vector bool short vec_sel (vector bool short,
7861 vector bool short vec_sel (vector bool short,
7863 vector unsigned short);
7864 vector signed char vec_sel (vector signed char,
7867 vector signed char vec_sel (vector signed char,
7869 vector unsigned char);
7870 vector unsigned char vec_sel (vector unsigned char,
7871 vector unsigned char,
7873 vector unsigned char vec_sel (vector unsigned char,
7874 vector unsigned char,
7875 vector unsigned char);
7876 vector bool char vec_sel (vector bool char,
7879 vector bool char vec_sel (vector bool char,
7881 vector unsigned char);
7883 vector signed char vec_sl (vector signed char,
7884 vector unsigned char);
7885 vector unsigned char vec_sl (vector unsigned char,
7886 vector unsigned char);
7887 vector signed short vec_sl (vector signed short, vector unsigned short);
7888 vector unsigned short vec_sl (vector unsigned short,
7889 vector unsigned short);
7890 vector signed int vec_sl (vector signed int, vector unsigned int);
7891 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
7893 vector signed int vec_vslw (vector signed int, vector unsigned int);
7894 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
7896 vector signed short vec_vslh (vector signed short,
7897 vector unsigned short);
7898 vector unsigned short vec_vslh (vector unsigned short,
7899 vector unsigned short);
7901 vector signed char vec_vslb (vector signed char, vector unsigned char);
7902 vector unsigned char vec_vslb (vector unsigned char,
7903 vector unsigned char);
7905 vector float vec_sld (vector float, vector float, const int);
7906 vector signed int vec_sld (vector signed int,
7909 vector unsigned int vec_sld (vector unsigned int,
7910 vector unsigned int,
7912 vector bool int vec_sld (vector bool int,
7915 vector signed short vec_sld (vector signed short,
7916 vector signed short,
7918 vector unsigned short vec_sld (vector unsigned short,
7919 vector unsigned short,
7921 vector bool short vec_sld (vector bool short,
7924 vector pixel vec_sld (vector pixel,
7927 vector signed char vec_sld (vector signed char,
7930 vector unsigned char vec_sld (vector unsigned char,
7931 vector unsigned char,
7933 vector bool char vec_sld (vector bool char,
7937 vector signed int vec_sll (vector signed int,
7938 vector unsigned int);
7939 vector signed int vec_sll (vector signed int,
7940 vector unsigned short);
7941 vector signed int vec_sll (vector signed int,
7942 vector unsigned char);
7943 vector unsigned int vec_sll (vector unsigned int,
7944 vector unsigned int);
7945 vector unsigned int vec_sll (vector unsigned int,
7946 vector unsigned short);
7947 vector unsigned int vec_sll (vector unsigned int,
7948 vector unsigned char);
7949 vector bool int vec_sll (vector bool int,
7950 vector unsigned int);
7951 vector bool int vec_sll (vector bool int,
7952 vector unsigned short);
7953 vector bool int vec_sll (vector bool int,
7954 vector unsigned char);
7955 vector signed short vec_sll (vector signed short,
7956 vector unsigned int);
7957 vector signed short vec_sll (vector signed short,
7958 vector unsigned short);
7959 vector signed short vec_sll (vector signed short,
7960 vector unsigned char);
7961 vector unsigned short vec_sll (vector unsigned short,
7962 vector unsigned int);
7963 vector unsigned short vec_sll (vector unsigned short,
7964 vector unsigned short);
7965 vector unsigned short vec_sll (vector unsigned short,
7966 vector unsigned char);
7967 vector bool short vec_sll (vector bool short, vector unsigned int);
7968 vector bool short vec_sll (vector bool short, vector unsigned short);
7969 vector bool short vec_sll (vector bool short, vector unsigned char);
7970 vector pixel vec_sll (vector pixel, vector unsigned int);
7971 vector pixel vec_sll (vector pixel, vector unsigned short);
7972 vector pixel vec_sll (vector pixel, vector unsigned char);
7973 vector signed char vec_sll (vector signed char, vector unsigned int);
7974 vector signed char vec_sll (vector signed char, vector unsigned short);
7975 vector signed char vec_sll (vector signed char, vector unsigned char);
7976 vector unsigned char vec_sll (vector unsigned char,
7977 vector unsigned int);
7978 vector unsigned char vec_sll (vector unsigned char,
7979 vector unsigned short);
7980 vector unsigned char vec_sll (vector unsigned char,
7981 vector unsigned char);
7982 vector bool char vec_sll (vector bool char, vector unsigned int);
7983 vector bool char vec_sll (vector bool char, vector unsigned short);
7984 vector bool char vec_sll (vector bool char, vector unsigned char);
7986 vector float vec_slo (vector float, vector signed char);
7987 vector float vec_slo (vector float, vector unsigned char);
7988 vector signed int vec_slo (vector signed int, vector signed char);
7989 vector signed int vec_slo (vector signed int, vector unsigned char);
7990 vector unsigned int vec_slo (vector unsigned int, vector signed char);
7991 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
7992 vector signed short vec_slo (vector signed short, vector signed char);
7993 vector signed short vec_slo (vector signed short, vector unsigned char);
7994 vector unsigned short vec_slo (vector unsigned short,
7995 vector signed char);
7996 vector unsigned short vec_slo (vector unsigned short,
7997 vector unsigned char);
7998 vector pixel vec_slo (vector pixel, vector signed char);
7999 vector pixel vec_slo (vector pixel, vector unsigned char);
8000 vector signed char vec_slo (vector signed char, vector signed char);
8001 vector signed char vec_slo (vector signed char, vector unsigned char);
8002 vector unsigned char vec_slo (vector unsigned char, vector signed char);
8003 vector unsigned char vec_slo (vector unsigned char,
8004 vector unsigned char);
8006 vector signed char vec_splat (vector signed char, const int);
8007 vector unsigned char vec_splat (vector unsigned char, const int);
8008 vector bool char vec_splat (vector bool char, const int);
8009 vector signed short vec_splat (vector signed short, const int);
8010 vector unsigned short vec_splat (vector unsigned short, const int);
8011 vector bool short vec_splat (vector bool short, const int);
8012 vector pixel vec_splat (vector pixel, const int);
8013 vector float vec_splat (vector float, const int);
8014 vector signed int vec_splat (vector signed int, const int);
8015 vector unsigned int vec_splat (vector unsigned int, const int);
8016 vector bool int vec_splat (vector bool int, const int);
8018 vector float vec_vspltw (vector float, const int);
8019 vector signed int vec_vspltw (vector signed int, const int);
8020 vector unsigned int vec_vspltw (vector unsigned int, const int);
8021 vector bool int vec_vspltw (vector bool int, const int);
8023 vector bool short vec_vsplth (vector bool short, const int);
8024 vector signed short vec_vsplth (vector signed short, const int);
8025 vector unsigned short vec_vsplth (vector unsigned short, const int);
8026 vector pixel vec_vsplth (vector pixel, const int);
8028 vector signed char vec_vspltb (vector signed char, const int);
8029 vector unsigned char vec_vspltb (vector unsigned char, const int);
8030 vector bool char vec_vspltb (vector bool char, const int);
8032 vector signed char vec_splat_s8 (const int);
8034 vector signed short vec_splat_s16 (const int);
8036 vector signed int vec_splat_s32 (const int);
8038 vector unsigned char vec_splat_u8 (const int);
8040 vector unsigned short vec_splat_u16 (const int);
8042 vector unsigned int vec_splat_u32 (const int);
8044 vector signed char vec_sr (vector signed char, vector unsigned char);
8045 vector unsigned char vec_sr (vector unsigned char,
8046 vector unsigned char);
8047 vector signed short vec_sr (vector signed short,
8048 vector unsigned short);
8049 vector unsigned short vec_sr (vector unsigned short,
8050 vector unsigned short);
8051 vector signed int vec_sr (vector signed int, vector unsigned int);
8052 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
8054 vector signed int vec_vsrw (vector signed int, vector unsigned int);
8055 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
8057 vector signed short vec_vsrh (vector signed short,
8058 vector unsigned short);
8059 vector unsigned short vec_vsrh (vector unsigned short,
8060 vector unsigned short);
8062 vector signed char vec_vsrb (vector signed char, vector unsigned char);
8063 vector unsigned char vec_vsrb (vector unsigned char,
8064 vector unsigned char);
8066 vector signed char vec_sra (vector signed char, vector unsigned char);
8067 vector unsigned char vec_sra (vector unsigned char,
8068 vector unsigned char);
8069 vector signed short vec_sra (vector signed short,
8070 vector unsigned short);
8071 vector unsigned short vec_sra (vector unsigned short,
8072 vector unsigned short);
8073 vector signed int vec_sra (vector signed int, vector unsigned int);
8074 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
8076 vector signed int vec_vsraw (vector signed int, vector unsigned int);
8077 vector unsigned int vec_vsraw (vector unsigned int,
8078 vector unsigned int);
8080 vector signed short vec_vsrah (vector signed short,
8081 vector unsigned short);
8082 vector unsigned short vec_vsrah (vector unsigned short,
8083 vector unsigned short);
8085 vector signed char vec_vsrab (vector signed char, vector unsigned char);
8086 vector unsigned char vec_vsrab (vector unsigned char,
8087 vector unsigned char);
8089 vector signed int vec_srl (vector signed int, vector unsigned int);
8090 vector signed int vec_srl (vector signed int, vector unsigned short);
8091 vector signed int vec_srl (vector signed int, vector unsigned char);
8092 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
8093 vector unsigned int vec_srl (vector unsigned int,
8094 vector unsigned short);
8095 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
8096 vector bool int vec_srl (vector bool int, vector unsigned int);
8097 vector bool int vec_srl (vector bool int, vector unsigned short);
8098 vector bool int vec_srl (vector bool int, vector unsigned char);
8099 vector signed short vec_srl (vector signed short, vector unsigned int);
8100 vector signed short vec_srl (vector signed short,
8101 vector unsigned short);
8102 vector signed short vec_srl (vector signed short, vector unsigned char);
8103 vector unsigned short vec_srl (vector unsigned short,
8104 vector unsigned int);
8105 vector unsigned short vec_srl (vector unsigned short,
8106 vector unsigned short);
8107 vector unsigned short vec_srl (vector unsigned short,
8108 vector unsigned char);
8109 vector bool short vec_srl (vector bool short, vector unsigned int);
8110 vector bool short vec_srl (vector bool short, vector unsigned short);
8111 vector bool short vec_srl (vector bool short, vector unsigned char);
8112 vector pixel vec_srl (vector pixel, vector unsigned int);
8113 vector pixel vec_srl (vector pixel, vector unsigned short);
8114 vector pixel vec_srl (vector pixel, vector unsigned char);
8115 vector signed char vec_srl (vector signed char, vector unsigned int);
8116 vector signed char vec_srl (vector signed char, vector unsigned short);
8117 vector signed char vec_srl (vector signed char, vector unsigned char);
8118 vector unsigned char vec_srl (vector unsigned char,
8119 vector unsigned int);
8120 vector unsigned char vec_srl (vector unsigned char,
8121 vector unsigned short);
8122 vector unsigned char vec_srl (vector unsigned char,
8123 vector unsigned char);
8124 vector bool char vec_srl (vector bool char, vector unsigned int);
8125 vector bool char vec_srl (vector bool char, vector unsigned short);
8126 vector bool char vec_srl (vector bool char, vector unsigned char);
8128 vector float vec_sro (vector float, vector signed char);
8129 vector float vec_sro (vector float, vector unsigned char);
8130 vector signed int vec_sro (vector signed int, vector signed char);
8131 vector signed int vec_sro (vector signed int, vector unsigned char);
8132 vector unsigned int vec_sro (vector unsigned int, vector signed char);
8133 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
8134 vector signed short vec_sro (vector signed short, vector signed char);
8135 vector signed short vec_sro (vector signed short, vector unsigned char);
8136 vector unsigned short vec_sro (vector unsigned short,
8137 vector signed char);
8138 vector unsigned short vec_sro (vector unsigned short,
8139 vector unsigned char);
8140 vector pixel vec_sro (vector pixel, vector signed char);
8141 vector pixel vec_sro (vector pixel, vector unsigned char);
8142 vector signed char vec_sro (vector signed char, vector signed char);
8143 vector signed char vec_sro (vector signed char, vector unsigned char);
8144 vector unsigned char vec_sro (vector unsigned char, vector signed char);
8145 vector unsigned char vec_sro (vector unsigned char,
8146 vector unsigned char);
8148 void vec_st (vector float, int, vector float *);
8149 void vec_st (vector float, int, float *);
8150 void vec_st (vector signed int, int, vector signed int *);
8151 void vec_st (vector signed int, int, int *);
8152 void vec_st (vector unsigned int, int, vector unsigned int *);
8153 void vec_st (vector unsigned int, int, unsigned int *);
8154 void vec_st (vector bool int, int, vector bool int *);
8155 void vec_st (vector bool int, int, unsigned int *);
8156 void vec_st (vector bool int, int, int *);
8157 void vec_st (vector signed short, int, vector signed short *);
8158 void vec_st (vector signed short, int, short *);
8159 void vec_st (vector unsigned short, int, vector unsigned short *);
8160 void vec_st (vector unsigned short, int, unsigned short *);
8161 void vec_st (vector bool short, int, vector bool short *);
8162 void vec_st (vector bool short, int, unsigned short *);
8163 void vec_st (vector pixel, int, vector pixel *);
8164 void vec_st (vector pixel, int, unsigned short *);
8165 void vec_st (vector pixel, int, short *);
8166 void vec_st (vector bool short, int, short *);
8167 void vec_st (vector signed char, int, vector signed char *);
8168 void vec_st (vector signed char, int, signed char *);
8169 void vec_st (vector unsigned char, int, vector unsigned char *);
8170 void vec_st (vector unsigned char, int, unsigned char *);
8171 void vec_st (vector bool char, int, vector bool char *);
8172 void vec_st (vector bool char, int, unsigned char *);
8173 void vec_st (vector bool char, int, signed char *);
8175 void vec_ste (vector signed char, int, signed char *);
8176 void vec_ste (vector unsigned char, int, unsigned char *);
8177 void vec_ste (vector bool char, int, signed char *);
8178 void vec_ste (vector bool char, int, unsigned char *);
8179 void vec_ste (vector signed short, int, short *);
8180 void vec_ste (vector unsigned short, int, unsigned short *);
8181 void vec_ste (vector bool short, int, short *);
8182 void vec_ste (vector bool short, int, unsigned short *);
8183 void vec_ste (vector pixel, int, short *);
8184 void vec_ste (vector pixel, int, unsigned short *);
8185 void vec_ste (vector float, int, float *);
8186 void vec_ste (vector signed int, int, int *);
8187 void vec_ste (vector unsigned int, int, unsigned int *);
8188 void vec_ste (vector bool int, int, int *);
8189 void vec_ste (vector bool int, int, unsigned int *);
8191 void vec_stvewx (vector float, int, float *);
8192 void vec_stvewx (vector signed int, int, int *);
8193 void vec_stvewx (vector unsigned int, int, unsigned int *);
8194 void vec_stvewx (vector bool int, int, int *);
8195 void vec_stvewx (vector bool int, int, unsigned int *);
8197 void vec_stvehx (vector signed short, int, short *);
8198 void vec_stvehx (vector unsigned short, int, unsigned short *);
8199 void vec_stvehx (vector bool short, int, short *);
8200 void vec_stvehx (vector bool short, int, unsigned short *);
8201 void vec_stvehx (vector pixel, int, short *);
8202 void vec_stvehx (vector pixel, int, unsigned short *);
8204 void vec_stvebx (vector signed char, int, signed char *);
8205 void vec_stvebx (vector unsigned char, int, unsigned char *);
8206 void vec_stvebx (vector bool char, int, signed char *);
8207 void vec_stvebx (vector bool char, int, unsigned char *);
8209 void vec_stl (vector float, int, vector float *);
8210 void vec_stl (vector float, int, float *);
8211 void vec_stl (vector signed int, int, vector signed int *);
8212 void vec_stl (vector signed int, int, int *);
8213 void vec_stl (vector unsigned int, int, vector unsigned int *);
8214 void vec_stl (vector unsigned int, int, unsigned int *);
8215 void vec_stl (vector bool int, int, vector bool int *);
8216 void vec_stl (vector bool int, int, unsigned int *);
8217 void vec_stl (vector bool int, int, int *);
8218 void vec_stl (vector signed short, int, vector signed short *);
8219 void vec_stl (vector signed short, int, short *);
8220 void vec_stl (vector unsigned short, int, vector unsigned short *);
8221 void vec_stl (vector unsigned short, int, unsigned short *);
8222 void vec_stl (vector bool short, int, vector bool short *);
8223 void vec_stl (vector bool short, int, unsigned short *);
8224 void vec_stl (vector bool short, int, short *);
8225 void vec_stl (vector pixel, int, vector pixel *);
8226 void vec_stl (vector pixel, int, unsigned short *);
8227 void vec_stl (vector pixel, int, short *);
8228 void vec_stl (vector signed char, int, vector signed char *);
8229 void vec_stl (vector signed char, int, signed char *);
8230 void vec_stl (vector unsigned char, int, vector unsigned char *);
8231 void vec_stl (vector unsigned char, int, unsigned char *);
8232 void vec_stl (vector bool char, int, vector bool char *);
8233 void vec_stl (vector bool char, int, unsigned char *);
8234 void vec_stl (vector bool char, int, signed char *);
8236 vector signed char vec_sub (vector bool char, vector signed char);
8237 vector signed char vec_sub (vector signed char, vector bool char);
8238 vector signed char vec_sub (vector signed char, vector signed char);
8239 vector unsigned char vec_sub (vector bool char, vector unsigned char);
8240 vector unsigned char vec_sub (vector unsigned char, vector bool char);
8241 vector unsigned char vec_sub (vector unsigned char,
8242 vector unsigned char);
8243 vector signed short vec_sub (vector bool short, vector signed short);
8244 vector signed short vec_sub (vector signed short, vector bool short);
8245 vector signed short vec_sub (vector signed short, vector signed short);
8246 vector unsigned short vec_sub (vector bool short,
8247 vector unsigned short);
8248 vector unsigned short vec_sub (vector unsigned short,
8250 vector unsigned short vec_sub (vector unsigned short,
8251 vector unsigned short);
8252 vector signed int vec_sub (vector bool int, vector signed int);
8253 vector signed int vec_sub (vector signed int, vector bool int);
8254 vector signed int vec_sub (vector signed int, vector signed int);
8255 vector unsigned int vec_sub (vector bool int, vector unsigned int);
8256 vector unsigned int vec_sub (vector unsigned int, vector bool int);
8257 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
8258 vector float vec_sub (vector float, vector float);
8260 vector float vec_vsubfp (vector float, vector float);
8262 vector signed int vec_vsubuwm (vector bool int, vector signed int);
8263 vector signed int vec_vsubuwm (vector signed int, vector bool int);
8264 vector signed int vec_vsubuwm (vector signed int, vector signed int);
8265 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
8266 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
8267 vector unsigned int vec_vsubuwm (vector unsigned int,
8268 vector unsigned int);
8270 vector signed short vec_vsubuhm (vector bool short,
8271 vector signed short);
8272 vector signed short vec_vsubuhm (vector signed short,
8274 vector signed short vec_vsubuhm (vector signed short,
8275 vector signed short);
8276 vector unsigned short vec_vsubuhm (vector bool short,
8277 vector unsigned short);
8278 vector unsigned short vec_vsubuhm (vector unsigned short,
8280 vector unsigned short vec_vsubuhm (vector unsigned short,
8281 vector unsigned short);
8283 vector signed char vec_vsububm (vector bool char, vector signed char);
8284 vector signed char vec_vsububm (vector signed char, vector bool char);
8285 vector signed char vec_vsububm (vector signed char, vector signed char);
8286 vector unsigned char vec_vsububm (vector bool char,
8287 vector unsigned char);
8288 vector unsigned char vec_vsububm (vector unsigned char,
8290 vector unsigned char vec_vsububm (vector unsigned char,
8291 vector unsigned char);
8293 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
8295 vector unsigned char vec_subs (vector bool char, vector unsigned char);
8296 vector unsigned char vec_subs (vector unsigned char, vector bool char);
8297 vector unsigned char vec_subs (vector unsigned char,
8298 vector unsigned char);
8299 vector signed char vec_subs (vector bool char, vector signed char);
8300 vector signed char vec_subs (vector signed char, vector bool char);
8301 vector signed char vec_subs (vector signed char, vector signed char);
8302 vector unsigned short vec_subs (vector bool short,
8303 vector unsigned short);
8304 vector unsigned short vec_subs (vector unsigned short,
8306 vector unsigned short vec_subs (vector unsigned short,
8307 vector unsigned short);
8308 vector signed short vec_subs (vector bool short, vector signed short);
8309 vector signed short vec_subs (vector signed short, vector bool short);
8310 vector signed short vec_subs (vector signed short, vector signed short);
8311 vector unsigned int vec_subs (vector bool int, vector unsigned int);
8312 vector unsigned int vec_subs (vector unsigned int, vector bool int);
8313 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
8314 vector signed int vec_subs (vector bool int, vector signed int);
8315 vector signed int vec_subs (vector signed int, vector bool int);
8316 vector signed int vec_subs (vector signed int, vector signed int);
8318 vector signed int vec_vsubsws (vector bool int, vector signed int);
8319 vector signed int vec_vsubsws (vector signed int, vector bool int);
8320 vector signed int vec_vsubsws (vector signed int, vector signed int);
8322 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
8323 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
8324 vector unsigned int vec_vsubuws (vector unsigned int,
8325 vector unsigned int);
8327 vector signed short vec_vsubshs (vector bool short,
8328 vector signed short);
8329 vector signed short vec_vsubshs (vector signed short,
8331 vector signed short vec_vsubshs (vector signed short,
8332 vector signed short);
8334 vector unsigned short vec_vsubuhs (vector bool short,
8335 vector unsigned short);
8336 vector unsigned short vec_vsubuhs (vector unsigned short,
8338 vector unsigned short vec_vsubuhs (vector unsigned short,
8339 vector unsigned short);
8341 vector signed char vec_vsubsbs (vector bool char, vector signed char);
8342 vector signed char vec_vsubsbs (vector signed char, vector bool char);
8343 vector signed char vec_vsubsbs (vector signed char, vector signed char);
8345 vector unsigned char vec_vsububs (vector bool char,
8346 vector unsigned char);
8347 vector unsigned char vec_vsububs (vector unsigned char,
8349 vector unsigned char vec_vsububs (vector unsigned char,
8350 vector unsigned char);
8352 vector unsigned int vec_sum4s (vector unsigned char,
8353 vector unsigned int);
8354 vector signed int vec_sum4s (vector signed char, vector signed int);
8355 vector signed int vec_sum4s (vector signed short, vector signed int);
8357 vector signed int vec_vsum4shs (vector signed short, vector signed int);
8359 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
8361 vector unsigned int vec_vsum4ubs (vector unsigned char,
8362 vector unsigned int);
8364 vector signed int vec_sum2s (vector signed int, vector signed int);
8366 vector signed int vec_sums (vector signed int, vector signed int);
8368 vector float vec_trunc (vector float);
8370 vector signed short vec_unpackh (vector signed char);
8371 vector bool short vec_unpackh (vector bool char);
8372 vector signed int vec_unpackh (vector signed short);
8373 vector bool int vec_unpackh (vector bool short);
8374 vector unsigned int vec_unpackh (vector pixel);
8376 vector bool int vec_vupkhsh (vector bool short);
8377 vector signed int vec_vupkhsh (vector signed short);
8379 vector unsigned int vec_vupkhpx (vector pixel);
8381 vector bool short vec_vupkhsb (vector bool char);
8382 vector signed short vec_vupkhsb (vector signed char);
8384 vector signed short vec_unpackl (vector signed char);
8385 vector bool short vec_unpackl (vector bool char);
8386 vector unsigned int vec_unpackl (vector pixel);
8387 vector signed int vec_unpackl (vector signed short);
8388 vector bool int vec_unpackl (vector bool short);
8390 vector unsigned int vec_vupklpx (vector pixel);
8392 vector bool int vec_vupklsh (vector bool short);
8393 vector signed int vec_vupklsh (vector signed short);
8395 vector bool short vec_vupklsb (vector bool char);
8396 vector signed short vec_vupklsb (vector signed char);
8398 vector float vec_xor (vector float, vector float);
8399 vector float vec_xor (vector float, vector bool int);
8400 vector float vec_xor (vector bool int, vector float);
8401 vector bool int vec_xor (vector bool int, vector bool int);
8402 vector signed int vec_xor (vector bool int, vector signed int);
8403 vector signed int vec_xor (vector signed int, vector bool int);
8404 vector signed int vec_xor (vector signed int, vector signed int);
8405 vector unsigned int vec_xor (vector bool int, vector unsigned int);
8406 vector unsigned int vec_xor (vector unsigned int, vector bool int);
8407 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
8408 vector bool short vec_xor (vector bool short, vector bool short);
8409 vector signed short vec_xor (vector bool short, vector signed short);
8410 vector signed short vec_xor (vector signed short, vector bool short);
8411 vector signed short vec_xor (vector signed short, vector signed short);
8412 vector unsigned short vec_xor (vector bool short,
8413 vector unsigned short);
8414 vector unsigned short vec_xor (vector unsigned short,
8416 vector unsigned short vec_xor (vector unsigned short,
8417 vector unsigned short);
8418 vector signed char vec_xor (vector bool char, vector signed char);
8419 vector bool char vec_xor (vector bool char, vector bool char);
8420 vector signed char vec_xor (vector signed char, vector bool char);
8421 vector signed char vec_xor (vector signed char, vector signed char);
8422 vector unsigned char vec_xor (vector bool char, vector unsigned char);
8423 vector unsigned char vec_xor (vector unsigned char, vector bool char);
8424 vector unsigned char vec_xor (vector unsigned char,
8425 vector unsigned char);
8427 int vec_all_eq (vector signed char, vector bool char);
8428 int vec_all_eq (vector signed char, vector signed char);
8429 int vec_all_eq (vector unsigned char, vector bool char);
8430 int vec_all_eq (vector unsigned char, vector unsigned char);
8431 int vec_all_eq (vector bool char, vector bool char);
8432 int vec_all_eq (vector bool char, vector unsigned char);
8433 int vec_all_eq (vector bool char, vector signed char);
8434 int vec_all_eq (vector signed short, vector bool short);
8435 int vec_all_eq (vector signed short, vector signed short);
8436 int vec_all_eq (vector unsigned short, vector bool short);
8437 int vec_all_eq (vector unsigned short, vector unsigned short);
8438 int vec_all_eq (vector bool short, vector bool short);
8439 int vec_all_eq (vector bool short, vector unsigned short);
8440 int vec_all_eq (vector bool short, vector signed short);
8441 int vec_all_eq (vector pixel, vector pixel);
8442 int vec_all_eq (vector signed int, vector bool int);
8443 int vec_all_eq (vector signed int, vector signed int);
8444 int vec_all_eq (vector unsigned int, vector bool int);
8445 int vec_all_eq (vector unsigned int, vector unsigned int);
8446 int vec_all_eq (vector bool int, vector bool int);
8447 int vec_all_eq (vector bool int, vector unsigned int);
8448 int vec_all_eq (vector bool int, vector signed int);
8449 int vec_all_eq (vector float, vector float);
8451 int vec_all_ge (vector bool char, vector unsigned char);
8452 int vec_all_ge (vector unsigned char, vector bool char);
8453 int vec_all_ge (vector unsigned char, vector unsigned char);
8454 int vec_all_ge (vector bool char, vector signed char);
8455 int vec_all_ge (vector signed char, vector bool char);
8456 int vec_all_ge (vector signed char, vector signed char);
8457 int vec_all_ge (vector bool short, vector unsigned short);
8458 int vec_all_ge (vector unsigned short, vector bool short);
8459 int vec_all_ge (vector unsigned short, vector unsigned short);
8460 int vec_all_ge (vector signed short, vector signed short);
8461 int vec_all_ge (vector bool short, vector signed short);
8462 int vec_all_ge (vector signed short, vector bool short);
8463 int vec_all_ge (vector bool int, vector unsigned int);
8464 int vec_all_ge (vector unsigned int, vector bool int);
8465 int vec_all_ge (vector unsigned int, vector unsigned int);
8466 int vec_all_ge (vector bool int, vector signed int);
8467 int vec_all_ge (vector signed int, vector bool int);
8468 int vec_all_ge (vector signed int, vector signed int);
8469 int vec_all_ge (vector float, vector float);
8471 int vec_all_gt (vector bool char, vector unsigned char);
8472 int vec_all_gt (vector unsigned char, vector bool char);
8473 int vec_all_gt (vector unsigned char, vector unsigned char);
8474 int vec_all_gt (vector bool char, vector signed char);
8475 int vec_all_gt (vector signed char, vector bool char);
8476 int vec_all_gt (vector signed char, vector signed char);
8477 int vec_all_gt (vector bool short, vector unsigned short);
8478 int vec_all_gt (vector unsigned short, vector bool short);
8479 int vec_all_gt (vector unsigned short, vector unsigned short);
8480 int vec_all_gt (vector bool short, vector signed short);
8481 int vec_all_gt (vector signed short, vector bool short);
8482 int vec_all_gt (vector signed short, vector signed short);
8483 int vec_all_gt (vector bool int, vector unsigned int);
8484 int vec_all_gt (vector unsigned int, vector bool int);
8485 int vec_all_gt (vector unsigned int, vector unsigned int);
8486 int vec_all_gt (vector bool int, vector signed int);
8487 int vec_all_gt (vector signed int, vector bool int);
8488 int vec_all_gt (vector signed int, vector signed int);
8489 int vec_all_gt (vector float, vector float);
8491 int vec_all_in (vector float, vector float);
8493 int vec_all_le (vector bool char, vector unsigned char);
8494 int vec_all_le (vector unsigned char, vector bool char);
8495 int vec_all_le (vector unsigned char, vector unsigned char);
8496 int vec_all_le (vector bool char, vector signed char);
8497 int vec_all_le (vector signed char, vector bool char);
8498 int vec_all_le (vector signed char, vector signed char);
8499 int vec_all_le (vector bool short, vector unsigned short);
8500 int vec_all_le (vector unsigned short, vector bool short);
8501 int vec_all_le (vector unsigned short, vector unsigned short);
8502 int vec_all_le (vector bool short, vector signed short);
8503 int vec_all_le (vector signed short, vector bool short);
8504 int vec_all_le (vector signed short, vector signed short);
8505 int vec_all_le (vector bool int, vector unsigned int);
8506 int vec_all_le (vector unsigned int, vector bool int);
8507 int vec_all_le (vector unsigned int, vector unsigned int);
8508 int vec_all_le (vector bool int, vector signed int);
8509 int vec_all_le (vector signed int, vector bool int);
8510 int vec_all_le (vector signed int, vector signed int);
8511 int vec_all_le (vector float, vector float);
8513 int vec_all_lt (vector bool char, vector unsigned char);
8514 int vec_all_lt (vector unsigned char, vector bool char);
8515 int vec_all_lt (vector unsigned char, vector unsigned char);
8516 int vec_all_lt (vector bool char, vector signed char);
8517 int vec_all_lt (vector signed char, vector bool char);
8518 int vec_all_lt (vector signed char, vector signed char);
8519 int vec_all_lt (vector bool short, vector unsigned short);
8520 int vec_all_lt (vector unsigned short, vector bool short);
8521 int vec_all_lt (vector unsigned short, vector unsigned short);
8522 int vec_all_lt (vector bool short, vector signed short);
8523 int vec_all_lt (vector signed short, vector bool short);
8524 int vec_all_lt (vector signed short, vector signed short);
8525 int vec_all_lt (vector bool int, vector unsigned int);
8526 int vec_all_lt (vector unsigned int, vector bool int);
8527 int vec_all_lt (vector unsigned int, vector unsigned int);
8528 int vec_all_lt (vector bool int, vector signed int);
8529 int vec_all_lt (vector signed int, vector bool int);
8530 int vec_all_lt (vector signed int, vector signed int);
8531 int vec_all_lt (vector float, vector float);
8533 int vec_all_nan (vector float);
8535 int vec_all_ne (vector signed char, vector bool char);
8536 int vec_all_ne (vector signed char, vector signed char);
8537 int vec_all_ne (vector unsigned char, vector bool char);
8538 int vec_all_ne (vector unsigned char, vector unsigned char);
8539 int vec_all_ne (vector bool char, vector bool char);
8540 int vec_all_ne (vector bool char, vector unsigned char);
8541 int vec_all_ne (vector bool char, vector signed char);
8542 int vec_all_ne (vector signed short, vector bool short);
8543 int vec_all_ne (vector signed short, vector signed short);
8544 int vec_all_ne (vector unsigned short, vector bool short);
8545 int vec_all_ne (vector unsigned short, vector unsigned short);
8546 int vec_all_ne (vector bool short, vector bool short);
8547 int vec_all_ne (vector bool short, vector unsigned short);
8548 int vec_all_ne (vector bool short, vector signed short);
8549 int vec_all_ne (vector pixel, vector pixel);
8550 int vec_all_ne (vector signed int, vector bool int);
8551 int vec_all_ne (vector signed int, vector signed int);
8552 int vec_all_ne (vector unsigned int, vector bool int);
8553 int vec_all_ne (vector unsigned int, vector unsigned int);
8554 int vec_all_ne (vector bool int, vector bool int);
8555 int vec_all_ne (vector bool int, vector unsigned int);
8556 int vec_all_ne (vector bool int, vector signed int);
8557 int vec_all_ne (vector float, vector float);
8559 int vec_all_nge (vector float, vector float);
8561 int vec_all_ngt (vector float, vector float);
8563 int vec_all_nle (vector float, vector float);
8565 int vec_all_nlt (vector float, vector float);
8567 int vec_all_numeric (vector float);
8569 int vec_any_eq (vector signed char, vector bool char);
8570 int vec_any_eq (vector signed char, vector signed char);
8571 int vec_any_eq (vector unsigned char, vector bool char);
8572 int vec_any_eq (vector unsigned char, vector unsigned char);
8573 int vec_any_eq (vector bool char, vector bool char);
8574 int vec_any_eq (vector bool char, vector unsigned char);
8575 int vec_any_eq (vector bool char, vector signed char);
8576 int vec_any_eq (vector signed short, vector bool short);
8577 int vec_any_eq (vector signed short, vector signed short);
8578 int vec_any_eq (vector unsigned short, vector bool short);
8579 int vec_any_eq (vector unsigned short, vector unsigned short);
8580 int vec_any_eq (vector bool short, vector bool short);
8581 int vec_any_eq (vector bool short, vector unsigned short);
8582 int vec_any_eq (vector bool short, vector signed short);
8583 int vec_any_eq (vector pixel, vector pixel);
8584 int vec_any_eq (vector signed int, vector bool int);
8585 int vec_any_eq (vector signed int, vector signed int);
8586 int vec_any_eq (vector unsigned int, vector bool int);
8587 int vec_any_eq (vector unsigned int, vector unsigned int);
8588 int vec_any_eq (vector bool int, vector bool int);
8589 int vec_any_eq (vector bool int, vector unsigned int);
8590 int vec_any_eq (vector bool int, vector signed int);
8591 int vec_any_eq (vector float, vector float);
8593 int vec_any_ge (vector signed char, vector bool char);
8594 int vec_any_ge (vector unsigned char, vector bool char);
8595 int vec_any_ge (vector unsigned char, vector unsigned char);
8596 int vec_any_ge (vector signed char, vector signed char);
8597 int vec_any_ge (vector bool char, vector unsigned char);
8598 int vec_any_ge (vector bool char, vector signed char);
8599 int vec_any_ge (vector unsigned short, vector bool short);
8600 int vec_any_ge (vector unsigned short, vector unsigned short);
8601 int vec_any_ge (vector signed short, vector signed short);
8602 int vec_any_ge (vector signed short, vector bool short);
8603 int vec_any_ge (vector bool short, vector unsigned short);
8604 int vec_any_ge (vector bool short, vector signed short);
8605 int vec_any_ge (vector signed int, vector bool int);
8606 int vec_any_ge (vector unsigned int, vector bool int);
8607 int vec_any_ge (vector unsigned int, vector unsigned int);
8608 int vec_any_ge (vector signed int, vector signed int);
8609 int vec_any_ge (vector bool int, vector unsigned int);
8610 int vec_any_ge (vector bool int, vector signed int);
8611 int vec_any_ge (vector float, vector float);
8613 int vec_any_gt (vector bool char, vector unsigned char);
8614 int vec_any_gt (vector unsigned char, vector bool char);
8615 int vec_any_gt (vector unsigned char, vector unsigned char);
8616 int vec_any_gt (vector bool char, vector signed char);
8617 int vec_any_gt (vector signed char, vector bool char);
8618 int vec_any_gt (vector signed char, vector signed char);
8619 int vec_any_gt (vector bool short, vector unsigned short);
8620 int vec_any_gt (vector unsigned short, vector bool short);
8621 int vec_any_gt (vector unsigned short, vector unsigned short);
8622 int vec_any_gt (vector bool short, vector signed short);
8623 int vec_any_gt (vector signed short, vector bool short);
8624 int vec_any_gt (vector signed short, vector signed short);
8625 int vec_any_gt (vector bool int, vector unsigned int);
8626 int vec_any_gt (vector unsigned int, vector bool int);
8627 int vec_any_gt (vector unsigned int, vector unsigned int);
8628 int vec_any_gt (vector bool int, vector signed int);
8629 int vec_any_gt (vector signed int, vector bool int);
8630 int vec_any_gt (vector signed int, vector signed int);
8631 int vec_any_gt (vector float, vector float);
8633 int vec_any_le (vector bool char, vector unsigned char);
8634 int vec_any_le (vector unsigned char, vector bool char);
8635 int vec_any_le (vector unsigned char, vector unsigned char);
8636 int vec_any_le (vector bool char, vector signed char);
8637 int vec_any_le (vector signed char, vector bool char);
8638 int vec_any_le (vector signed char, vector signed char);
8639 int vec_any_le (vector bool short, vector unsigned short);
8640 int vec_any_le (vector unsigned short, vector bool short);
8641 int vec_any_le (vector unsigned short, vector unsigned short);
8642 int vec_any_le (vector bool short, vector signed short);
8643 int vec_any_le (vector signed short, vector bool short);
8644 int vec_any_le (vector signed short, vector signed short);
8645 int vec_any_le (vector bool int, vector unsigned int);
8646 int vec_any_le (vector unsigned int, vector bool int);
8647 int vec_any_le (vector unsigned int, vector unsigned int);
8648 int vec_any_le (vector bool int, vector signed int);
8649 int vec_any_le (vector signed int, vector bool int);
8650 int vec_any_le (vector signed int, vector signed int);
8651 int vec_any_le (vector float, vector float);
8653 int vec_any_lt (vector bool char, vector unsigned char);
8654 int vec_any_lt (vector unsigned char, vector bool char);
8655 int vec_any_lt (vector unsigned char, vector unsigned char);
8656 int vec_any_lt (vector bool char, vector signed char);
8657 int vec_any_lt (vector signed char, vector bool char);
8658 int vec_any_lt (vector signed char, vector signed char);
8659 int vec_any_lt (vector bool short, vector unsigned short);
8660 int vec_any_lt (vector unsigned short, vector bool short);
8661 int vec_any_lt (vector unsigned short, vector unsigned short);
8662 int vec_any_lt (vector bool short, vector signed short);
8663 int vec_any_lt (vector signed short, vector bool short);
8664 int vec_any_lt (vector signed short, vector signed short);
8665 int vec_any_lt (vector bool int, vector unsigned int);
8666 int vec_any_lt (vector unsigned int, vector bool int);
8667 int vec_any_lt (vector unsigned int, vector unsigned int);
8668 int vec_any_lt (vector bool int, vector signed int);
8669 int vec_any_lt (vector signed int, vector bool int);
8670 int vec_any_lt (vector signed int, vector signed int);
8671 int vec_any_lt (vector float, vector float);
8673 int vec_any_nan (vector float);
8675 int vec_any_ne (vector signed char, vector bool char);
8676 int vec_any_ne (vector signed char, vector signed char);
8677 int vec_any_ne (vector unsigned char, vector bool char);
8678 int vec_any_ne (vector unsigned char, vector unsigned char);
8679 int vec_any_ne (vector bool char, vector bool char);
8680 int vec_any_ne (vector bool char, vector unsigned char);
8681 int vec_any_ne (vector bool char, vector signed char);
8682 int vec_any_ne (vector signed short, vector bool short);
8683 int vec_any_ne (vector signed short, vector signed short);
8684 int vec_any_ne (vector unsigned short, vector bool short);
8685 int vec_any_ne (vector unsigned short, vector unsigned short);
8686 int vec_any_ne (vector bool short, vector bool short);
8687 int vec_any_ne (vector bool short, vector unsigned short);
8688 int vec_any_ne (vector bool short, vector signed short);
8689 int vec_any_ne (vector pixel, vector pixel);
8690 int vec_any_ne (vector signed int, vector bool int);
8691 int vec_any_ne (vector signed int, vector signed int);
8692 int vec_any_ne (vector unsigned int, vector bool int);
8693 int vec_any_ne (vector unsigned int, vector unsigned int);
8694 int vec_any_ne (vector bool int, vector bool int);
8695 int vec_any_ne (vector bool int, vector unsigned int);
8696 int vec_any_ne (vector bool int, vector signed int);
8697 int vec_any_ne (vector float, vector float);
8699 int vec_any_nge (vector float, vector float);
8701 int vec_any_ngt (vector float, vector float);
8703 int vec_any_nle (vector float, vector float);
8705 int vec_any_nlt (vector float, vector float);
8707 int vec_any_numeric (vector float);
8709 int vec_any_out (vector float, vector float);
8712 @node SPARC VIS Built-in Functions
8713 @subsection SPARC VIS Built-in Functions
8715 GCC supports SIMD operations on the SPARC using both the generic vector
8716 extensions (@pxref{Vector Extensions}) as well as built-in functions for
8717 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
8718 switch, the VIS extension is exposed as the following built-in functions:
8721 typedef int v2si __attribute__ ((vector_size (8)));
8722 typedef short v4hi __attribute__ ((vector_size (8)));
8723 typedef short v2hi __attribute__ ((vector_size (4)));
8724 typedef char v8qi __attribute__ ((vector_size (8)));
8725 typedef char v4qi __attribute__ ((vector_size (4)));
8727 void * __builtin_vis_alignaddr (void *, long);
8728 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
8729 v2si __builtin_vis_faligndatav2si (v2si, v2si);
8730 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
8731 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
8733 v4hi __builtin_vis_fexpand (v4qi);
8735 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
8736 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
8737 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
8738 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
8739 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
8740 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
8741 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
8743 v4qi __builtin_vis_fpack16 (v4hi);
8744 v8qi __builtin_vis_fpack32 (v2si, v2si);
8745 v2hi __builtin_vis_fpackfix (v2si);
8746 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
8748 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
8751 @node Target Format Checks
8752 @section Format Checks Specific to Particular Target Machines
8754 For some target machines, GCC supports additional options to the
8756 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
8759 * Solaris Format Checks::
8762 @node Solaris Format Checks
8763 @subsection Solaris Format Checks
8765 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
8766 check. @code{cmn_err} accepts a subset of the standard @code{printf}
8767 conversions, and the two-argument @code{%b} conversion for displaying
8768 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
8771 @section Pragmas Accepted by GCC
8775 GCC supports several types of pragmas, primarily in order to compile
8776 code originally written for other compilers. Note that in general
8777 we do not recommend the use of pragmas; @xref{Function Attributes},
8778 for further explanation.
8782 * RS/6000 and PowerPC Pragmas::
8785 * Symbol-Renaming Pragmas::
8786 * Structure-Packing Pragmas::
8791 @subsection ARM Pragmas
8793 The ARM target defines pragmas for controlling the default addition of
8794 @code{long_call} and @code{short_call} attributes to functions.
8795 @xref{Function Attributes}, for information about the effects of these
8800 @cindex pragma, long_calls
8801 Set all subsequent functions to have the @code{long_call} attribute.
8804 @cindex pragma, no_long_calls
8805 Set all subsequent functions to have the @code{short_call} attribute.
8807 @item long_calls_off
8808 @cindex pragma, long_calls_off
8809 Do not affect the @code{long_call} or @code{short_call} attributes of
8810 subsequent functions.
8813 @node RS/6000 and PowerPC Pragmas
8814 @subsection RS/6000 and PowerPC Pragmas
8816 The RS/6000 and PowerPC targets define one pragma for controlling
8817 whether or not the @code{longcall} attribute is added to function
8818 declarations by default. This pragma overrides the @option{-mlongcall}
8819 option, but not the @code{longcall} and @code{shortcall} attributes.
8820 @xref{RS/6000 and PowerPC Options}, for more information about when long
8821 calls are and are not necessary.
8825 @cindex pragma, longcall
8826 Apply the @code{longcall} attribute to all subsequent function
8830 Do not apply the @code{longcall} attribute to subsequent function
8834 @c Describe c4x pragmas here.
8835 @c Describe h8300 pragmas here.
8836 @c Describe sh pragmas here.
8837 @c Describe v850 pragmas here.
8839 @node Darwin Pragmas
8840 @subsection Darwin Pragmas
8842 The following pragmas are available for all architectures running the
8843 Darwin operating system. These are useful for compatibility with other
8847 @item mark @var{tokens}@dots{}
8848 @cindex pragma, mark
8849 This pragma is accepted, but has no effect.
8851 @item options align=@var{alignment}
8852 @cindex pragma, options align
8853 This pragma sets the alignment of fields in structures. The values of
8854 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
8855 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
8856 properly; to restore the previous setting, use @code{reset} for the
8859 @item segment @var{tokens}@dots{}
8860 @cindex pragma, segment
8861 This pragma is accepted, but has no effect.
8863 @item unused (@var{var} [, @var{var}]@dots{})
8864 @cindex pragma, unused
8865 This pragma declares variables to be possibly unused. GCC will not
8866 produce warnings for the listed variables. The effect is similar to
8867 that of the @code{unused} attribute, except that this pragma may appear
8868 anywhere within the variables' scopes.
8871 @node Solaris Pragmas
8872 @subsection Solaris Pragmas
8874 The Solaris target supports @code{#pragma redefine_extname}
8875 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
8876 @code{#pragma} directives for compatibility with the system compiler.
8879 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
8880 @cindex pragma, align
8882 Increase the minimum alignment of each @var{variable} to @var{alignment}.
8883 This is the same as GCC's @code{aligned} attribute @pxref{Variable
8884 Attributes}). Macro expansion occurs on the arguments to this pragma
8885 when compiling C and Objective-C. It does not currently occur when
8886 compiling C++, but this is a bug which may be fixed in a future
8889 @item fini (@var{function} [, @var{function}]...)
8890 @cindex pragma, fini
8892 This pragma causes each listed @var{function} to be called after
8893 main, or during shared module unloading, by adding a call to the
8894 @code{.fini} section.
8896 @item init (@var{function} [, @var{function}]...)
8897 @cindex pragma, init
8899 This pragma causes each listed @var{function} to be called during
8900 initialization (before @code{main}) or during shared module loading, by
8901 adding a call to the @code{.init} section.
8905 @node Symbol-Renaming Pragmas
8906 @subsection Symbol-Renaming Pragmas
8908 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
8909 supports two @code{#pragma} directives which change the name used in
8910 assembly for a given declaration. These pragmas are only available on
8911 platforms whose system headers need them. To get this effect on all
8912 platforms supported by GCC, use the asm labels extension (@pxref{Asm
8916 @item redefine_extname @var{oldname} @var{newname}
8917 @cindex pragma, redefine_extname
8919 This pragma gives the C function @var{oldname} the assembly symbol
8920 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
8921 will be defined if this pragma is available (currently only on
8924 @item extern_prefix @var{string}
8925 @cindex pragma, extern_prefix
8927 This pragma causes all subsequent external function and variable
8928 declarations to have @var{string} prepended to their assembly symbols.
8929 This effect may be terminated with another @code{extern_prefix} pragma
8930 whose argument is an empty string. The preprocessor macro
8931 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
8932 available (currently only on Tru64 UNIX)@.
8935 These pragmas and the asm labels extension interact in a complicated
8936 manner. Here are some corner cases you may want to be aware of.
8939 @item Both pragmas silently apply only to declarations with external
8940 linkage. Asm labels do not have this restriction.
8942 @item In C++, both pragmas silently apply only to declarations with
8943 ``C'' linkage. Again, asm labels do not have this restriction.
8945 @item If any of the three ways of changing the assembly name of a
8946 declaration is applied to a declaration whose assembly name has
8947 already been determined (either by a previous use of one of these
8948 features, or because the compiler needed the assembly name in order to
8949 generate code), and the new name is different, a warning issues and
8950 the name does not change.
8952 @item The @var{oldname} used by @code{#pragma redefine_extname} is
8953 always the C-language name.
8955 @item If @code{#pragma extern_prefix} is in effect, and a declaration
8956 occurs with an asm label attached, the prefix is silently ignored for
8959 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
8960 apply to the same declaration, whichever triggered first wins, and a
8961 warning issues if they contradict each other. (We would like to have
8962 @code{#pragma redefine_extname} always win, for consistency with asm
8963 labels, but if @code{#pragma extern_prefix} triggers first we have no
8964 way of knowing that that happened.)
8967 @node Structure-Packing Pragmas
8968 @subsection Structure-Packing Pragmas
8970 For compatibility with Win32, GCC supports a set of @code{#pragma}
8971 directives which change the maximum alignment of members of structures,
8972 unions, and classes subsequently defined. The @var{n} value below always
8973 is required to be a small power of two and specifies the new alignment
8977 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
8978 @item @code{#pragma pack()} sets the alignment to the one that was in
8979 effect when compilation started (see also command line option
8980 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
8981 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
8982 setting on an internal stack and then optionally sets the new alignment.
8983 @item @code{#pragma pack(pop)} restores the alignment setting to the one
8984 saved at the top of the internal stack (and removes that stack entry).
8985 Note that @code{#pragma pack([@var{n}])} does not influence this internal
8986 stack; thus it is possible to have @code{#pragma pack(push)} followed by
8987 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
8988 @code{#pragma pack(pop)}.
8992 @subsection Weak Pragmas
8994 For compatibility with SVR4, GCC supports a set of @code{#pragma}
8995 directives for declaring symbols to be weak, and defining weak
8999 @item #pragma weak @var{symbol}
9000 @cindex pragma, weak
9001 This pragma declares @var{symbol} to be weak, as if the declaration
9002 had the attribute of the same name. The pragma may appear before
9003 or after the declaration of @var{symbol}, but must appear before
9004 either its first use or its definition. It is not an error for
9005 @var{symbol} to never be defined at all.
9007 @item #pragma weak @var{symbol1} = @var{symbol2}
9008 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
9009 It is an error if @var{symbol2} is not defined in the current
9013 @node Unnamed Fields
9014 @section Unnamed struct/union fields within structs/unions
9018 For compatibility with other compilers, GCC allows you to define
9019 a structure or union that contains, as fields, structures and unions
9020 without names. For example:
9033 In this example, the user would be able to access members of the unnamed
9034 union with code like @samp{foo.b}. Note that only unnamed structs and
9035 unions are allowed, you may not have, for example, an unnamed
9038 You must never create such structures that cause ambiguous field definitions.
9039 For example, this structure:
9050 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
9051 Such constructs are not supported and must be avoided. In the future,
9052 such constructs may be detected and treated as compilation errors.
9054 @opindex fms-extensions
9055 Unless @option{-fms-extensions} is used, the unnamed field must be a
9056 structure or union definition without a tag (for example, @samp{struct
9057 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
9058 also be a definition with a tag such as @samp{struct foo @{ int a;
9059 @};}, a reference to a previously defined structure or union such as
9060 @samp{struct foo;}, or a reference to a @code{typedef} name for a
9061 previously defined structure or union type.
9064 @section Thread-Local Storage
9065 @cindex Thread-Local Storage
9066 @cindex @acronym{TLS}
9069 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
9070 are allocated such that there is one instance of the variable per extant
9071 thread. The run-time model GCC uses to implement this originates
9072 in the IA-64 processor-specific ABI, but has since been migrated
9073 to other processors as well. It requires significant support from
9074 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
9075 system libraries (@file{libc.so} and @file{libpthread.so}), so it
9076 is not available everywhere.
9078 At the user level, the extension is visible with a new storage
9079 class keyword: @code{__thread}. For example:
9083 extern __thread struct state s;
9084 static __thread char *p;
9087 The @code{__thread} specifier may be used alone, with the @code{extern}
9088 or @code{static} specifiers, but with no other storage class specifier.
9089 When used with @code{extern} or @code{static}, @code{__thread} must appear
9090 immediately after the other storage class specifier.
9092 The @code{__thread} specifier may be applied to any global, file-scoped
9093 static, function-scoped static, or static data member of a class. It may
9094 not be applied to block-scoped automatic or non-static data member.
9096 When the address-of operator is applied to a thread-local variable, it is
9097 evaluated at run-time and returns the address of the current thread's
9098 instance of that variable. An address so obtained may be used by any
9099 thread. When a thread terminates, any pointers to thread-local variables
9100 in that thread become invalid.
9102 No static initialization may refer to the address of a thread-local variable.
9104 In C++, if an initializer is present for a thread-local variable, it must
9105 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
9108 See @uref{http://people.redhat.com/drepper/tls.pdf,
9109 ELF Handling For Thread-Local Storage} for a detailed explanation of
9110 the four thread-local storage addressing models, and how the run-time
9111 is expected to function.
9114 * C99 Thread-Local Edits::
9115 * C++98 Thread-Local Edits::
9118 @node C99 Thread-Local Edits
9119 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
9121 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
9122 that document the exact semantics of the language extension.
9126 @cite{5.1.2 Execution environments}
9128 Add new text after paragraph 1
9131 Within either execution environment, a @dfn{thread} is a flow of
9132 control within a program. It is implementation defined whether
9133 or not there may be more than one thread associated with a program.
9134 It is implementation defined how threads beyond the first are
9135 created, the name and type of the function called at thread
9136 startup, and how threads may be terminated. However, objects
9137 with thread storage duration shall be initialized before thread
9142 @cite{6.2.4 Storage durations of objects}
9144 Add new text before paragraph 3
9147 An object whose identifier is declared with the storage-class
9148 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
9149 Its lifetime is the entire execution of the thread, and its
9150 stored value is initialized only once, prior to thread startup.
9154 @cite{6.4.1 Keywords}
9156 Add @code{__thread}.
9159 @cite{6.7.1 Storage-class specifiers}
9161 Add @code{__thread} to the list of storage class specifiers in
9164 Change paragraph 2 to
9167 With the exception of @code{__thread}, at most one storage-class
9168 specifier may be given [@dots{}]. The @code{__thread} specifier may
9169 be used alone, or immediately following @code{extern} or
9173 Add new text after paragraph 6
9176 The declaration of an identifier for a variable that has
9177 block scope that specifies @code{__thread} shall also
9178 specify either @code{extern} or @code{static}.
9180 The @code{__thread} specifier shall be used only with
9185 @node C++98 Thread-Local Edits
9186 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
9188 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
9189 that document the exact semantics of the language extension.
9193 @b{[intro.execution]}
9195 New text after paragraph 4
9198 A @dfn{thread} is a flow of control within the abstract machine.
9199 It is implementation defined whether or not there may be more than
9203 New text after paragraph 7
9206 It is unspecified whether additional action must be taken to
9207 ensure when and whether side effects are visible to other threads.
9213 Add @code{__thread}.
9216 @b{[basic.start.main]}
9218 Add after paragraph 5
9221 The thread that begins execution at the @code{main} function is called
9222 the @dfn{main thread}. It is implementation defined how functions
9223 beginning threads other than the main thread are designated or typed.
9224 A function so designated, as well as the @code{main} function, is called
9225 a @dfn{thread startup function}. It is implementation defined what
9226 happens if a thread startup function returns. It is implementation
9227 defined what happens to other threads when any thread calls @code{exit}.
9231 @b{[basic.start.init]}
9233 Add after paragraph 4
9236 The storage for an object of thread storage duration shall be
9237 statically initialized before the first statement of the thread startup
9238 function. An object of thread storage duration shall not require
9239 dynamic initialization.
9243 @b{[basic.start.term]}
9245 Add after paragraph 3
9248 The type of an object with thread storage duration shall not have a
9249 non-trivial destructor, nor shall it be an array type whose elements
9250 (directly or indirectly) have non-trivial destructors.
9256 Add ``thread storage duration'' to the list in paragraph 1.
9261 Thread, static, and automatic storage durations are associated with
9262 objects introduced by declarations [@dots{}].
9265 Add @code{__thread} to the list of specifiers in paragraph 3.
9268 @b{[basic.stc.thread]}
9270 New section before @b{[basic.stc.static]}
9273 The keyword @code{__thread} applied to a non-local object gives the
9274 object thread storage duration.
9276 A local variable or class data member declared both @code{static}
9277 and @code{__thread} gives the variable or member thread storage
9282 @b{[basic.stc.static]}
9287 All objects which have neither thread storage duration, dynamic
9288 storage duration nor are local [@dots{}].
9294 Add @code{__thread} to the list in paragraph 1.
9299 With the exception of @code{__thread}, at most one
9300 @var{storage-class-specifier} shall appear in a given
9301 @var{decl-specifier-seq}. The @code{__thread} specifier may
9302 be used alone, or immediately following the @code{extern} or
9303 @code{static} specifiers. [@dots{}]
9306 Add after paragraph 5
9309 The @code{__thread} specifier can be applied only to the names of objects
9310 and to anonymous unions.
9316 Add after paragraph 6
9319 Non-@code{static} members shall not be @code{__thread}.
9323 @node C++ Extensions
9324 @chapter Extensions to the C++ Language
9325 @cindex extensions, C++ language
9326 @cindex C++ language extensions
9328 The GNU compiler provides these extensions to the C++ language (and you
9329 can also use most of the C language extensions in your C++ programs). If you
9330 want to write code that checks whether these features are available, you can
9331 test for the GNU compiler the same way as for C programs: check for a
9332 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
9333 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
9334 Predefined Macros,cpp,The GNU C Preprocessor}).
9337 * Volatiles:: What constitutes an access to a volatile object.
9338 * Restricted Pointers:: C99 restricted pointers and references.
9339 * Vague Linkage:: Where G++ puts inlines, vtables and such.
9340 * C++ Interface:: You can use a single C++ header file for both
9341 declarations and definitions.
9342 * Template Instantiation:: Methods for ensuring that exactly one copy of
9343 each needed template instantiation is emitted.
9344 * Bound member functions:: You can extract a function pointer to the
9345 method denoted by a @samp{->*} or @samp{.*} expression.
9346 * C++ Attributes:: Variable, function, and type attributes for C++ only.
9347 * Strong Using:: Strong using-directives for namespace composition.
9348 * Java Exceptions:: Tweaking exception handling to work with Java.
9349 * Deprecated Features:: Things will disappear from g++.
9350 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
9354 @section When is a Volatile Object Accessed?
9355 @cindex accessing volatiles
9356 @cindex volatile read
9357 @cindex volatile write
9358 @cindex volatile access
9360 Both the C and C++ standard have the concept of volatile objects. These
9361 are normally accessed by pointers and used for accessing hardware. The
9362 standards encourage compilers to refrain from optimizations
9363 concerning accesses to volatile objects that it might perform on
9364 non-volatile objects. The C standard leaves it implementation defined
9365 as to what constitutes a volatile access. The C++ standard omits to
9366 specify this, except to say that C++ should behave in a similar manner
9367 to C with respect to volatiles, where possible. The minimum either
9368 standard specifies is that at a sequence point all previous accesses to
9369 volatile objects have stabilized and no subsequent accesses have
9370 occurred. Thus an implementation is free to reorder and combine
9371 volatile accesses which occur between sequence points, but cannot do so
9372 for accesses across a sequence point. The use of volatiles does not
9373 allow you to violate the restriction on updating objects multiple times
9374 within a sequence point.
9376 In most expressions, it is intuitively obvious what is a read and what is
9377 a write. For instance
9380 volatile int *dst = @var{somevalue};
9381 volatile int *src = @var{someothervalue};
9386 will cause a read of the volatile object pointed to by @var{src} and stores the
9387 value into the volatile object pointed to by @var{dst}. There is no
9388 guarantee that these reads and writes are atomic, especially for objects
9389 larger than @code{int}.
9391 Less obvious expressions are where something which looks like an access
9392 is used in a void context. An example would be,
9395 volatile int *src = @var{somevalue};
9399 With C, such expressions are rvalues, and as rvalues cause a read of
9400 the object, GCC interprets this as a read of the volatile being pointed
9401 to. The C++ standard specifies that such expressions do not undergo
9402 lvalue to rvalue conversion, and that the type of the dereferenced
9403 object may be incomplete. The C++ standard does not specify explicitly
9404 that it is this lvalue to rvalue conversion which is responsible for
9405 causing an access. However, there is reason to believe that it is,
9406 because otherwise certain simple expressions become undefined. However,
9407 because it would surprise most programmers, G++ treats dereferencing a
9408 pointer to volatile object of complete type in a void context as a read
9409 of the object. When the object has incomplete type, G++ issues a
9414 struct T @{int m;@};
9415 volatile S *ptr1 = @var{somevalue};
9416 volatile T *ptr2 = @var{somevalue};
9421 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
9422 causes a read of the object pointed to. If you wish to force an error on
9423 the first case, you must force a conversion to rvalue with, for instance
9424 a static cast, @code{static_cast<S>(*ptr1)}.
9426 When using a reference to volatile, G++ does not treat equivalent
9427 expressions as accesses to volatiles, but instead issues a warning that
9428 no volatile is accessed. The rationale for this is that otherwise it
9429 becomes difficult to determine where volatile access occur, and not
9430 possible to ignore the return value from functions returning volatile
9431 references. Again, if you wish to force a read, cast the reference to
9434 @node Restricted Pointers
9435 @section Restricting Pointer Aliasing
9436 @cindex restricted pointers
9437 @cindex restricted references
9438 @cindex restricted this pointer
9440 As with the C front end, G++ understands the C99 feature of restricted pointers,
9441 specified with the @code{__restrict__}, or @code{__restrict} type
9442 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
9443 language flag, @code{restrict} is not a keyword in C++.
9445 In addition to allowing restricted pointers, you can specify restricted
9446 references, which indicate that the reference is not aliased in the local
9450 void fn (int *__restrict__ rptr, int &__restrict__ rref)
9457 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
9458 @var{rref} refers to a (different) unaliased integer.
9460 You may also specify whether a member function's @var{this} pointer is
9461 unaliased by using @code{__restrict__} as a member function qualifier.
9464 void T::fn () __restrict__
9471 Within the body of @code{T::fn}, @var{this} will have the effective
9472 definition @code{T *__restrict__ const this}. Notice that the
9473 interpretation of a @code{__restrict__} member function qualifier is
9474 different to that of @code{const} or @code{volatile} qualifier, in that it
9475 is applied to the pointer rather than the object. This is consistent with
9476 other compilers which implement restricted pointers.
9478 As with all outermost parameter qualifiers, @code{__restrict__} is
9479 ignored in function definition matching. This means you only need to
9480 specify @code{__restrict__} in a function definition, rather than
9481 in a function prototype as well.
9484 @section Vague Linkage
9485 @cindex vague linkage
9487 There are several constructs in C++ which require space in the object
9488 file but are not clearly tied to a single translation unit. We say that
9489 these constructs have ``vague linkage''. Typically such constructs are
9490 emitted wherever they are needed, though sometimes we can be more
9494 @item Inline Functions
9495 Inline functions are typically defined in a header file which can be
9496 included in many different compilations. Hopefully they can usually be
9497 inlined, but sometimes an out-of-line copy is necessary, if the address
9498 of the function is taken or if inlining fails. In general, we emit an
9499 out-of-line copy in all translation units where one is needed. As an
9500 exception, we only emit inline virtual functions with the vtable, since
9501 it will always require a copy.
9503 Local static variables and string constants used in an inline function
9504 are also considered to have vague linkage, since they must be shared
9505 between all inlined and out-of-line instances of the function.
9509 C++ virtual functions are implemented in most compilers using a lookup
9510 table, known as a vtable. The vtable contains pointers to the virtual
9511 functions provided by a class, and each object of the class contains a
9512 pointer to its vtable (or vtables, in some multiple-inheritance
9513 situations). If the class declares any non-inline, non-pure virtual
9514 functions, the first one is chosen as the ``key method'' for the class,
9515 and the vtable is only emitted in the translation unit where the key
9518 @emph{Note:} If the chosen key method is later defined as inline, the
9519 vtable will still be emitted in every translation unit which defines it.
9520 Make sure that any inline virtuals are declared inline in the class
9521 body, even if they are not defined there.
9523 @item type_info objects
9526 C++ requires information about types to be written out in order to
9527 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
9528 For polymorphic classes (classes with virtual functions), the type_info
9529 object is written out along with the vtable so that @samp{dynamic_cast}
9530 can determine the dynamic type of a class object at runtime. For all
9531 other types, we write out the type_info object when it is used: when
9532 applying @samp{typeid} to an expression, throwing an object, or
9533 referring to a type in a catch clause or exception specification.
9535 @item Template Instantiations
9536 Most everything in this section also applies to template instantiations,
9537 but there are other options as well.
9538 @xref{Template Instantiation,,Where's the Template?}.
9542 When used with GNU ld version 2.8 or later on an ELF system such as
9543 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
9544 these constructs will be discarded at link time. This is known as
9547 On targets that don't support COMDAT, but do support weak symbols, GCC
9548 will use them. This way one copy will override all the others, but
9549 the unused copies will still take up space in the executable.
9551 For targets which do not support either COMDAT or weak symbols,
9552 most entities with vague linkage will be emitted as local symbols to
9553 avoid duplicate definition errors from the linker. This will not happen
9554 for local statics in inlines, however, as having multiple copies will
9555 almost certainly break things.
9557 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
9558 another way to control placement of these constructs.
9561 @section #pragma interface and implementation
9563 @cindex interface and implementation headers, C++
9564 @cindex C++ interface and implementation headers
9565 @cindex pragmas, interface and implementation
9567 @code{#pragma interface} and @code{#pragma implementation} provide the
9568 user with a way of explicitly directing the compiler to emit entities
9569 with vague linkage (and debugging information) in a particular
9572 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
9573 most cases, because of COMDAT support and the ``key method'' heuristic
9574 mentioned in @ref{Vague Linkage}. Using them can actually cause your
9575 program to grow due to unnecessary out-of-line copies of inline
9576 functions. Currently (3.4) the only benefit of these
9577 @code{#pragma}s is reduced duplication of debugging information, and
9578 that should be addressed soon on DWARF 2 targets with the use of
9582 @item #pragma interface
9583 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
9584 @kindex #pragma interface
9585 Use this directive in @emph{header files} that define object classes, to save
9586 space in most of the object files that use those classes. Normally,
9587 local copies of certain information (backup copies of inline member
9588 functions, debugging information, and the internal tables that implement
9589 virtual functions) must be kept in each object file that includes class
9590 definitions. You can use this pragma to avoid such duplication. When a
9591 header file containing @samp{#pragma interface} is included in a
9592 compilation, this auxiliary information will not be generated (unless
9593 the main input source file itself uses @samp{#pragma implementation}).
9594 Instead, the object files will contain references to be resolved at link
9597 The second form of this directive is useful for the case where you have
9598 multiple headers with the same name in different directories. If you
9599 use this form, you must specify the same string to @samp{#pragma
9602 @item #pragma implementation
9603 @itemx #pragma implementation "@var{objects}.h"
9604 @kindex #pragma implementation
9605 Use this pragma in a @emph{main input file}, when you want full output from
9606 included header files to be generated (and made globally visible). The
9607 included header file, in turn, should use @samp{#pragma interface}.
9608 Backup copies of inline member functions, debugging information, and the
9609 internal tables used to implement virtual functions are all generated in
9610 implementation files.
9612 @cindex implied @code{#pragma implementation}
9613 @cindex @code{#pragma implementation}, implied
9614 @cindex naming convention, implementation headers
9615 If you use @samp{#pragma implementation} with no argument, it applies to
9616 an include file with the same basename@footnote{A file's @dfn{basename}
9617 was the name stripped of all leading path information and of trailing
9618 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
9619 file. For example, in @file{allclass.cc}, giving just
9620 @samp{#pragma implementation}
9621 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
9623 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
9624 an implementation file whenever you would include it from
9625 @file{allclass.cc} even if you never specified @samp{#pragma
9626 implementation}. This was deemed to be more trouble than it was worth,
9627 however, and disabled.
9629 Use the string argument if you want a single implementation file to
9630 include code from multiple header files. (You must also use
9631 @samp{#include} to include the header file; @samp{#pragma
9632 implementation} only specifies how to use the file---it doesn't actually
9635 There is no way to split up the contents of a single header file into
9636 multiple implementation files.
9639 @cindex inlining and C++ pragmas
9640 @cindex C++ pragmas, effect on inlining
9641 @cindex pragmas in C++, effect on inlining
9642 @samp{#pragma implementation} and @samp{#pragma interface} also have an
9643 effect on function inlining.
9645 If you define a class in a header file marked with @samp{#pragma
9646 interface}, the effect on an inline function defined in that class is
9647 similar to an explicit @code{extern} declaration---the compiler emits
9648 no code at all to define an independent version of the function. Its
9649 definition is used only for inlining with its callers.
9651 @opindex fno-implement-inlines
9652 Conversely, when you include the same header file in a main source file
9653 that declares it as @samp{#pragma implementation}, the compiler emits
9654 code for the function itself; this defines a version of the function
9655 that can be found via pointers (or by callers compiled without
9656 inlining). If all calls to the function can be inlined, you can avoid
9657 emitting the function by compiling with @option{-fno-implement-inlines}.
9658 If any calls were not inlined, you will get linker errors.
9660 @node Template Instantiation
9661 @section Where's the Template?
9662 @cindex template instantiation
9664 C++ templates are the first language feature to require more
9665 intelligence from the environment than one usually finds on a UNIX
9666 system. Somehow the compiler and linker have to make sure that each
9667 template instance occurs exactly once in the executable if it is needed,
9668 and not at all otherwise. There are two basic approaches to this
9669 problem, which are referred to as the Borland model and the Cfront model.
9673 Borland C++ solved the template instantiation problem by adding the code
9674 equivalent of common blocks to their linker; the compiler emits template
9675 instances in each translation unit that uses them, and the linker
9676 collapses them together. The advantage of this model is that the linker
9677 only has to consider the object files themselves; there is no external
9678 complexity to worry about. This disadvantage is that compilation time
9679 is increased because the template code is being compiled repeatedly.
9680 Code written for this model tends to include definitions of all
9681 templates in the header file, since they must be seen to be
9685 The AT&T C++ translator, Cfront, solved the template instantiation
9686 problem by creating the notion of a template repository, an
9687 automatically maintained place where template instances are stored. A
9688 more modern version of the repository works as follows: As individual
9689 object files are built, the compiler places any template definitions and
9690 instantiations encountered in the repository. At link time, the link
9691 wrapper adds in the objects in the repository and compiles any needed
9692 instances that were not previously emitted. The advantages of this
9693 model are more optimal compilation speed and the ability to use the
9694 system linker; to implement the Borland model a compiler vendor also
9695 needs to replace the linker. The disadvantages are vastly increased
9696 complexity, and thus potential for error; for some code this can be
9697 just as transparent, but in practice it can been very difficult to build
9698 multiple programs in one directory and one program in multiple
9699 directories. Code written for this model tends to separate definitions
9700 of non-inline member templates into a separate file, which should be
9701 compiled separately.
9704 When used with GNU ld version 2.8 or later on an ELF system such as
9705 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
9706 Borland model. On other systems, G++ implements neither automatic
9709 A future version of G++ will support a hybrid model whereby the compiler
9710 will emit any instantiations for which the template definition is
9711 included in the compile, and store template definitions and
9712 instantiation context information into the object file for the rest.
9713 The link wrapper will extract that information as necessary and invoke
9714 the compiler to produce the remaining instantiations. The linker will
9715 then combine duplicate instantiations.
9717 In the mean time, you have the following options for dealing with
9718 template instantiations:
9723 Compile your template-using code with @option{-frepo}. The compiler will
9724 generate files with the extension @samp{.rpo} listing all of the
9725 template instantiations used in the corresponding object files which
9726 could be instantiated there; the link wrapper, @samp{collect2}, will
9727 then update the @samp{.rpo} files to tell the compiler where to place
9728 those instantiations and rebuild any affected object files. The
9729 link-time overhead is negligible after the first pass, as the compiler
9730 will continue to place the instantiations in the same files.
9732 This is your best option for application code written for the Borland
9733 model, as it will just work. Code written for the Cfront model will
9734 need to be modified so that the template definitions are available at
9735 one or more points of instantiation; usually this is as simple as adding
9736 @code{#include <tmethods.cc>} to the end of each template header.
9738 For library code, if you want the library to provide all of the template
9739 instantiations it needs, just try to link all of its object files
9740 together; the link will fail, but cause the instantiations to be
9741 generated as a side effect. Be warned, however, that this may cause
9742 conflicts if multiple libraries try to provide the same instantiations.
9743 For greater control, use explicit instantiation as described in the next
9747 @opindex fno-implicit-templates
9748 Compile your code with @option{-fno-implicit-templates} to disable the
9749 implicit generation of template instances, and explicitly instantiate
9750 all the ones you use. This approach requires more knowledge of exactly
9751 which instances you need than do the others, but it's less
9752 mysterious and allows greater control. You can scatter the explicit
9753 instantiations throughout your program, perhaps putting them in the
9754 translation units where the instances are used or the translation units
9755 that define the templates themselves; you can put all of the explicit
9756 instantiations you need into one big file; or you can create small files
9763 template class Foo<int>;
9764 template ostream& operator <<
9765 (ostream&, const Foo<int>&);
9768 for each of the instances you need, and create a template instantiation
9771 If you are using Cfront-model code, you can probably get away with not
9772 using @option{-fno-implicit-templates} when compiling files that don't
9773 @samp{#include} the member template definitions.
9775 If you use one big file to do the instantiations, you may want to
9776 compile it without @option{-fno-implicit-templates} so you get all of the
9777 instances required by your explicit instantiations (but not by any
9778 other files) without having to specify them as well.
9780 G++ has extended the template instantiation syntax given in the ISO
9781 standard to allow forward declaration of explicit instantiations
9782 (with @code{extern}), instantiation of the compiler support data for a
9783 template class (i.e.@: the vtable) without instantiating any of its
9784 members (with @code{inline}), and instantiation of only the static data
9785 members of a template class, without the support data or member
9786 functions (with (@code{static}):
9789 extern template int max (int, int);
9790 inline template class Foo<int>;
9791 static template class Foo<int>;
9795 Do nothing. Pretend G++ does implement automatic instantiation
9796 management. Code written for the Borland model will work fine, but
9797 each translation unit will contain instances of each of the templates it
9798 uses. In a large program, this can lead to an unacceptable amount of code
9802 @node Bound member functions
9803 @section Extracting the function pointer from a bound pointer to member function
9805 @cindex pointer to member function
9806 @cindex bound pointer to member function
9808 In C++, pointer to member functions (PMFs) are implemented using a wide
9809 pointer of sorts to handle all the possible call mechanisms; the PMF
9810 needs to store information about how to adjust the @samp{this} pointer,
9811 and if the function pointed to is virtual, where to find the vtable, and
9812 where in the vtable to look for the member function. If you are using
9813 PMFs in an inner loop, you should really reconsider that decision. If
9814 that is not an option, you can extract the pointer to the function that
9815 would be called for a given object/PMF pair and call it directly inside
9816 the inner loop, to save a bit of time.
9818 Note that you will still be paying the penalty for the call through a
9819 function pointer; on most modern architectures, such a call defeats the
9820 branch prediction features of the CPU@. This is also true of normal
9821 virtual function calls.
9823 The syntax for this extension is
9827 extern int (A::*fp)();
9828 typedef int (*fptr)(A *);
9830 fptr p = (fptr)(a.*fp);
9833 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
9834 no object is needed to obtain the address of the function. They can be
9835 converted to function pointers directly:
9838 fptr p1 = (fptr)(&A::foo);
9841 @opindex Wno-pmf-conversions
9842 You must specify @option{-Wno-pmf-conversions} to use this extension.
9844 @node C++ Attributes
9845 @section C++-Specific Variable, Function, and Type Attributes
9847 Some attributes only make sense for C++ programs.
9850 @item init_priority (@var{priority})
9851 @cindex init_priority attribute
9854 In Standard C++, objects defined at namespace scope are guaranteed to be
9855 initialized in an order in strict accordance with that of their definitions
9856 @emph{in a given translation unit}. No guarantee is made for initializations
9857 across translation units. However, GNU C++ allows users to control the
9858 order of initialization of objects defined at namespace scope with the
9859 @code{init_priority} attribute by specifying a relative @var{priority},
9860 a constant integral expression currently bounded between 101 and 65535
9861 inclusive. Lower numbers indicate a higher priority.
9863 In the following example, @code{A} would normally be created before
9864 @code{B}, but the @code{init_priority} attribute has reversed that order:
9867 Some_Class A __attribute__ ((init_priority (2000)));
9868 Some_Class B __attribute__ ((init_priority (543)));
9872 Note that the particular values of @var{priority} do not matter; only their
9875 @item java_interface
9876 @cindex java_interface attribute
9878 This type attribute informs C++ that the class is a Java interface. It may
9879 only be applied to classes declared within an @code{extern "Java"} block.
9880 Calls to methods declared in this interface will be dispatched using GCJ's
9881 interface table mechanism, instead of regular virtual table dispatch.
9885 See also @xref{Strong Using}.
9888 @section Strong Using
9890 @strong{Caution:} The semantics of this extension are not fully
9891 defined. Users should refrain from using this extension as its
9892 semantics may change subtly over time. It is possible that this
9893 extension wil be removed in future versions of G++.
9895 A using-directive with @code{__attribute ((strong))} is stronger
9896 than a normal using-directive in two ways:
9900 Templates from the used namespace can be specialized as though they were members of the using namespace.
9903 The using namespace is considered an associated namespace of all
9904 templates in the used namespace for purposes of argument-dependent
9908 This is useful for composing a namespace transparently from
9909 implementation namespaces. For example:
9914 template <class T> struct A @{ @};
9916 using namespace debug __attribute ((__strong__));
9917 template <> struct A<int> @{ @}; // @r{ok to specialize}
9919 template <class T> void f (A<T>);
9924 f (std::A<float>()); // @r{lookup finds} std::f
9929 @node Java Exceptions
9930 @section Java Exceptions
9932 The Java language uses a slightly different exception handling model
9933 from C++. Normally, GNU C++ will automatically detect when you are
9934 writing C++ code that uses Java exceptions, and handle them
9935 appropriately. However, if C++ code only needs to execute destructors
9936 when Java exceptions are thrown through it, GCC will guess incorrectly.
9937 Sample problematic code is:
9940 struct S @{ ~S(); @};
9941 extern void bar(); // @r{is written in Java, and may throw exceptions}
9950 The usual effect of an incorrect guess is a link failure, complaining of
9951 a missing routine called @samp{__gxx_personality_v0}.
9953 You can inform the compiler that Java exceptions are to be used in a
9954 translation unit, irrespective of what it might think, by writing
9955 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
9956 @samp{#pragma} must appear before any functions that throw or catch
9957 exceptions, or run destructors when exceptions are thrown through them.
9959 You cannot mix Java and C++ exceptions in the same translation unit. It
9960 is believed to be safe to throw a C++ exception from one file through
9961 another file compiled for the Java exception model, or vice versa, but
9962 there may be bugs in this area.
9964 @node Deprecated Features
9965 @section Deprecated Features
9967 In the past, the GNU C++ compiler was extended to experiment with new
9968 features, at a time when the C++ language was still evolving. Now that
9969 the C++ standard is complete, some of those features are superseded by
9970 superior alternatives. Using the old features might cause a warning in
9971 some cases that the feature will be dropped in the future. In other
9972 cases, the feature might be gone already.
9974 While the list below is not exhaustive, it documents some of the options
9975 that are now deprecated:
9978 @item -fexternal-templates
9979 @itemx -falt-external-templates
9980 These are two of the many ways for G++ to implement template
9981 instantiation. @xref{Template Instantiation}. The C++ standard clearly
9982 defines how template definitions have to be organized across
9983 implementation units. G++ has an implicit instantiation mechanism that
9984 should work just fine for standard-conforming code.
9986 @item -fstrict-prototype
9987 @itemx -fno-strict-prototype
9988 Previously it was possible to use an empty prototype parameter list to
9989 indicate an unspecified number of parameters (like C), rather than no
9990 parameters, as C++ demands. This feature has been removed, except where
9991 it is required for backwards compatibility @xref{Backwards Compatibility}.
9994 G++ allows a virtual function returning @samp{void *} to be overridden
9995 by one returning a different pointer type. This extension to the
9996 covariant return type rules is now deprecated and will be removed from a
9999 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
10000 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
10001 and will be removed in a future version. Code using these operators
10002 should be modified to use @code{std::min} and @code{std::max} instead.
10004 The named return value extension has been deprecated, and is now
10007 The use of initializer lists with new expressions has been deprecated,
10008 and is now removed from G++.
10010 Floating and complex non-type template parameters have been deprecated,
10011 and are now removed from G++.
10013 The implicit typename extension has been deprecated and is now
10016 The use of default arguments in function pointers, function typedefs and
10017 and other places where they are not permitted by the standard is
10018 deprecated and will be removed from a future version of G++.
10020 G++ allows floating-point literals to appear in integral constant expressions,
10021 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
10022 This extension is deprecated and will be removed from a future version.
10024 G++ allows static data members of const floating-point type to be declared
10025 with an initializer in a class definition. The standard only allows
10026 initializers for static members of const integral types and const
10027 enumeration types so this extension has been deprecated and will be removed
10028 from a future version.
10030 @node Backwards Compatibility
10031 @section Backwards Compatibility
10032 @cindex Backwards Compatibility
10033 @cindex ARM [Annotated C++ Reference Manual]
10035 Now that there is a definitive ISO standard C++, G++ has a specification
10036 to adhere to. The C++ language evolved over time, and features that
10037 used to be acceptable in previous drafts of the standard, such as the ARM
10038 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
10039 compilation of C++ written to such drafts, G++ contains some backwards
10040 compatibilities. @emph{All such backwards compatibility features are
10041 liable to disappear in future versions of G++.} They should be considered
10042 deprecated @xref{Deprecated Features}.
10046 If a variable is declared at for scope, it used to remain in scope until
10047 the end of the scope which contained the for statement (rather than just
10048 within the for scope). G++ retains this, but issues a warning, if such a
10049 variable is accessed outside the for scope.
10051 @item Implicit C language
10052 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
10053 scope to set the language. On such systems, all header files are
10054 implicitly scoped inside a C language scope. Also, an empty prototype
10055 @code{()} will be treated as an unspecified number of arguments, rather
10056 than no arguments, as C++ demands.