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 * Other Builtins:: Other built-in functions.
74 * Target Builtins:: Built-in functions specific to particular targets.
75 * Target Format Checks:: Format checks specific to particular targets.
76 * Pragmas:: Pragmas accepted by GCC.
77 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
78 * Thread-Local:: Per-thread variables.
82 @section Statements and Declarations in Expressions
83 @cindex statements inside expressions
84 @cindex declarations inside expressions
85 @cindex expressions containing statements
86 @cindex macros, statements in expressions
88 @c the above section title wrapped and causes an underfull hbox.. i
89 @c changed it from "within" to "in". --mew 4feb93
90 A compound statement enclosed in parentheses may appear as an expression
91 in GNU C@. This allows you to use loops, switches, and local variables
94 Recall that a compound statement is a sequence of statements surrounded
95 by braces; in this construct, parentheses go around the braces. For
99 (@{ int y = foo (); int z;
106 is a valid (though slightly more complex than necessary) expression
107 for the absolute value of @code{foo ()}.
109 The last thing in the compound statement should be an expression
110 followed by a semicolon; the value of this subexpression serves as the
111 value of the entire construct. (If you use some other kind of statement
112 last within the braces, the construct has type @code{void}, and thus
113 effectively no value.)
115 This feature is especially useful in making macro definitions ``safe'' (so
116 that they evaluate each operand exactly once). For example, the
117 ``maximum'' function is commonly defined as a macro in standard C as
121 #define max(a,b) ((a) > (b) ? (a) : (b))
125 @cindex side effects, macro argument
126 But this definition computes either @var{a} or @var{b} twice, with bad
127 results if the operand has side effects. In GNU C, if you know the
128 type of the operands (here taken as @code{int}), you can define
129 the macro safely as follows:
132 #define maxint(a,b) \
133 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
136 Embedded statements are not allowed in constant expressions, such as
137 the value of an enumeration constant, the width of a bit-field, or
138 the initial value of a static variable.
140 If you don't know the type of the operand, you can still do this, but you
141 must use @code{typeof} (@pxref{Typeof}).
143 In G++, the result value of a statement expression undergoes array and
144 function pointer decay, and is returned by value to the enclosing
145 expression. For instance, if @code{A} is a class, then
154 will construct a temporary @code{A} object to hold the result of the
155 statement expression, and that will be used to invoke @code{Foo}.
156 Therefore the @code{this} pointer observed by @code{Foo} will not be the
159 Any temporaries created within a statement within a statement expression
160 will be destroyed at the statement's end. This makes statement
161 expressions inside macros slightly different from function calls. In
162 the latter case temporaries introduced during argument evaluation will
163 be destroyed at the end of the statement that includes the function
164 call. In the statement expression case they will be destroyed during
165 the statement expression. For instance,
168 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
169 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
179 will have different places where temporaries are destroyed. For the
180 @code{macro} case, the temporary @code{X} will be destroyed just after
181 the initialization of @code{b}. In the @code{function} case that
182 temporary will be destroyed when the function returns.
184 These considerations mean that it is probably a bad idea to use
185 statement-expressions of this form in header files that are designed to
186 work with C++. (Note that some versions of the GNU C Library contained
187 header files using statement-expression that lead to precisely this
190 Jumping into a statement expression with @code{goto} or using a
191 @code{switch} statement outside the statement expression with a
192 @code{case} or @code{default} label inside the statement expression is
193 not permitted. Jumping into a statement expression with a computed
194 @code{goto} (@pxref{Labels as Values}) yields undefined behavior.
195 Jumping out of a statement expression is permitted, but if the
196 statement expression is part of a larger expression then it is
197 unspecified which other subexpressions of that expression have been
198 evaluated except where the language definition requires certain
199 subexpressions to be evaluated before or after the statement
200 expression. In any case, as with a function call the evaluation of a
201 statement expression is not interleaved with the evaluation of other
202 parts of the containing expression. For example,
205 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
209 will call @code{foo} and @code{bar1} and will not call @code{baz} but
210 may or may not call @code{bar2}. If @code{bar2} is called, it will be
211 called after @code{foo} and before @code{bar1}
214 @section Locally Declared Labels
216 @cindex macros, local labels
218 GCC allows you to declare @dfn{local labels} in any nested block
219 scope. A local label is just like an ordinary label, but you can
220 only reference it (with a @code{goto} statement, or by taking its
221 address) within the block in which it was declared.
223 A local label declaration looks like this:
226 __label__ @var{label};
233 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
236 Local label declarations must come at the beginning of the block,
237 before any ordinary declarations or statements.
239 The label declaration defines the label @emph{name}, but does not define
240 the label itself. You must do this in the usual way, with
241 @code{@var{label}:}, within the statements of the statement expression.
243 The local label feature is useful for complex macros. If a macro
244 contains nested loops, a @code{goto} can be useful for breaking out of
245 them. However, an ordinary label whose scope is the whole function
246 cannot be used: if the macro can be expanded several times in one
247 function, the label will be multiply defined in that function. A
248 local label avoids this problem. For example:
251 #define SEARCH(value, array, target) \
254 typeof (target) _SEARCH_target = (target); \
255 typeof (*(array)) *_SEARCH_array = (array); \
258 for (i = 0; i < max; i++) \
259 for (j = 0; j < max; j++) \
260 if (_SEARCH_array[i][j] == _SEARCH_target) \
261 @{ (value) = i; goto found; @} \
267 This could also be written using a statement-expression:
270 #define SEARCH(array, target) \
273 typeof (target) _SEARCH_target = (target); \
274 typeof (*(array)) *_SEARCH_array = (array); \
277 for (i = 0; i < max; i++) \
278 for (j = 0; j < max; j++) \
279 if (_SEARCH_array[i][j] == _SEARCH_target) \
280 @{ value = i; goto found; @} \
287 Local label declarations also make the labels they declare visible to
288 nested functions, if there are any. @xref{Nested Functions}, for details.
290 @node Labels as Values
291 @section Labels as Values
292 @cindex labels as values
293 @cindex computed gotos
294 @cindex goto with computed label
295 @cindex address of a label
297 You can get the address of a label defined in the current function
298 (or a containing function) with the unary operator @samp{&&}. The
299 value has type @code{void *}. This value is a constant and can be used
300 wherever a constant of that type is valid. For example:
308 To use these values, you need to be able to jump to one. This is done
309 with the computed goto statement@footnote{The analogous feature in
310 Fortran is called an assigned goto, but that name seems inappropriate in
311 C, where one can do more than simply store label addresses in label
312 variables.}, @code{goto *@var{exp};}. For example,
319 Any expression of type @code{void *} is allowed.
321 One way of using these constants is in initializing a static array that
322 will serve as a jump table:
325 static void *array[] = @{ &&foo, &&bar, &&hack @};
328 Then you can select a label with indexing, like this:
335 Note that this does not check whether the subscript is in bounds---array
336 indexing in C never does that.
338 Such an array of label values serves a purpose much like that of the
339 @code{switch} statement. The @code{switch} statement is cleaner, so
340 use that rather than an array unless the problem does not fit a
341 @code{switch} statement very well.
343 Another use of label values is in an interpreter for threaded code.
344 The labels within the interpreter function can be stored in the
345 threaded code for super-fast dispatching.
347 You may not use this mechanism to jump to code in a different function.
348 If you do that, totally unpredictable things will happen. The best way to
349 avoid this is to store the label address only in automatic variables and
350 never pass it as an argument.
352 An alternate way to write the above example is
355 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
357 goto *(&&foo + array[i]);
361 This is more friendly to code living in shared libraries, as it reduces
362 the number of dynamic relocations that are needed, and by consequence,
363 allows the data to be read-only.
365 @node Nested Functions
366 @section Nested Functions
367 @cindex nested functions
368 @cindex downward funargs
371 A @dfn{nested function} is a function defined inside another function.
372 (Nested functions are not supported for GNU C++.) The nested function's
373 name is local to the block where it is defined. For example, here we
374 define a nested function named @code{square}, and call it twice:
378 foo (double a, double b)
380 double square (double z) @{ return z * z; @}
382 return square (a) + square (b);
387 The nested function can access all the variables of the containing
388 function that are visible at the point of its definition. This is
389 called @dfn{lexical scoping}. For example, here we show a nested
390 function which uses an inherited variable named @code{offset}:
394 bar (int *array, int offset, int size)
396 int access (int *array, int index)
397 @{ return array[index + offset]; @}
400 for (i = 0; i < size; i++)
401 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
406 Nested function definitions are permitted within functions in the places
407 where variable definitions are allowed; that is, in any block, mixed
408 with the other declarations and statements in the block.
410 It is possible to call the nested function from outside the scope of its
411 name by storing its address or passing the address to another function:
414 hack (int *array, int size)
416 void store (int index, int value)
417 @{ array[index] = value; @}
419 intermediate (store, size);
423 Here, the function @code{intermediate} receives the address of
424 @code{store} as an argument. If @code{intermediate} calls @code{store},
425 the arguments given to @code{store} are used to store into @code{array}.
426 But this technique works only so long as the containing function
427 (@code{hack}, in this example) does not exit.
429 If you try to call the nested function through its address after the
430 containing function has exited, all hell will break loose. If you try
431 to call it after a containing scope level has exited, and if it refers
432 to some of the variables that are no longer in scope, you may be lucky,
433 but it's not wise to take the risk. If, however, the nested function
434 does not refer to anything that has gone out of scope, you should be
437 GCC implements taking the address of a nested function using a technique
438 called @dfn{trampolines}. A paper describing them is available as
441 @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}.
443 A nested function can jump to a label inherited from a containing
444 function, provided the label was explicitly declared in the containing
445 function (@pxref{Local Labels}). Such a jump returns instantly to the
446 containing function, exiting the nested function which did the
447 @code{goto} and any intermediate functions as well. Here is an example:
451 bar (int *array, int offset, int size)
454 int access (int *array, int index)
458 return array[index + offset];
462 for (i = 0; i < size; i++)
463 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
467 /* @r{Control comes here from @code{access}
468 if it detects an error.} */
475 A nested function always has no linkage. Declaring one with
476 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
477 before its definition, use @code{auto} (which is otherwise meaningless
478 for function declarations).
481 bar (int *array, int offset, int size)
484 auto int access (int *, int);
486 int access (int *array, int index)
490 return array[index + offset];
496 @node Constructing Calls
497 @section Constructing Function Calls
498 @cindex constructing calls
499 @cindex forwarding calls
501 Using the built-in functions described below, you can record
502 the arguments a function received, and call another function
503 with the same arguments, without knowing the number or types
506 You can also record the return value of that function call,
507 and later return that value, without knowing what data type
508 the function tried to return (as long as your caller expects
511 However, these built-in functions may interact badly with some
512 sophisticated features or other extensions of the language. It
513 is, therefore, not recommended to use them outside very simple
514 functions acting as mere forwarders for their arguments.
516 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
517 This built-in function returns a pointer to data
518 describing how to perform a call with the same arguments as were passed
519 to the current function.
521 The function saves the arg pointer register, structure value address,
522 and all registers that might be used to pass arguments to a function
523 into a block of memory allocated on the stack. Then it returns the
524 address of that block.
527 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
528 This built-in function invokes @var{function}
529 with a copy of the parameters described by @var{arguments}
532 The value of @var{arguments} should be the value returned by
533 @code{__builtin_apply_args}. The argument @var{size} specifies the size
534 of the stack argument data, in bytes.
536 This function returns a pointer to data describing
537 how to return whatever value was returned by @var{function}. The data
538 is saved in a block of memory allocated on the stack.
540 It is not always simple to compute the proper value for @var{size}. The
541 value is used by @code{__builtin_apply} to compute the amount of data
542 that should be pushed on the stack and copied from the incoming argument
546 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
547 This built-in function returns the value described by @var{result} from
548 the containing function. You should specify, for @var{result}, a value
549 returned by @code{__builtin_apply}.
553 @section Referring to a Type with @code{typeof}
556 @cindex macros, types of arguments
558 Another way to refer to the type of an expression is with @code{typeof}.
559 The syntax of using of this keyword looks like @code{sizeof}, but the
560 construct acts semantically like a type name defined with @code{typedef}.
562 There are two ways of writing the argument to @code{typeof}: with an
563 expression or with a type. Here is an example with an expression:
570 This assumes that @code{x} is an array of pointers to functions;
571 the type described is that of the values of the functions.
573 Here is an example with a typename as the argument:
580 Here the type described is that of pointers to @code{int}.
582 If you are writing a header file that must work when included in ISO C
583 programs, write @code{__typeof__} instead of @code{typeof}.
584 @xref{Alternate Keywords}.
586 A @code{typeof}-construct can be used anywhere a typedef name could be
587 used. For example, you can use it in a declaration, in a cast, or inside
588 of @code{sizeof} or @code{typeof}.
590 @code{typeof} is often useful in conjunction with the
591 statements-within-expressions feature. Here is how the two together can
592 be used to define a safe ``maximum'' macro that operates on any
593 arithmetic type and evaluates each of its arguments exactly once:
597 (@{ typeof (a) _a = (a); \
598 typeof (b) _b = (b); \
599 _a > _b ? _a : _b; @})
602 @cindex underscores in variables in macros
603 @cindex @samp{_} in variables in macros
604 @cindex local variables in macros
605 @cindex variables, local, in macros
606 @cindex macros, local variables in
608 The reason for using names that start with underscores for the local
609 variables is to avoid conflicts with variable names that occur within the
610 expressions that are substituted for @code{a} and @code{b}. Eventually we
611 hope to design a new form of declaration syntax that allows you to declare
612 variables whose scopes start only after their initializers; this will be a
613 more reliable way to prevent such conflicts.
616 Some more examples of the use of @code{typeof}:
620 This declares @code{y} with the type of what @code{x} points to.
627 This declares @code{y} as an array of such values.
634 This declares @code{y} as an array of pointers to characters:
637 typeof (typeof (char *)[4]) y;
641 It is equivalent to the following traditional C declaration:
647 To see the meaning of the declaration using @code{typeof}, and why it
648 might be a useful way to write, rewrite it with these macros:
651 #define pointer(T) typeof(T *)
652 #define array(T, N) typeof(T [N])
656 Now the declaration can be rewritten this way:
659 array (pointer (char), 4) y;
663 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
664 pointers to @code{char}.
667 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
668 a more limited extension which permitted one to write
671 typedef @var{T} = @var{expr};
675 with the effect of declaring @var{T} to have the type of the expression
676 @var{expr}. This extension does not work with GCC 3 (versions between
677 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
678 relies on it should be rewritten to use @code{typeof}:
681 typedef typeof(@var{expr}) @var{T};
685 This will work with all versions of GCC@.
688 @section Conditionals with Omitted Operands
689 @cindex conditional expressions, extensions
690 @cindex omitted middle-operands
691 @cindex middle-operands, omitted
692 @cindex extensions, @code{?:}
693 @cindex @code{?:} extensions
695 The middle operand in a conditional expression may be omitted. Then
696 if the first operand is nonzero, its value is the value of the conditional
699 Therefore, the expression
706 has the value of @code{x} if that is nonzero; otherwise, the value of
709 This example is perfectly equivalent to
715 @cindex side effect in ?:
716 @cindex ?: side effect
718 In this simple case, the ability to omit the middle operand is not
719 especially useful. When it becomes useful is when the first operand does,
720 or may (if it is a macro argument), contain a side effect. Then repeating
721 the operand in the middle would perform the side effect twice. Omitting
722 the middle operand uses the value already computed without the undesirable
723 effects of recomputing it.
726 @section Double-Word Integers
727 @cindex @code{long long} data types
728 @cindex double-word arithmetic
729 @cindex multiprecision arithmetic
730 @cindex @code{LL} integer suffix
731 @cindex @code{ULL} integer suffix
733 ISO C99 supports data types for integers that are at least 64 bits wide,
734 and as an extension GCC supports them in C89 mode and in C++.
735 Simply write @code{long long int} for a signed integer, or
736 @code{unsigned long long int} for an unsigned integer. To make an
737 integer constant of type @code{long long int}, add the suffix @samp{LL}
738 to the integer. To make an integer constant of type @code{unsigned long
739 long int}, add the suffix @samp{ULL} to the integer.
741 You can use these types in arithmetic like any other integer types.
742 Addition, subtraction, and bitwise boolean operations on these types
743 are open-coded on all types of machines. Multiplication is open-coded
744 if the machine supports fullword-to-doubleword a widening multiply
745 instruction. Division and shifts are open-coded only on machines that
746 provide special support. The operations that are not open-coded use
747 special library routines that come with GCC@.
749 There may be pitfalls when you use @code{long long} types for function
750 arguments, unless you declare function prototypes. If a function
751 expects type @code{int} for its argument, and you pass a value of type
752 @code{long long int}, confusion will result because the caller and the
753 subroutine will disagree about the number of bytes for the argument.
754 Likewise, if the function expects @code{long long int} and you pass
755 @code{int}. The best way to avoid such problems is to use prototypes.
758 @section Complex Numbers
759 @cindex complex numbers
760 @cindex @code{_Complex} keyword
761 @cindex @code{__complex__} keyword
763 ISO C99 supports complex floating data types, and as an extension GCC
764 supports them in C89 mode and in C++, and supports complex integer data
765 types which are not part of ISO C99. You can declare complex types
766 using the keyword @code{_Complex}. As an extension, the older GNU
767 keyword @code{__complex__} is also supported.
769 For example, @samp{_Complex double x;} declares @code{x} as a
770 variable whose real part and imaginary part are both of type
771 @code{double}. @samp{_Complex short int y;} declares @code{y} to
772 have real and imaginary parts of type @code{short int}; this is not
773 likely to be useful, but it shows that the set of complex types is
776 To write a constant with a complex data type, use the suffix @samp{i} or
777 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
778 has type @code{_Complex float} and @code{3i} has type
779 @code{_Complex int}. Such a constant always has a pure imaginary
780 value, but you can form any complex value you like by adding one to a
781 real constant. This is a GNU extension; if you have an ISO C99
782 conforming C library (such as GNU libc), and want to construct complex
783 constants of floating type, you should include @code{<complex.h>} and
784 use the macros @code{I} or @code{_Complex_I} instead.
786 @cindex @code{__real__} keyword
787 @cindex @code{__imag__} keyword
788 To extract the real part of a complex-valued expression @var{exp}, write
789 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
790 extract the imaginary part. This is a GNU extension; for values of
791 floating type, you should use the ISO C99 functions @code{crealf},
792 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
793 @code{cimagl}, declared in @code{<complex.h>} and also provided as
794 built-in functions by GCC@.
796 @cindex complex conjugation
797 The operator @samp{~} performs complex conjugation when used on a value
798 with a complex type. This is a GNU extension; for values of
799 floating type, you should use the ISO C99 functions @code{conjf},
800 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
801 provided as built-in functions by GCC@.
803 GCC can allocate complex automatic variables in a noncontiguous
804 fashion; it's even possible for the real part to be in a register while
805 the imaginary part is on the stack (or vice-versa). Only the DWARF2
806 debug info format can represent this, so use of DWARF2 is recommended.
807 If you are using the stabs debug info format, GCC describes a noncontiguous
808 complex variable as if it were two separate variables of noncomplex type.
809 If the variable's actual name is @code{foo}, the two fictitious
810 variables are named @code{foo$real} and @code{foo$imag}. You can
811 examine and set these two fictitious variables with your debugger.
817 ISO C99 supports floating-point numbers written not only in the usual
818 decimal notation, such as @code{1.55e1}, but also numbers such as
819 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
820 supports this in C89 mode (except in some cases when strictly
821 conforming) and in C++. In that format the
822 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
823 mandatory. The exponent is a decimal number that indicates the power of
824 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
831 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
832 is the same as @code{1.55e1}.
834 Unlike for floating-point numbers in the decimal notation the exponent
835 is always required in the hexadecimal notation. Otherwise the compiler
836 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
837 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
838 extension for floating-point constants of type @code{float}.
841 @section Arrays of Length Zero
842 @cindex arrays of length zero
843 @cindex zero-length arrays
844 @cindex length-zero arrays
845 @cindex flexible array members
847 Zero-length arrays are allowed in GNU C@. They are very useful as the
848 last element of a structure which is really a header for a variable-length
857 struct line *thisline = (struct line *)
858 malloc (sizeof (struct line) + this_length);
859 thisline->length = this_length;
862 In ISO C90, you would have to give @code{contents} a length of 1, which
863 means either you waste space or complicate the argument to @code{malloc}.
865 In ISO C99, you would use a @dfn{flexible array member}, which is
866 slightly different in syntax and semantics:
870 Flexible array members are written as @code{contents[]} without
874 Flexible array members have incomplete type, and so the @code{sizeof}
875 operator may not be applied. As a quirk of the original implementation
876 of zero-length arrays, @code{sizeof} evaluates to zero.
879 Flexible array members may only appear as the last member of a
880 @code{struct} that is otherwise non-empty.
883 A structure containing a flexible array member, or a union containing
884 such a structure (possibly recursively), may not be a member of a
885 structure or an element of an array. (However, these uses are
886 permitted by GCC as extensions.)
889 GCC versions before 3.0 allowed zero-length arrays to be statically
890 initialized, as if they were flexible arrays. In addition to those
891 cases that were useful, it also allowed initializations in situations
892 that would corrupt later data. Non-empty initialization of zero-length
893 arrays is now treated like any case where there are more initializer
894 elements than the array holds, in that a suitable warning about "excess
895 elements in array" is given, and the excess elements (all of them, in
896 this case) are ignored.
898 Instead GCC allows static initialization of flexible array members.
899 This is equivalent to defining a new structure containing the original
900 structure followed by an array of sufficient size to contain the data.
901 I.e.@: in the following, @code{f1} is constructed as if it were declared
907 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
910 struct f1 f1; int data[3];
911 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
915 The convenience of this extension is that @code{f1} has the desired
916 type, eliminating the need to consistently refer to @code{f2.f1}.
918 This has symmetry with normal static arrays, in that an array of
919 unknown size is also written with @code{[]}.
921 Of course, this extension only makes sense if the extra data comes at
922 the end of a top-level object, as otherwise we would be overwriting
923 data at subsequent offsets. To avoid undue complication and confusion
924 with initialization of deeply nested arrays, we simply disallow any
925 non-empty initialization except when the structure is the top-level
929 struct foo @{ int x; int y[]; @};
930 struct bar @{ struct foo z; @};
932 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
933 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
934 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
935 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
938 @node Empty Structures
939 @section Structures With No Members
940 @cindex empty structures
941 @cindex zero-size structures
943 GCC permits a C structure to have no members:
950 The structure will have size zero. In C++, empty structures are part
951 of the language. G++ treats empty structures as if they had a single
952 member of type @code{char}.
954 @node Variable Length
955 @section Arrays of Variable Length
956 @cindex variable-length arrays
957 @cindex arrays of variable length
960 Variable-length automatic arrays are allowed in ISO C99, and as an
961 extension GCC accepts them in C89 mode and in C++. (However, GCC's
962 implementation of variable-length arrays does not yet conform in detail
963 to the ISO C99 standard.) These arrays are
964 declared like any other automatic arrays, but with a length that is not
965 a constant expression. The storage is allocated at the point of
966 declaration and deallocated when the brace-level is exited. For
971 concat_fopen (char *s1, char *s2, char *mode)
973 char str[strlen (s1) + strlen (s2) + 1];
976 return fopen (str, mode);
980 @cindex scope of a variable length array
981 @cindex variable-length array scope
982 @cindex deallocating variable length arrays
983 Jumping or breaking out of the scope of the array name deallocates the
984 storage. Jumping into the scope is not allowed; you get an error
987 @cindex @code{alloca} vs variable-length arrays
988 You can use the function @code{alloca} to get an effect much like
989 variable-length arrays. The function @code{alloca} is available in
990 many other C implementations (but not in all). On the other hand,
991 variable-length arrays are more elegant.
993 There are other differences between these two methods. Space allocated
994 with @code{alloca} exists until the containing @emph{function} returns.
995 The space for a variable-length array is deallocated as soon as the array
996 name's scope ends. (If you use both variable-length arrays and
997 @code{alloca} in the same function, deallocation of a variable-length array
998 will also deallocate anything more recently allocated with @code{alloca}.)
1000 You can also use variable-length arrays as arguments to functions:
1004 tester (int len, char data[len][len])
1010 The length of an array is computed once when the storage is allocated
1011 and is remembered for the scope of the array in case you access it with
1014 If you want to pass the array first and the length afterward, you can
1015 use a forward declaration in the parameter list---another GNU extension.
1019 tester (int len; char data[len][len], int len)
1025 @cindex parameter forward declaration
1026 The @samp{int len} before the semicolon is a @dfn{parameter forward
1027 declaration}, and it serves the purpose of making the name @code{len}
1028 known when the declaration of @code{data} is parsed.
1030 You can write any number of such parameter forward declarations in the
1031 parameter list. They can be separated by commas or semicolons, but the
1032 last one must end with a semicolon, which is followed by the ``real''
1033 parameter declarations. Each forward declaration must match a ``real''
1034 declaration in parameter name and data type. ISO C99 does not support
1035 parameter forward declarations.
1037 @node Variadic Macros
1038 @section Macros with a Variable Number of Arguments.
1039 @cindex variable number of arguments
1040 @cindex macro with variable arguments
1041 @cindex rest argument (in macro)
1042 @cindex variadic macros
1044 In the ISO C standard of 1999, a macro can be declared to accept a
1045 variable number of arguments much as a function can. The syntax for
1046 defining the macro is similar to that of a function. Here is an
1050 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1053 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1054 such a macro, it represents the zero or more tokens until the closing
1055 parenthesis that ends the invocation, including any commas. This set of
1056 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1057 wherever it appears. See the CPP manual for more information.
1059 GCC has long supported variadic macros, and used a different syntax that
1060 allowed you to give a name to the variable arguments just like any other
1061 argument. Here is an example:
1064 #define debug(format, args...) fprintf (stderr, format, args)
1067 This is in all ways equivalent to the ISO C example above, but arguably
1068 more readable and descriptive.
1070 GNU CPP has two further variadic macro extensions, and permits them to
1071 be used with either of the above forms of macro definition.
1073 In standard C, you are not allowed to leave the variable argument out
1074 entirely; but you are allowed to pass an empty argument. For example,
1075 this invocation is invalid in ISO C, because there is no comma after
1082 GNU CPP permits you to completely omit the variable arguments in this
1083 way. In the above examples, the compiler would complain, though since
1084 the expansion of the macro still has the extra comma after the format
1087 To help solve this problem, CPP behaves specially for variable arguments
1088 used with the token paste operator, @samp{##}. If instead you write
1091 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1094 and if the variable arguments are omitted or empty, the @samp{##}
1095 operator causes the preprocessor to remove the comma before it. If you
1096 do provide some variable arguments in your macro invocation, GNU CPP
1097 does not complain about the paste operation and instead places the
1098 variable arguments after the comma. Just like any other pasted macro
1099 argument, these arguments are not macro expanded.
1101 @node Escaped Newlines
1102 @section Slightly Looser Rules for Escaped Newlines
1103 @cindex escaped newlines
1104 @cindex newlines (escaped)
1106 Recently, the preprocessor has relaxed its treatment of escaped
1107 newlines. Previously, the newline had to immediately follow a
1108 backslash. The current implementation allows whitespace in the form
1109 of spaces, horizontal and vertical tabs, and form feeds between the
1110 backslash and the subsequent newline. The preprocessor issues a
1111 warning, but treats it as a valid escaped newline and combines the two
1112 lines to form a single logical line. This works within comments and
1113 tokens, as well as between tokens. Comments are @emph{not} treated as
1114 whitespace for the purposes of this relaxation, since they have not
1115 yet been replaced with spaces.
1118 @section Non-Lvalue Arrays May Have Subscripts
1119 @cindex subscripting
1120 @cindex arrays, non-lvalue
1122 @cindex subscripting and function values
1123 In ISO C99, arrays that are not lvalues still decay to pointers, and
1124 may be subscripted, although they may not be modified or used after
1125 the next sequence point and the unary @samp{&} operator may not be
1126 applied to them. As an extension, GCC allows such arrays to be
1127 subscripted in C89 mode, though otherwise they do not decay to
1128 pointers outside C99 mode. For example,
1129 this is valid in GNU C though not valid in C89:
1133 struct foo @{int a[4];@};
1139 return f().a[index];
1145 @section Arithmetic on @code{void}- and Function-Pointers
1146 @cindex void pointers, arithmetic
1147 @cindex void, size of pointer to
1148 @cindex function pointers, arithmetic
1149 @cindex function, size of pointer to
1151 In GNU C, addition and subtraction operations are supported on pointers to
1152 @code{void} and on pointers to functions. This is done by treating the
1153 size of a @code{void} or of a function as 1.
1155 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1156 and on function types, and returns 1.
1158 @opindex Wpointer-arith
1159 The option @option{-Wpointer-arith} requests a warning if these extensions
1163 @section Non-Constant Initializers
1164 @cindex initializers, non-constant
1165 @cindex non-constant initializers
1167 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1168 automatic variable are not required to be constant expressions in GNU C@.
1169 Here is an example of an initializer with run-time varying elements:
1172 foo (float f, float g)
1174 float beat_freqs[2] = @{ f-g, f+g @};
1179 @node Compound Literals
1180 @section Compound Literals
1181 @cindex constructor expressions
1182 @cindex initializations in expressions
1183 @cindex structures, constructor expression
1184 @cindex expressions, constructor
1185 @cindex compound literals
1186 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1188 ISO C99 supports compound literals. A compound literal looks like
1189 a cast containing an initializer. Its value is an object of the
1190 type specified in the cast, containing the elements specified in
1191 the initializer; it is an lvalue. As an extension, GCC supports
1192 compound literals in C89 mode and in C++.
1194 Usually, the specified type is a structure. Assume that
1195 @code{struct foo} and @code{structure} are declared as shown:
1198 struct foo @{int a; char b[2];@} structure;
1202 Here is an example of constructing a @code{struct foo} with a compound literal:
1205 structure = ((struct foo) @{x + y, 'a', 0@});
1209 This is equivalent to writing the following:
1213 struct foo temp = @{x + y, 'a', 0@};
1218 You can also construct an array. If all the elements of the compound literal
1219 are (made up of) simple constant expressions, suitable for use in
1220 initializers of objects of static storage duration, then the compound
1221 literal can be coerced to a pointer to its first element and used in
1222 such an initializer, as shown here:
1225 char **foo = (char *[]) @{ "x", "y", "z" @};
1228 Compound literals for scalar types and union types are is
1229 also allowed, but then the compound literal is equivalent
1232 As a GNU extension, GCC allows initialization of objects with static storage
1233 duration by compound literals (which is not possible in ISO C99, because
1234 the initializer is not a constant).
1235 It is handled as if the object was initialized only with the bracket
1236 enclosed list if compound literal's and object types match.
1237 The initializer list of the compound literal must be constant.
1238 If the object being initialized has array type of unknown size, the size is
1239 determined by compound literal size.
1242 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1243 static int y[] = (int []) @{1, 2, 3@};
1244 static int z[] = (int [3]) @{1@};
1248 The above lines are equivalent to the following:
1250 static struct foo x = @{1, 'a', 'b'@};
1251 static int y[] = @{1, 2, 3@};
1252 static int z[] = @{1, 0, 0@};
1255 @node Designated Inits
1256 @section Designated Initializers
1257 @cindex initializers with labeled elements
1258 @cindex labeled elements in initializers
1259 @cindex case labels in initializers
1260 @cindex designated initializers
1262 Standard C89 requires the elements of an initializer to appear in a fixed
1263 order, the same as the order of the elements in the array or structure
1266 In ISO C99 you can give the elements in any order, specifying the array
1267 indices or structure field names they apply to, and GNU C allows this as
1268 an extension in C89 mode as well. This extension is not
1269 implemented in GNU C++.
1271 To specify an array index, write
1272 @samp{[@var{index}] =} before the element value. For example,
1275 int a[6] = @{ [4] = 29, [2] = 15 @};
1282 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1286 The index values must be constant expressions, even if the array being
1287 initialized is automatic.
1289 An alternative syntax for this which has been obsolete since GCC 2.5 but
1290 GCC still accepts is to write @samp{[@var{index}]} before the element
1291 value, with no @samp{=}.
1293 To initialize a range of elements to the same value, write
1294 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1295 extension. For example,
1298 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1302 If the value in it has side-effects, the side-effects will happen only once,
1303 not for each initialized field by the range initializer.
1306 Note that the length of the array is the highest value specified
1309 In a structure initializer, specify the name of a field to initialize
1310 with @samp{.@var{fieldname} =} before the element value. For example,
1311 given the following structure,
1314 struct point @{ int x, y; @};
1318 the following initialization
1321 struct point p = @{ .y = yvalue, .x = xvalue @};
1328 struct point p = @{ xvalue, yvalue @};
1331 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1332 @samp{@var{fieldname}:}, as shown here:
1335 struct point p = @{ y: yvalue, x: xvalue @};
1339 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1340 @dfn{designator}. You can also use a designator (or the obsolete colon
1341 syntax) when initializing a union, to specify which element of the union
1342 should be used. For example,
1345 union foo @{ int i; double d; @};
1347 union foo f = @{ .d = 4 @};
1351 will convert 4 to a @code{double} to store it in the union using
1352 the second element. By contrast, casting 4 to type @code{union foo}
1353 would store it into the union as the integer @code{i}, since it is
1354 an integer. (@xref{Cast to Union}.)
1356 You can combine this technique of naming elements with ordinary C
1357 initialization of successive elements. Each initializer element that
1358 does not have a designator applies to the next consecutive element of the
1359 array or structure. For example,
1362 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1369 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1372 Labeling the elements of an array initializer is especially useful
1373 when the indices are characters or belong to an @code{enum} type.
1378 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1379 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1382 @cindex designator lists
1383 You can also write a series of @samp{.@var{fieldname}} and
1384 @samp{[@var{index}]} designators before an @samp{=} to specify a
1385 nested subobject to initialize; the list is taken relative to the
1386 subobject corresponding to the closest surrounding brace pair. For
1387 example, with the @samp{struct point} declaration above:
1390 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1394 If the same field is initialized multiple times, it will have value from
1395 the last initialization. If any such overridden initialization has
1396 side-effect, it is unspecified whether the side-effect happens or not.
1397 Currently, GCC will discard them and issue a warning.
1400 @section Case Ranges
1402 @cindex ranges in case statements
1404 You can specify a range of consecutive values in a single @code{case} label,
1408 case @var{low} ... @var{high}:
1412 This has the same effect as the proper number of individual @code{case}
1413 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1415 This feature is especially useful for ranges of ASCII character codes:
1421 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1422 it may be parsed wrong when you use it with integer values. For example,
1437 @section Cast to a Union Type
1438 @cindex cast to a union
1439 @cindex union, casting to a
1441 A cast to union type is similar to other casts, except that the type
1442 specified is a union type. You can specify the type either with
1443 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1444 a constructor though, not a cast, and hence does not yield an lvalue like
1445 normal casts. (@xref{Compound Literals}.)
1447 The types that may be cast to the union type are those of the members
1448 of the union. Thus, given the following union and variables:
1451 union foo @{ int i; double d; @};
1457 both @code{x} and @code{y} can be cast to type @code{union foo}.
1459 Using the cast as the right-hand side of an assignment to a variable of
1460 union type is equivalent to storing in a member of the union:
1465 u = (union foo) x @equiv{} u.i = x
1466 u = (union foo) y @equiv{} u.d = y
1469 You can also use the union cast as a function argument:
1472 void hack (union foo);
1474 hack ((union foo) x);
1477 @node Mixed Declarations
1478 @section Mixed Declarations and Code
1479 @cindex mixed declarations and code
1480 @cindex declarations, mixed with code
1481 @cindex code, mixed with declarations
1483 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1484 within compound statements. As an extension, GCC also allows this in
1485 C89 mode. For example, you could do:
1494 Each identifier is visible from where it is declared until the end of
1495 the enclosing block.
1497 @node Function Attributes
1498 @section Declaring Attributes of Functions
1499 @cindex function attributes
1500 @cindex declaring attributes of functions
1501 @cindex functions that never return
1502 @cindex functions that have no side effects
1503 @cindex functions in arbitrary sections
1504 @cindex functions that behave like malloc
1505 @cindex @code{volatile} applied to function
1506 @cindex @code{const} applied to function
1507 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1508 @cindex functions with non-null pointer arguments
1509 @cindex functions that are passed arguments in registers on the 386
1510 @cindex functions that pop the argument stack on the 386
1511 @cindex functions that do not pop the argument stack on the 386
1513 In GNU C, you declare certain things about functions called in your program
1514 which help the compiler optimize function calls and check your code more
1517 The keyword @code{__attribute__} allows you to specify special
1518 attributes when making a declaration. This keyword is followed by an
1519 attribute specification inside double parentheses. The following
1520 attributes are currently defined for functions on all targets:
1521 @code{noreturn}, @code{noinline}, @code{always_inline},
1522 @code{pure}, @code{const}, @code{nothrow}, @code{sentinel},
1523 @code{format}, @code{format_arg}, @code{no_instrument_function},
1524 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1525 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
1526 @code{alias}, @code{warn_unused_result} and @code{nonnull}. Several other
1527 attributes are defined for functions on particular target systems. Other
1528 attributes, including @code{section} are supported for variables declarations
1529 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1531 You may also specify attributes with @samp{__} preceding and following
1532 each keyword. This allows you to use them in header files without
1533 being concerned about a possible macro of the same name. For example,
1534 you may use @code{__noreturn__} instead of @code{noreturn}.
1536 @xref{Attribute Syntax}, for details of the exact syntax for using
1540 @c Keep this table alphabetized by attribute name. Treat _ as space.
1542 @item alias ("@var{target}")
1543 @cindex @code{alias} attribute
1544 The @code{alias} attribute causes the declaration to be emitted as an
1545 alias for another symbol, which must be specified. For instance,
1548 void __f () @{ /* @r{Do something.} */; @}
1549 void f () __attribute__ ((weak, alias ("__f")));
1552 declares @samp{f} to be a weak alias for @samp{__f}. In C++, the
1553 mangled name for the target must be used. It is an error if @samp{__f}
1554 is not defined in the same translation unit.
1556 Not all target machines support this attribute.
1559 @cindex @code{always_inline} function attribute
1560 Generally, functions are not inlined unless optimization is specified.
1561 For functions declared inline, this attribute inlines the function even
1562 if no optimization level was specified.
1565 @cindex functions that do pop the argument stack on the 386
1567 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1568 assume that the calling function will pop off the stack space used to
1569 pass arguments. This is
1570 useful to override the effects of the @option{-mrtd} switch.
1573 @cindex @code{const} function attribute
1574 Many functions do not examine any values except their arguments, and
1575 have no effects except the return value. Basically this is just slightly
1576 more strict class than the @code{pure} attribute below, since function is not
1577 allowed to read global memory.
1579 @cindex pointer arguments
1580 Note that a function that has pointer arguments and examines the data
1581 pointed to must @emph{not} be declared @code{const}. Likewise, a
1582 function that calls a non-@code{const} function usually must not be
1583 @code{const}. It does not make sense for a @code{const} function to
1586 The attribute @code{const} is not implemented in GCC versions earlier
1587 than 2.5. An alternative way to declare that a function has no side
1588 effects, which works in the current version and in some older versions,
1592 typedef int intfn ();
1594 extern const intfn square;
1597 This approach does not work in GNU C++ from 2.6.0 on, since the language
1598 specifies that the @samp{const} must be attached to the return value.
1602 @cindex @code{constructor} function attribute
1603 @cindex @code{destructor} function attribute
1604 The @code{constructor} attribute causes the function to be called
1605 automatically before execution enters @code{main ()}. Similarly, the
1606 @code{destructor} attribute causes the function to be called
1607 automatically after @code{main ()} has completed or @code{exit ()} has
1608 been called. Functions with these attributes are useful for
1609 initializing data that will be used implicitly during the execution of
1612 These attributes are not currently implemented for Objective-C@.
1615 @cindex @code{deprecated} attribute.
1616 The @code{deprecated} attribute results in a warning if the function
1617 is used anywhere in the source file. This is useful when identifying
1618 functions that are expected to be removed in a future version of a
1619 program. The warning also includes the location of the declaration
1620 of the deprecated function, to enable users to easily find further
1621 information about why the function is deprecated, or what they should
1622 do instead. Note that the warnings only occurs for uses:
1625 int old_fn () __attribute__ ((deprecated));
1627 int (*fn_ptr)() = old_fn;
1630 results in a warning on line 3 but not line 2.
1632 The @code{deprecated} attribute can also be used for variables and
1633 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
1636 @cindex @code{__declspec(dllexport)}
1637 On Microsoft Windows targets and Symbian OS targets the
1638 @code{dllexport} attribute causes the compiler to provide a global
1639 pointer to a pointer in a DLL, so that it can be referenced with the
1640 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
1641 name is formed by combining @code{_imp__} and the function or variable
1644 You can use @code{__declspec(dllexport)} as a synonym for
1645 @code{__attribute__ ((dllexport))} for compatibility with other
1648 On systems that support the @code{visibility} attribute, this
1649 attribute also implies ``default'' visibility, unless a
1650 @code{visibility} attribute is explicitly specified. You should avoid
1651 the use of @code{dllexport} with ``hidden'' or ``internal''
1652 visibility; in the future GCC may issue an error for those cases.
1654 Currently, the @code{dllexport} attribute is ignored for inlined
1655 functions, unless the @option{-fkeep-inline-functions} flag has been
1656 used. The attribute is also ignored for undefined symbols.
1658 When applied to C++ classes, the attribute marks defined non-inlined
1659 member functions and static data members as exports. Static consts
1660 initialized in-class are not marked unless they are also defined
1663 For Microsoft Windows targets there are alternative methods for
1664 including the symbol in the DLL's export table such as using a
1665 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
1666 the @option{--export-all} linker flag.
1669 @cindex @code{__declspec(dllimport)}
1670 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
1671 attribute causes the compiler to reference a function or variable via
1672 a global pointer to a pointer that is set up by the DLL exporting the
1673 symbol. The attribute implies @code{extern} storage. On Microsoft
1674 Windows targets, the pointer name is formed by combining @code{_imp__}
1675 and the function or variable name.
1677 You can use @code{__declspec(dllimport)} as a synonym for
1678 @code{__attribute__ ((dllimport))} for compatibility with other
1681 Currently, the attribute is ignored for inlined functions. If the
1682 attribute is applied to a symbol @emph{definition}, an error is reported.
1683 If a symbol previously declared @code{dllimport} is later defined, the
1684 attribute is ignored in subsequent references, and a warning is emitted.
1685 The attribute is also overridden by a subsequent declaration as
1688 When applied to C++ classes, the attribute marks non-inlined
1689 member functions and static data members as imports. However, the
1690 attribute is ignored for virtual methods to allow creation of vtables
1693 On the SH Symbian OS target the @code{dllimport} attribute also has
1694 another affect---it can cause the vtable and run-time type information
1695 for a class to be exported. This happens when the class has a
1696 dllimport'ed constructor or a non-inline, non-pure virtual function
1697 and, for either of those two conditions, the class also has a inline
1698 constructor or destructor and has a key function that is defined in
1699 the current translation unit.
1701 For Microsoft Windows based targets the use of the @code{dllimport}
1702 attribute on functions is not necessary, but provides a small
1703 performance benefit by eliminating a thunk in the DLL@. The use of the
1704 @code{dllimport} attribute on imported variables was required on older
1705 versions of the GNU linker, but can now be avoided by passing the
1706 @option{--enable-auto-import} switch to the GNU linker. As with
1707 functions, using the attribute for a variable eliminates a thunk in
1710 One drawback to using this attribute is that a pointer to a function
1711 or variable marked as @code{dllimport} cannot be used as a constant
1712 address. On Microsoft Windows targets, the attribute can be disabled
1713 for functions by setting the @option{-mnop-fun-dllimport} flag.
1716 @cindex eight bit data on the H8/300, H8/300H, and H8S
1717 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1718 variable should be placed into the eight bit data section.
1719 The compiler will generate more efficient code for certain operations
1720 on data in the eight bit data area. Note the eight bit data area is limited to
1723 You must use GAS and GLD from GNU binutils version 2.7 or later for
1724 this attribute to work correctly.
1726 @item exception_handler
1727 @cindex exception handler functions on the Blackfin processor
1728 Use this attribute on the Blackfin to indicate that the specified function
1729 is an exception handler. The compiler will generate function entry and
1730 exit sequences suitable for use in an exception handler when this
1731 attribute is present.
1734 @cindex functions which handle memory bank switching
1735 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
1736 use a calling convention that takes care of switching memory banks when
1737 entering and leaving a function. This calling convention is also the
1738 default when using the @option{-mlong-calls} option.
1740 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
1741 to call and return from a function.
1743 On 68HC11 the compiler will generate a sequence of instructions
1744 to invoke a board-specific routine to switch the memory bank and call the
1745 real function. The board-specific routine simulates a @code{call}.
1746 At the end of a function, it will jump to a board-specific routine
1747 instead of using @code{rts}. The board-specific return routine simulates
1751 @cindex functions that pop the argument stack on the 386
1752 On the Intel 386, the @code{fastcall} attribute causes the compiler to
1753 pass the first two arguments in the registers ECX and EDX@. Subsequent
1754 arguments are passed on the stack. The called function will pop the
1755 arguments off the stack. If the number of arguments is variable all
1756 arguments are pushed on the stack.
1758 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
1759 @cindex @code{format} function attribute
1761 The @code{format} attribute specifies that a function takes @code{printf},
1762 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
1763 should be type-checked against a format string. For example, the
1768 my_printf (void *my_object, const char *my_format, ...)
1769 __attribute__ ((format (printf, 2, 3)));
1773 causes the compiler to check the arguments in calls to @code{my_printf}
1774 for consistency with the @code{printf} style format string argument
1777 The parameter @var{archetype} determines how the format string is
1778 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
1779 or @code{strfmon}. (You can also use @code{__printf__},
1780 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
1781 parameter @var{string-index} specifies which argument is the format
1782 string argument (starting from 1), while @var{first-to-check} is the
1783 number of the first argument to check against the format string. For
1784 functions where the arguments are not available to be checked (such as
1785 @code{vprintf}), specify the third parameter as zero. In this case the
1786 compiler only checks the format string for consistency. For
1787 @code{strftime} formats, the third parameter is required to be zero.
1788 Since non-static C++ methods have an implicit @code{this} argument, the
1789 arguments of such methods should be counted from two, not one, when
1790 giving values for @var{string-index} and @var{first-to-check}.
1792 In the example above, the format string (@code{my_format}) is the second
1793 argument of the function @code{my_print}, and the arguments to check
1794 start with the third argument, so the correct parameters for the format
1795 attribute are 2 and 3.
1797 @opindex ffreestanding
1798 @opindex fno-builtin
1799 The @code{format} attribute allows you to identify your own functions
1800 which take format strings as arguments, so that GCC can check the
1801 calls to these functions for errors. The compiler always (unless
1802 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
1803 for the standard library functions @code{printf}, @code{fprintf},
1804 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
1805 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
1806 warnings are requested (using @option{-Wformat}), so there is no need to
1807 modify the header file @file{stdio.h}. In C99 mode, the functions
1808 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
1809 @code{vsscanf} are also checked. Except in strictly conforming C
1810 standard modes, the X/Open function @code{strfmon} is also checked as
1811 are @code{printf_unlocked} and @code{fprintf_unlocked}.
1812 @xref{C Dialect Options,,Options Controlling C Dialect}.
1814 The target may provide additional types of format checks.
1815 @xref{Target Format Checks,,Format Checks Specific to Particular
1818 @item format_arg (@var{string-index})
1819 @cindex @code{format_arg} function attribute
1820 @opindex Wformat-nonliteral
1821 The @code{format_arg} attribute specifies that a function takes a format
1822 string for a @code{printf}, @code{scanf}, @code{strftime} or
1823 @code{strfmon} style function and modifies it (for example, to translate
1824 it into another language), so the result can be passed to a
1825 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
1826 function (with the remaining arguments to the format function the same
1827 as they would have been for the unmodified string). For example, the
1832 my_dgettext (char *my_domain, const char *my_format)
1833 __attribute__ ((format_arg (2)));
1837 causes the compiler to check the arguments in calls to a @code{printf},
1838 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
1839 format string argument is a call to the @code{my_dgettext} function, for
1840 consistency with the format string argument @code{my_format}. If the
1841 @code{format_arg} attribute had not been specified, all the compiler
1842 could tell in such calls to format functions would be that the format
1843 string argument is not constant; this would generate a warning when
1844 @option{-Wformat-nonliteral} is used, but the calls could not be checked
1845 without the attribute.
1847 The parameter @var{string-index} specifies which argument is the format
1848 string argument (starting from one). Since non-static C++ methods have
1849 an implicit @code{this} argument, the arguments of such methods should
1850 be counted from two.
1852 The @code{format-arg} attribute allows you to identify your own
1853 functions which modify format strings, so that GCC can check the
1854 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
1855 type function whose operands are a call to one of your own function.
1856 The compiler always treats @code{gettext}, @code{dgettext}, and
1857 @code{dcgettext} in this manner except when strict ISO C support is
1858 requested by @option{-ansi} or an appropriate @option{-std} option, or
1859 @option{-ffreestanding} or @option{-fno-builtin}
1860 is used. @xref{C Dialect Options,,Options
1861 Controlling C Dialect}.
1863 @item function_vector
1864 @cindex calling functions through the function vector on the H8/300 processors
1865 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1866 function should be called through the function vector. Calling a
1867 function through the function vector will reduce code size, however;
1868 the function vector has a limited size (maximum 128 entries on the H8/300
1869 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
1871 You must use GAS and GLD from GNU binutils version 2.7 or later for
1872 this attribute to work correctly.
1875 @cindex interrupt handler functions
1876 Use this attribute on the ARM, AVR, C4x, M32R/D and Xstormy16 ports to indicate
1877 that the specified function is an interrupt handler. The compiler will
1878 generate function entry and exit sequences suitable for use in an
1879 interrupt handler when this attribute is present.
1881 Note, interrupt handlers for the Blackfin, m68k, H8/300, H8/300H, H8S, and
1882 SH processors can be specified via the @code{interrupt_handler} attribute.
1884 Note, on the AVR, interrupts will be enabled inside the function.
1886 Note, for the ARM, you can specify the kind of interrupt to be handled by
1887 adding an optional parameter to the interrupt attribute like this:
1890 void f () __attribute__ ((interrupt ("IRQ")));
1893 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
1895 @item interrupt_handler
1896 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
1897 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
1898 indicate that the specified function is an interrupt handler. The compiler
1899 will generate function entry and exit sequences suitable for use in an
1900 interrupt handler when this attribute is present.
1903 @cindex User stack pointer in interrupts on the Blackfin
1904 When used together with @code{interrupt_handler}, @code{exception_handler}
1905 or @code{nmi_handler}, code will be generated to load the stack pointer
1906 from the USP register in the function prologue.
1908 @item long_call/short_call
1909 @cindex indirect calls on ARM
1910 This attribute specifies how a particular function is called on
1911 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
1912 command line switch and @code{#pragma long_calls} settings. The
1913 @code{long_call} attribute causes the compiler to always call the
1914 function by first loading its address into a register and then using the
1915 contents of that register. The @code{short_call} attribute always places
1916 the offset to the function from the call site into the @samp{BL}
1917 instruction directly.
1919 @item longcall/shortcall
1920 @cindex functions called via pointer on the RS/6000 and PowerPC
1921 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute causes
1922 the compiler to always call this function via a pointer, just as it would if
1923 the @option{-mlongcall} option had been specified. The @code{shortcall}
1924 attribute causes the compiler not to do this. These attributes override
1925 both the @option{-mlongcall} switch and, on the RS/6000 and PowerPC, the
1926 @code{#pragma longcall} setting.
1928 @xref{RS/6000 and PowerPC Options}, for more information on whether long
1929 calls are necessary.
1932 @cindex @code{malloc} attribute
1933 The @code{malloc} attribute is used to tell the compiler that a function
1934 may be treated as if any non-@code{NULL} pointer it returns cannot
1935 alias any other pointer valid when the function returns.
1936 This will often improve optimization.
1937 Standard functions with this property include @code{malloc} and
1938 @code{calloc}. @code{realloc}-like functions have this property as
1939 long as the old pointer is never referred to (including comparing it
1940 to the new pointer) after the function returns a non-@code{NULL}
1943 @item model (@var{model-name})
1944 @cindex function addressability on the M32R/D
1945 @cindex variable addressability on the IA-64
1947 On the M32R/D, use this attribute to set the addressability of an
1948 object, and of the code generated for a function. The identifier
1949 @var{model-name} is one of @code{small}, @code{medium}, or
1950 @code{large}, representing each of the code models.
1952 Small model objects live in the lower 16MB of memory (so that their
1953 addresses can be loaded with the @code{ld24} instruction), and are
1954 callable with the @code{bl} instruction.
1956 Medium model objects may live anywhere in the 32-bit address space (the
1957 compiler will generate @code{seth/add3} instructions to load their addresses),
1958 and are callable with the @code{bl} instruction.
1960 Large model objects may live anywhere in the 32-bit address space (the
1961 compiler will generate @code{seth/add3} instructions to load their addresses),
1962 and may not be reachable with the @code{bl} instruction (the compiler will
1963 generate the much slower @code{seth/add3/jl} instruction sequence).
1965 On IA-64, use this attribute to set the addressability of an object.
1966 At present, the only supported identifier for @var{model-name} is
1967 @code{small}, indicating addressability via ``small'' (22-bit)
1968 addresses (so that their addresses can be loaded with the @code{addl}
1969 instruction). Caveat: such addressing is by definition not position
1970 independent and hence this attribute must not be used for objects
1971 defined by shared libraries.
1974 @cindex function without a prologue/epilogue code
1975 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
1976 specified function does not need prologue/epilogue sequences generated by
1977 the compiler. It is up to the programmer to provide these sequences.
1980 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
1981 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
1982 use the normal calling convention based on @code{jsr} and @code{rts}.
1983 This attribute can be used to cancel the effect of the @option{-mlong-calls}
1987 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
1988 Use this attribute together with @code{interrupt_handler},
1989 @code{exception_handler} or @code{nmi_handler} to indicate that the function
1990 entry code should enable nested interrupts or exceptions.
1993 @cindex NMI handler functions on the Blackfin processor
1994 Use this attribute on the Blackfin to indicate that the specified function
1995 is an NMI handler. The compiler will generate function entry and
1996 exit sequences suitable for use in an NMI handler when this
1997 attribute is present.
1999 @item no_instrument_function
2000 @cindex @code{no_instrument_function} function attribute
2001 @opindex finstrument-functions
2002 If @option{-finstrument-functions} is given, profiling function calls will
2003 be generated at entry and exit of most user-compiled functions.
2004 Functions with this attribute will not be so instrumented.
2007 @cindex @code{noinline} function attribute
2008 This function attribute prevents a function from being considered for
2011 @item nonnull (@var{arg-index}, @dots{})
2012 @cindex @code{nonnull} function attribute
2013 The @code{nonnull} attribute specifies that some function parameters should
2014 be non-null pointers. For instance, the declaration:
2018 my_memcpy (void *dest, const void *src, size_t len)
2019 __attribute__((nonnull (1, 2)));
2023 causes the compiler to check that, in calls to @code{my_memcpy},
2024 arguments @var{dest} and @var{src} are non-null. If the compiler
2025 determines that a null pointer is passed in an argument slot marked
2026 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2027 is issued. The compiler may also choose to make optimizations based
2028 on the knowledge that certain function arguments will not be null.
2030 If no argument index list is given to the @code{nonnull} attribute,
2031 all pointer arguments are marked as non-null. To illustrate, the
2032 following declaration is equivalent to the previous example:
2036 my_memcpy (void *dest, const void *src, size_t len)
2037 __attribute__((nonnull));
2041 @cindex @code{noreturn} function attribute
2042 A few standard library functions, such as @code{abort} and @code{exit},
2043 cannot return. GCC knows this automatically. Some programs define
2044 their own functions that never return. You can declare them
2045 @code{noreturn} to tell the compiler this fact. For example,
2049 void fatal () __attribute__ ((noreturn));
2052 fatal (/* @r{@dots{}} */)
2054 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2060 The @code{noreturn} keyword tells the compiler to assume that
2061 @code{fatal} cannot return. It can then optimize without regard to what
2062 would happen if @code{fatal} ever did return. This makes slightly
2063 better code. More importantly, it helps avoid spurious warnings of
2064 uninitialized variables.
2066 The @code{noreturn} keyword does not affect the exceptional path when that
2067 applies: a @code{noreturn}-marked function may still return to the caller
2068 by throwing an exception or calling @code{longjmp}.
2070 Do not assume that registers saved by the calling function are
2071 restored before calling the @code{noreturn} function.
2073 It does not make sense for a @code{noreturn} function to have a return
2074 type other than @code{void}.
2076 The attribute @code{noreturn} is not implemented in GCC versions
2077 earlier than 2.5. An alternative way to declare that a function does
2078 not return, which works in the current version and in some older
2079 versions, is as follows:
2082 typedef void voidfn ();
2084 volatile voidfn fatal;
2087 This approach does not work in GNU C++.
2090 @cindex @code{nothrow} function attribute
2091 The @code{nothrow} attribute is used to inform the compiler that a
2092 function cannot throw an exception. For example, most functions in
2093 the standard C library can be guaranteed not to throw an exception
2094 with the notable exceptions of @code{qsort} and @code{bsearch} that
2095 take function pointer arguments. The @code{nothrow} attribute is not
2096 implemented in GCC versions earlier than 3.3.
2099 @cindex @code{pure} function attribute
2100 Many functions have no effects except the return value and their
2101 return value depends only on the parameters and/or global variables.
2102 Such a function can be subject
2103 to common subexpression elimination and loop optimization just as an
2104 arithmetic operator would be. These functions should be declared
2105 with the attribute @code{pure}. For example,
2108 int square (int) __attribute__ ((pure));
2112 says that the hypothetical function @code{square} is safe to call
2113 fewer times than the program says.
2115 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2116 Interesting non-pure functions are functions with infinite loops or those
2117 depending on volatile memory or other system resource, that may change between
2118 two consecutive calls (such as @code{feof} in a multithreading environment).
2120 The attribute @code{pure} is not implemented in GCC versions earlier
2123 @item regparm (@var{number})
2124 @cindex @code{regparm} attribute
2125 @cindex functions that are passed arguments in registers on the 386
2126 On the Intel 386, the @code{regparm} attribute causes the compiler to
2127 pass up to @var{number} integer arguments in registers EAX,
2128 EDX, and ECX instead of on the stack. Functions that take a
2129 variable number of arguments will continue to be passed all of their
2130 arguments on the stack.
2132 Beware that on some ELF systems this attribute is unsuitable for
2133 global functions in shared libraries with lazy binding (which is the
2134 default). Lazy binding will send the first call via resolving code in
2135 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2136 per the standard calling conventions. Solaris 8 is affected by this.
2137 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2138 safe since the loaders there save all registers. (Lazy binding can be
2139 disabled with the linker or the loader if desired, to avoid the
2143 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2144 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2145 all registers except the stack pointer should be saved in the prologue
2146 regardless of whether they are used or not.
2148 @item section ("@var{section-name}")
2149 @cindex @code{section} function attribute
2150 Normally, the compiler places the code it generates in the @code{text} section.
2151 Sometimes, however, you need additional sections, or you need certain
2152 particular functions to appear in special sections. The @code{section}
2153 attribute specifies that a function lives in a particular section.
2154 For example, the declaration:
2157 extern void foobar (void) __attribute__ ((section ("bar")));
2161 puts the function @code{foobar} in the @code{bar} section.
2163 Some file formats do not support arbitrary sections so the @code{section}
2164 attribute is not available on all platforms.
2165 If you need to map the entire contents of a module to a particular
2166 section, consider using the facilities of the linker instead.
2169 @cindex @code{sentinel} function attribute
2170 This function attribute ensures that a parameter in a function call is
2171 an explicit @code{NULL}. The attribute is only valid on variadic
2172 functions. By default, the sentinel is located at position zero, the
2173 last parameter of the function call. If an optional integer position
2174 argument P is supplied to the attribute, the sentinel must be located at
2175 position P counting backwards from the end of the argument list.
2178 __attribute__ ((sentinel))
2180 __attribute__ ((sentinel(0)))
2183 The attribute is automatically set with a position of 0 for the built-in
2184 functions @code{execl} and @code{execlp}. The built-in function
2185 @code{execle} has the attribute set with a position of 1.
2187 A valid @code{NULL} in this context is defined as zero with any pointer
2188 type. If your system defines the @code{NULL} macro with an integer type
2189 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2190 with a copy that redefines NULL appropriately.
2192 The warnings for missing or incorrect sentinels are enabled with
2196 See long_call/short_call.
2199 See longcall/shortcall.
2202 @cindex signal handler functions on the AVR processors
2203 Use this attribute on the AVR to indicate that the specified
2204 function is a signal handler. The compiler will generate function
2205 entry and exit sequences suitable for use in a signal handler when this
2206 attribute is present. Interrupts will be disabled inside the function.
2209 Use this attribute on the SH to indicate an @code{interrupt_handler}
2210 function should switch to an alternate stack. It expects a string
2211 argument that names a global variable holding the address of the
2216 void f () __attribute__ ((interrupt_handler,
2217 sp_switch ("alt_stack")));
2221 @cindex functions that pop the argument stack on the 386
2222 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2223 assume that the called function will pop off the stack space used to
2224 pass arguments, unless it takes a variable number of arguments.
2227 @cindex tiny data section on the H8/300H and H8S
2228 Use this attribute on the H8/300H and H8S to indicate that the specified
2229 variable should be placed into the tiny data section.
2230 The compiler will generate more efficient code for loads and stores
2231 on data in the tiny data section. Note the tiny data area is limited to
2232 slightly under 32kbytes of data.
2235 Use this attribute on the SH for an @code{interrupt_handler} to return using
2236 @code{trapa} instead of @code{rte}. This attribute expects an integer
2237 argument specifying the trap number to be used.
2240 @cindex @code{unused} attribute.
2241 This attribute, attached to a function, means that the function is meant
2242 to be possibly unused. GCC will not produce a warning for this
2246 @cindex @code{used} attribute.
2247 This attribute, attached to a function, means that code must be emitted
2248 for the function even if it appears that the function is not referenced.
2249 This is useful, for example, when the function is referenced only in
2252 @item visibility ("@var{visibility_type}")
2253 @cindex @code{visibility} attribute
2254 The @code{visibility} attribute on ELF targets causes the declaration
2255 to be emitted with default, hidden, protected or internal visibility.
2258 void __attribute__ ((visibility ("protected")))
2259 f () @{ /* @r{Do something.} */; @}
2260 int i __attribute__ ((visibility ("hidden")));
2263 See the ELF gABI for complete details, but the short story is:
2266 @c keep this list of visibilities in alphabetical order.
2269 Default visibility is the normal case for ELF@. This value is
2270 available for the visibility attribute to override other options
2271 that may change the assumed visibility of symbols.
2274 Hidden visibility indicates that the symbol will not be placed into
2275 the dynamic symbol table, so no other @dfn{module} (executable or
2276 shared library) can reference it directly.
2279 Internal visibility is like hidden visibility, but with additional
2280 processor specific semantics. Unless otherwise specified by the psABI,
2281 GCC defines internal visibility to mean that the function is @emph{never}
2282 called from another module. Note that hidden symbols, while they cannot
2283 be referenced directly by other modules, can be referenced indirectly via
2284 function pointers. By indicating that a symbol cannot be called from
2285 outside the module, GCC may for instance omit the load of a PIC register
2286 since it is known that the calling function loaded the correct value.
2289 Protected visibility indicates that the symbol will be placed in the
2290 dynamic symbol table, but that references within the defining module
2291 will bind to the local symbol. That is, the symbol cannot be overridden
2296 Not all ELF targets support this attribute.
2298 @item warn_unused_result
2299 @cindex @code{warn_unused_result} attribute
2300 The @code{warn_unused_result} attribute causes a warning to be emitted
2301 if a caller of the function with this attribute does not use its
2302 return value. This is useful for functions where not checking
2303 the result is either a security problem or always a bug, such as
2307 int fn () __attribute__ ((warn_unused_result));
2310 if (fn () < 0) return -1;
2316 results in warning on line 5.
2319 @cindex @code{weak} attribute
2320 The @code{weak} attribute causes the declaration to be emitted as a weak
2321 symbol rather than a global. This is primarily useful in defining
2322 library functions which can be overridden in user code, though it can
2323 also be used with non-function declarations. Weak symbols are supported
2324 for ELF targets, and also for a.out targets when using the GNU assembler
2329 You can specify multiple attributes in a declaration by separating them
2330 by commas within the double parentheses or by immediately following an
2331 attribute declaration with another attribute declaration.
2333 @cindex @code{#pragma}, reason for not using
2334 @cindex pragma, reason for not using
2335 Some people object to the @code{__attribute__} feature, suggesting that
2336 ISO C's @code{#pragma} should be used instead. At the time
2337 @code{__attribute__} was designed, there were two reasons for not doing
2342 It is impossible to generate @code{#pragma} commands from a macro.
2345 There is no telling what the same @code{#pragma} might mean in another
2349 These two reasons applied to almost any application that might have been
2350 proposed for @code{#pragma}. It was basically a mistake to use
2351 @code{#pragma} for @emph{anything}.
2353 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2354 to be generated from macros. In addition, a @code{#pragma GCC}
2355 namespace is now in use for GCC-specific pragmas. However, it has been
2356 found convenient to use @code{__attribute__} to achieve a natural
2357 attachment of attributes to their corresponding declarations, whereas
2358 @code{#pragma GCC} is of use for constructs that do not naturally form
2359 part of the grammar. @xref{Other Directives,,Miscellaneous
2360 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2362 @node Attribute Syntax
2363 @section Attribute Syntax
2364 @cindex attribute syntax
2366 This section describes the syntax with which @code{__attribute__} may be
2367 used, and the constructs to which attribute specifiers bind, for the C
2368 language. Some details may vary for C++ and Objective-C@. Because of
2369 infelicities in the grammar for attributes, some forms described here
2370 may not be successfully parsed in all cases.
2372 There are some problems with the semantics of attributes in C++. For
2373 example, there are no manglings for attributes, although they may affect
2374 code generation, so problems may arise when attributed types are used in
2375 conjunction with templates or overloading. Similarly, @code{typeid}
2376 does not distinguish between types with different attributes. Support
2377 for attributes in C++ may be restricted in future to attributes on
2378 declarations only, but not on nested declarators.
2380 @xref{Function Attributes}, for details of the semantics of attributes
2381 applying to functions. @xref{Variable Attributes}, for details of the
2382 semantics of attributes applying to variables. @xref{Type Attributes},
2383 for details of the semantics of attributes applying to structure, union
2384 and enumerated types.
2386 An @dfn{attribute specifier} is of the form
2387 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2388 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2389 each attribute is one of the following:
2393 Empty. Empty attributes are ignored.
2396 A word (which may be an identifier such as @code{unused}, or a reserved
2397 word such as @code{const}).
2400 A word, followed by, in parentheses, parameters for the attribute.
2401 These parameters take one of the following forms:
2405 An identifier. For example, @code{mode} attributes use this form.
2408 An identifier followed by a comma and a non-empty comma-separated list
2409 of expressions. For example, @code{format} attributes use this form.
2412 A possibly empty comma-separated list of expressions. For example,
2413 @code{format_arg} attributes use this form with the list being a single
2414 integer constant expression, and @code{alias} attributes use this form
2415 with the list being a single string constant.
2419 An @dfn{attribute specifier list} is a sequence of one or more attribute
2420 specifiers, not separated by any other tokens.
2422 In GNU C, an attribute specifier list may appear after the colon following a
2423 label, other than a @code{case} or @code{default} label. The only
2424 attribute it makes sense to use after a label is @code{unused}. This
2425 feature is intended for code generated by programs which contains labels
2426 that may be unused but which is compiled with @option{-Wall}. It would
2427 not normally be appropriate to use in it human-written code, though it
2428 could be useful in cases where the code that jumps to the label is
2429 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2430 such placement of attribute lists, as it is permissible for a
2431 declaration, which could begin with an attribute list, to be labelled in
2432 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2433 does not arise there.
2435 An attribute specifier list may appear as part of a @code{struct},
2436 @code{union} or @code{enum} specifier. It may go either immediately
2437 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2438 the closing brace. It is ignored if the content of the structure, union
2439 or enumerated type is not defined in the specifier in which the
2440 attribute specifier list is used---that is, in usages such as
2441 @code{struct __attribute__((foo)) bar} with no following opening brace.
2442 Where attribute specifiers follow the closing brace, they are considered
2443 to relate to the structure, union or enumerated type defined, not to any
2444 enclosing declaration the type specifier appears in, and the type
2445 defined is not complete until after the attribute specifiers.
2446 @c Otherwise, there would be the following problems: a shift/reduce
2447 @c conflict between attributes binding the struct/union/enum and
2448 @c binding to the list of specifiers/qualifiers; and "aligned"
2449 @c attributes could use sizeof for the structure, but the size could be
2450 @c changed later by "packed" attributes.
2452 Otherwise, an attribute specifier appears as part of a declaration,
2453 counting declarations of unnamed parameters and type names, and relates
2454 to that declaration (which may be nested in another declaration, for
2455 example in the case of a parameter declaration), or to a particular declarator
2456 within a declaration. Where an
2457 attribute specifier is applied to a parameter declared as a function or
2458 an array, it should apply to the function or array rather than the
2459 pointer to which the parameter is implicitly converted, but this is not
2460 yet correctly implemented.
2462 Any list of specifiers and qualifiers at the start of a declaration may
2463 contain attribute specifiers, whether or not such a list may in that
2464 context contain storage class specifiers. (Some attributes, however,
2465 are essentially in the nature of storage class specifiers, and only make
2466 sense where storage class specifiers may be used; for example,
2467 @code{section}.) There is one necessary limitation to this syntax: the
2468 first old-style parameter declaration in a function definition cannot
2469 begin with an attribute specifier, because such an attribute applies to
2470 the function instead by syntax described below (which, however, is not
2471 yet implemented in this case). In some other cases, attribute
2472 specifiers are permitted by this grammar but not yet supported by the
2473 compiler. All attribute specifiers in this place relate to the
2474 declaration as a whole. In the obsolescent usage where a type of
2475 @code{int} is implied by the absence of type specifiers, such a list of
2476 specifiers and qualifiers may be an attribute specifier list with no
2477 other specifiers or qualifiers.
2479 At present, the first parameter in a function prototype must have some
2480 type specifier which is not an attribute specifier; this resolves an
2481 ambiguity in the interpretation of @code{void f(int
2482 (__attribute__((foo)) x))}, but is subject to change. At present, if
2483 the parentheses of a function declarator contain only attributes then
2484 those attributes are ignored, rather than yielding an error or warning
2485 or implying a single parameter of type int, but this is subject to
2488 An attribute specifier list may appear immediately before a declarator
2489 (other than the first) in a comma-separated list of declarators in a
2490 declaration of more than one identifier using a single list of
2491 specifiers and qualifiers. Such attribute specifiers apply
2492 only to the identifier before whose declarator they appear. For
2496 __attribute__((noreturn)) void d0 (void),
2497 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2502 the @code{noreturn} attribute applies to all the functions
2503 declared; the @code{format} attribute only applies to @code{d1}.
2505 An attribute specifier list may appear immediately before the comma,
2506 @code{=} or semicolon terminating the declaration of an identifier other
2507 than a function definition. At present, such attribute specifiers apply
2508 to the declared object or function, but in future they may attach to the
2509 outermost adjacent declarator. In simple cases there is no difference,
2510 but, for example, in
2513 void (****f)(void) __attribute__((noreturn));
2517 at present the @code{noreturn} attribute applies to @code{f}, which
2518 causes a warning since @code{f} is not a function, but in future it may
2519 apply to the function @code{****f}. The precise semantics of what
2520 attributes in such cases will apply to are not yet specified. Where an
2521 assembler name for an object or function is specified (@pxref{Asm
2522 Labels}), at present the attribute must follow the @code{asm}
2523 specification; in future, attributes before the @code{asm} specification
2524 may apply to the adjacent declarator, and those after it to the declared
2527 An attribute specifier list may, in future, be permitted to appear after
2528 the declarator in a function definition (before any old-style parameter
2529 declarations or the function body).
2531 Attribute specifiers may be mixed with type qualifiers appearing inside
2532 the @code{[]} of a parameter array declarator, in the C99 construct by
2533 which such qualifiers are applied to the pointer to which the array is
2534 implicitly converted. Such attribute specifiers apply to the pointer,
2535 not to the array, but at present this is not implemented and they are
2538 An attribute specifier list may appear at the start of a nested
2539 declarator. At present, there are some limitations in this usage: the
2540 attributes correctly apply to the declarator, but for most individual
2541 attributes the semantics this implies are not implemented.
2542 When attribute specifiers follow the @code{*} of a pointer
2543 declarator, they may be mixed with any type qualifiers present.
2544 The following describes the formal semantics of this syntax. It will make the
2545 most sense if you are familiar with the formal specification of
2546 declarators in the ISO C standard.
2548 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2549 D1}, where @code{T} contains declaration specifiers that specify a type
2550 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2551 contains an identifier @var{ident}. The type specified for @var{ident}
2552 for derived declarators whose type does not include an attribute
2553 specifier is as in the ISO C standard.
2555 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2556 and the declaration @code{T D} specifies the type
2557 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2558 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2559 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2561 If @code{D1} has the form @code{*
2562 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2563 declaration @code{T D} specifies the type
2564 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2565 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2566 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2572 void (__attribute__((noreturn)) ****f) (void);
2576 specifies the type ``pointer to pointer to pointer to pointer to
2577 non-returning function returning @code{void}''. As another example,
2580 char *__attribute__((aligned(8))) *f;
2584 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2585 Note again that this does not work with most attributes; for example,
2586 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2587 is not yet supported.
2589 For compatibility with existing code written for compiler versions that
2590 did not implement attributes on nested declarators, some laxity is
2591 allowed in the placing of attributes. If an attribute that only applies
2592 to types is applied to a declaration, it will be treated as applying to
2593 the type of that declaration. If an attribute that only applies to
2594 declarations is applied to the type of a declaration, it will be treated
2595 as applying to that declaration; and, for compatibility with code
2596 placing the attributes immediately before the identifier declared, such
2597 an attribute applied to a function return type will be treated as
2598 applying to the function type, and such an attribute applied to an array
2599 element type will be treated as applying to the array type. If an
2600 attribute that only applies to function types is applied to a
2601 pointer-to-function type, it will be treated as applying to the pointer
2602 target type; if such an attribute is applied to a function return type
2603 that is not a pointer-to-function type, it will be treated as applying
2604 to the function type.
2606 @node Function Prototypes
2607 @section Prototypes and Old-Style Function Definitions
2608 @cindex function prototype declarations
2609 @cindex old-style function definitions
2610 @cindex promotion of formal parameters
2612 GNU C extends ISO C to allow a function prototype to override a later
2613 old-style non-prototype definition. Consider the following example:
2616 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2623 /* @r{Prototype function declaration.} */
2624 int isroot P((uid_t));
2626 /* @r{Old-style function definition.} */
2628 isroot (x) /* @r{??? lossage here ???} */
2635 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2636 not allow this example, because subword arguments in old-style
2637 non-prototype definitions are promoted. Therefore in this example the
2638 function definition's argument is really an @code{int}, which does not
2639 match the prototype argument type of @code{short}.
2641 This restriction of ISO C makes it hard to write code that is portable
2642 to traditional C compilers, because the programmer does not know
2643 whether the @code{uid_t} type is @code{short}, @code{int}, or
2644 @code{long}. Therefore, in cases like these GNU C allows a prototype
2645 to override a later old-style definition. More precisely, in GNU C, a
2646 function prototype argument type overrides the argument type specified
2647 by a later old-style definition if the former type is the same as the
2648 latter type before promotion. Thus in GNU C the above example is
2649 equivalent to the following:
2662 GNU C++ does not support old-style function definitions, so this
2663 extension is irrelevant.
2666 @section C++ Style Comments
2668 @cindex C++ comments
2669 @cindex comments, C++ style
2671 In GNU C, you may use C++ style comments, which start with @samp{//} and
2672 continue until the end of the line. Many other C implementations allow
2673 such comments, and they are included in the 1999 C standard. However,
2674 C++ style comments are not recognized if you specify an @option{-std}
2675 option specifying a version of ISO C before C99, or @option{-ansi}
2676 (equivalent to @option{-std=c89}).
2679 @section Dollar Signs in Identifier Names
2681 @cindex dollar signs in identifier names
2682 @cindex identifier names, dollar signs in
2684 In GNU C, you may normally use dollar signs in identifier names.
2685 This is because many traditional C implementations allow such identifiers.
2686 However, dollar signs in identifiers are not supported on a few target
2687 machines, typically because the target assembler does not allow them.
2689 @node Character Escapes
2690 @section The Character @key{ESC} in Constants
2692 You can use the sequence @samp{\e} in a string or character constant to
2693 stand for the ASCII character @key{ESC}.
2696 @section Inquiring on Alignment of Types or Variables
2698 @cindex type alignment
2699 @cindex variable alignment
2701 The keyword @code{__alignof__} allows you to inquire about how an object
2702 is aligned, or the minimum alignment usually required by a type. Its
2703 syntax is just like @code{sizeof}.
2705 For example, if the target machine requires a @code{double} value to be
2706 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2707 This is true on many RISC machines. On more traditional machine
2708 designs, @code{__alignof__ (double)} is 4 or even 2.
2710 Some machines never actually require alignment; they allow reference to any
2711 data type even at an odd address. For these machines, @code{__alignof__}
2712 reports the @emph{recommended} alignment of a type.
2714 If the operand of @code{__alignof__} is an lvalue rather than a type,
2715 its value is the required alignment for its type, taking into account
2716 any minimum alignment specified with GCC's @code{__attribute__}
2717 extension (@pxref{Variable Attributes}). For example, after this
2721 struct foo @{ int x; char y; @} foo1;
2725 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2726 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2728 It is an error to ask for the alignment of an incomplete type.
2730 @node Variable Attributes
2731 @section Specifying Attributes of Variables
2732 @cindex attribute of variables
2733 @cindex variable attributes
2735 The keyword @code{__attribute__} allows you to specify special
2736 attributes of variables or structure fields. This keyword is followed
2737 by an attribute specification inside double parentheses. Some
2738 attributes are currently defined generically for variables.
2739 Other attributes are defined for variables on particular target
2740 systems. Other attributes are available for functions
2741 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
2742 Other front ends might define more attributes
2743 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2745 You may also specify attributes with @samp{__} preceding and following
2746 each keyword. This allows you to use them in header files without
2747 being concerned about a possible macro of the same name. For example,
2748 you may use @code{__aligned__} instead of @code{aligned}.
2750 @xref{Attribute Syntax}, for details of the exact syntax for using
2754 @cindex @code{aligned} attribute
2755 @item aligned (@var{alignment})
2756 This attribute specifies a minimum alignment for the variable or
2757 structure field, measured in bytes. For example, the declaration:
2760 int x __attribute__ ((aligned (16))) = 0;
2764 causes the compiler to allocate the global variable @code{x} on a
2765 16-byte boundary. On a 68040, this could be used in conjunction with
2766 an @code{asm} expression to access the @code{move16} instruction which
2767 requires 16-byte aligned operands.
2769 You can also specify the alignment of structure fields. For example, to
2770 create a double-word aligned @code{int} pair, you could write:
2773 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2777 This is an alternative to creating a union with a @code{double} member
2778 that forces the union to be double-word aligned.
2780 As in the preceding examples, you can explicitly specify the alignment
2781 (in bytes) that you wish the compiler to use for a given variable or
2782 structure field. Alternatively, you can leave out the alignment factor
2783 and just ask the compiler to align a variable or field to the maximum
2784 useful alignment for the target machine you are compiling for. For
2785 example, you could write:
2788 short array[3] __attribute__ ((aligned));
2791 Whenever you leave out the alignment factor in an @code{aligned} attribute
2792 specification, the compiler automatically sets the alignment for the declared
2793 variable or field to the largest alignment which is ever used for any data
2794 type on the target machine you are compiling for. Doing this can often make
2795 copy operations more efficient, because the compiler can use whatever
2796 instructions copy the biggest chunks of memory when performing copies to
2797 or from the variables or fields that you have aligned this way.
2799 The @code{aligned} attribute can only increase the alignment; but you
2800 can decrease it by specifying @code{packed} as well. See below.
2802 Note that the effectiveness of @code{aligned} attributes may be limited
2803 by inherent limitations in your linker. On many systems, the linker is
2804 only able to arrange for variables to be aligned up to a certain maximum
2805 alignment. (For some linkers, the maximum supported alignment may
2806 be very very small.) If your linker is only able to align variables
2807 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
2808 in an @code{__attribute__} will still only provide you with 8 byte
2809 alignment. See your linker documentation for further information.
2811 @item cleanup (@var{cleanup_function})
2812 @cindex @code{cleanup} attribute
2813 The @code{cleanup} attribute runs a function when the variable goes
2814 out of scope. This attribute can only be applied to auto function
2815 scope variables; it may not be applied to parameters or variables
2816 with static storage duration. The function must take one parameter,
2817 a pointer to a type compatible with the variable. The return value
2818 of the function (if any) is ignored.
2820 If @option{-fexceptions} is enabled, then @var{cleanup_function}
2821 will be run during the stack unwinding that happens during the
2822 processing of the exception. Note that the @code{cleanup} attribute
2823 does not allow the exception to be caught, only to perform an action.
2824 It is undefined what happens if @var{cleanup_function} does not
2829 @cindex @code{common} attribute
2830 @cindex @code{nocommon} attribute
2833 The @code{common} attribute requests GCC to place a variable in
2834 ``common'' storage. The @code{nocommon} attribute requests the
2835 opposite---to allocate space for it directly.
2837 These attributes override the default chosen by the
2838 @option{-fno-common} and @option{-fcommon} flags respectively.
2841 @cindex @code{deprecated} attribute
2842 The @code{deprecated} attribute results in a warning if the variable
2843 is used anywhere in the source file. This is useful when identifying
2844 variables that are expected to be removed in a future version of a
2845 program. The warning also includes the location of the declaration
2846 of the deprecated variable, to enable users to easily find further
2847 information about why the variable is deprecated, or what they should
2848 do instead. Note that the warning only occurs for uses:
2851 extern int old_var __attribute__ ((deprecated));
2853 int new_fn () @{ return old_var; @}
2856 results in a warning on line 3 but not line 2.
2858 The @code{deprecated} attribute can also be used for functions and
2859 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
2861 @item mode (@var{mode})
2862 @cindex @code{mode} attribute
2863 This attribute specifies the data type for the declaration---whichever
2864 type corresponds to the mode @var{mode}. This in effect lets you
2865 request an integer or floating point type according to its width.
2867 You may also specify a mode of @samp{byte} or @samp{__byte__} to
2868 indicate the mode corresponding to a one-byte integer, @samp{word} or
2869 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
2870 or @samp{__pointer__} for the mode used to represent pointers.
2873 @cindex @code{packed} attribute
2874 The @code{packed} attribute specifies that a variable or structure field
2875 should have the smallest possible alignment---one byte for a variable,
2876 and one bit for a field, unless you specify a larger value with the
2877 @code{aligned} attribute.
2879 Here is a structure in which the field @code{x} is packed, so that it
2880 immediately follows @code{a}:
2886 int x[2] __attribute__ ((packed));
2890 @item section ("@var{section-name}")
2891 @cindex @code{section} variable attribute
2892 Normally, the compiler places the objects it generates in sections like
2893 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
2894 or you need certain particular variables to appear in special sections,
2895 for example to map to special hardware. The @code{section}
2896 attribute specifies that a variable (or function) lives in a particular
2897 section. For example, this small program uses several specific section names:
2900 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
2901 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
2902 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
2903 int init_data __attribute__ ((section ("INITDATA"))) = 0;
2907 /* @r{Initialize stack pointer} */
2908 init_sp (stack + sizeof (stack));
2910 /* @r{Initialize initialized data} */
2911 memcpy (&init_data, &data, &edata - &data);
2913 /* @r{Turn on the serial ports} */
2920 Use the @code{section} attribute with an @emph{initialized} definition
2921 of a @emph{global} variable, as shown in the example. GCC issues
2922 a warning and otherwise ignores the @code{section} attribute in
2923 uninitialized variable declarations.
2925 You may only use the @code{section} attribute with a fully initialized
2926 global definition because of the way linkers work. The linker requires
2927 each object be defined once, with the exception that uninitialized
2928 variables tentatively go in the @code{common} (or @code{bss}) section
2929 and can be multiply ``defined''. You can force a variable to be
2930 initialized with the @option{-fno-common} flag or the @code{nocommon}
2933 Some file formats do not support arbitrary sections so the @code{section}
2934 attribute is not available on all platforms.
2935 If you need to map the entire contents of a module to a particular
2936 section, consider using the facilities of the linker instead.
2939 @cindex @code{shared} variable attribute
2940 On Microsoft Windows, in addition to putting variable definitions in a named
2941 section, the section can also be shared among all running copies of an
2942 executable or DLL@. For example, this small program defines shared data
2943 by putting it in a named section @code{shared} and marking the section
2947 int foo __attribute__((section ("shared"), shared)) = 0;
2952 /* @r{Read and write foo. All running
2953 copies see the same value.} */
2959 You may only use the @code{shared} attribute along with @code{section}
2960 attribute with a fully initialized global definition because of the way
2961 linkers work. See @code{section} attribute for more information.
2963 The @code{shared} attribute is only available on Microsoft Windows@.
2965 @item tls_model ("@var{tls_model}")
2966 @cindex @code{tls_model} attribute
2967 The @code{tls_model} attribute sets thread-local storage model
2968 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
2969 overriding @option{-ftls-model=} command line switch on a per-variable
2971 The @var{tls_model} argument should be one of @code{global-dynamic},
2972 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
2974 Not all targets support this attribute.
2976 @item transparent_union
2977 This attribute, attached to a function parameter which is a union, means
2978 that the corresponding argument may have the type of any union member,
2979 but the argument is passed as if its type were that of the first union
2980 member. For more details see @xref{Type Attributes}. You can also use
2981 this attribute on a @code{typedef} for a union data type; then it
2982 applies to all function parameters with that type.
2985 This attribute, attached to a variable, means that the variable is meant
2986 to be possibly unused. GCC will not produce a warning for this
2989 @item vector_size (@var{bytes})
2990 This attribute specifies the vector size for the variable, measured in
2991 bytes. For example, the declaration:
2994 int foo __attribute__ ((vector_size (16)));
2998 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
2999 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3000 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3002 This attribute is only applicable to integral and float scalars,
3003 although arrays, pointers, and function return values are allowed in
3004 conjunction with this construct.
3006 Aggregates with this attribute are invalid, even if they are of the same
3007 size as a corresponding scalar. For example, the declaration:
3010 struct S @{ int a; @};
3011 struct S __attribute__ ((vector_size (16))) foo;
3015 is invalid even if the size of the structure is the same as the size of
3019 The @code{weak} attribute is described in @xref{Function Attributes}.
3022 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3025 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3029 @subsection M32R/D Variable Attributes
3031 One attribute is currently defined for the M32R/D@.
3034 @item model (@var{model-name})
3035 @cindex variable addressability on the M32R/D
3036 Use this attribute on the M32R/D to set the addressability of an object.
3037 The identifier @var{model-name} is one of @code{small}, @code{medium},
3038 or @code{large}, representing each of the code models.
3040 Small model objects live in the lower 16MB of memory (so that their
3041 addresses can be loaded with the @code{ld24} instruction).
3043 Medium and large model objects may live anywhere in the 32-bit address space
3044 (the compiler will generate @code{seth/add3} instructions to load their
3048 @subsection i386 Variable Attributes
3050 Two attributes are currently defined for i386 configurations:
3051 @code{ms_struct} and @code{gcc_struct}
3056 @cindex @code{ms_struct} attribute
3057 @cindex @code{gcc_struct} attribute
3059 If @code{packed} is used on a structure, or if bit-fields are used
3060 it may be that the Microsoft ABI packs them differently
3061 than GCC would normally pack them. Particularly when moving packed
3062 data between functions compiled with GCC and the native Microsoft compiler
3063 (either via function call or as data in a file), it may be necessary to access
3066 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3067 compilers to match the native Microsoft compiler.
3070 @subsection Xstormy16 Variable Attributes
3072 One attribute is currently defined for xstormy16 configurations:
3077 @cindex @code{below100} attribute
3079 If a variable has the @code{below100} attribute (@code{BELOW100} is
3080 allowed also), GCC will place the variable in the first 0x100 bytes of
3081 memory and use special opcodes to access it. Such variables will be
3082 placed in either the @code{.bss_below100} section or the
3083 @code{.data_below100} section.
3087 @node Type Attributes
3088 @section Specifying Attributes of Types
3089 @cindex attribute of types
3090 @cindex type attributes
3092 The keyword @code{__attribute__} allows you to specify special
3093 attributes of @code{struct} and @code{union} types when you define such
3094 types. This keyword is followed by an attribute specification inside
3095 double parentheses. Six attributes are currently defined for types:
3096 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3097 @code{deprecated} and @code{may_alias}. Other attributes are defined for
3098 functions (@pxref{Function Attributes}) and for variables
3099 (@pxref{Variable Attributes}).
3101 You may also specify any one of these attributes with @samp{__}
3102 preceding and following its keyword. This allows you to use these
3103 attributes in header files without being concerned about a possible
3104 macro of the same name. For example, you may use @code{__aligned__}
3105 instead of @code{aligned}.
3107 You may specify the @code{aligned} and @code{transparent_union}
3108 attributes either in a @code{typedef} declaration or just past the
3109 closing curly brace of a complete enum, struct or union type
3110 @emph{definition} and the @code{packed} attribute only past the closing
3111 brace of a definition.
3113 You may also specify attributes between the enum, struct or union
3114 tag and the name of the type rather than after the closing brace.
3116 @xref{Attribute Syntax}, for details of the exact syntax for using
3120 @cindex @code{aligned} attribute
3121 @item aligned (@var{alignment})
3122 This attribute specifies a minimum alignment (in bytes) for variables
3123 of the specified type. For example, the declarations:
3126 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3127 typedef int more_aligned_int __attribute__ ((aligned (8)));
3131 force the compiler to insure (as far as it can) that each variable whose
3132 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3133 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3134 variables of type @code{struct S} aligned to 8-byte boundaries allows
3135 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3136 store) instructions when copying one variable of type @code{struct S} to
3137 another, thus improving run-time efficiency.
3139 Note that the alignment of any given @code{struct} or @code{union} type
3140 is required by the ISO C standard to be at least a perfect multiple of
3141 the lowest common multiple of the alignments of all of the members of
3142 the @code{struct} or @code{union} in question. This means that you @emph{can}
3143 effectively adjust the alignment of a @code{struct} or @code{union}
3144 type by attaching an @code{aligned} attribute to any one of the members
3145 of such a type, but the notation illustrated in the example above is a
3146 more obvious, intuitive, and readable way to request the compiler to
3147 adjust the alignment of an entire @code{struct} or @code{union} type.
3149 As in the preceding example, you can explicitly specify the alignment
3150 (in bytes) that you wish the compiler to use for a given @code{struct}
3151 or @code{union} type. Alternatively, you can leave out the alignment factor
3152 and just ask the compiler to align a type to the maximum
3153 useful alignment for the target machine you are compiling for. For
3154 example, you could write:
3157 struct S @{ short f[3]; @} __attribute__ ((aligned));
3160 Whenever you leave out the alignment factor in an @code{aligned}
3161 attribute specification, the compiler automatically sets the alignment
3162 for the type to the largest alignment which is ever used for any data
3163 type on the target machine you are compiling for. Doing this can often
3164 make copy operations more efficient, because the compiler can use
3165 whatever instructions copy the biggest chunks of memory when performing
3166 copies to or from the variables which have types that you have aligned
3169 In the example above, if the size of each @code{short} is 2 bytes, then
3170 the size of the entire @code{struct S} type is 6 bytes. The smallest
3171 power of two which is greater than or equal to that is 8, so the
3172 compiler sets the alignment for the entire @code{struct S} type to 8
3175 Note that although you can ask the compiler to select a time-efficient
3176 alignment for a given type and then declare only individual stand-alone
3177 objects of that type, the compiler's ability to select a time-efficient
3178 alignment is primarily useful only when you plan to create arrays of
3179 variables having the relevant (efficiently aligned) type. If you
3180 declare or use arrays of variables of an efficiently-aligned type, then
3181 it is likely that your program will also be doing pointer arithmetic (or
3182 subscripting, which amounts to the same thing) on pointers to the
3183 relevant type, and the code that the compiler generates for these
3184 pointer arithmetic operations will often be more efficient for
3185 efficiently-aligned types than for other types.
3187 The @code{aligned} attribute can only increase the alignment; but you
3188 can decrease it by specifying @code{packed} as well. See below.
3190 Note that the effectiveness of @code{aligned} attributes may be limited
3191 by inherent limitations in your linker. On many systems, the linker is
3192 only able to arrange for variables to be aligned up to a certain maximum
3193 alignment. (For some linkers, the maximum supported alignment may
3194 be very very small.) If your linker is only able to align variables
3195 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3196 in an @code{__attribute__} will still only provide you with 8 byte
3197 alignment. See your linker documentation for further information.
3200 This attribute, attached to @code{struct} or @code{union} type
3201 definition, specifies that each member of the structure or union is
3202 placed to minimize the memory required. When attached to an @code{enum}
3203 definition, it indicates that the smallest integral type should be used.
3205 @opindex fshort-enums
3206 Specifying this attribute for @code{struct} and @code{union} types is
3207 equivalent to specifying the @code{packed} attribute on each of the
3208 structure or union members. Specifying the @option{-fshort-enums}
3209 flag on the line is equivalent to specifying the @code{packed}
3210 attribute on all @code{enum} definitions.
3212 In the following example @code{struct my_packed_struct}'s members are
3213 packed closely together, but the internal layout of its @code{s} member
3214 is not packed---to do that, @code{struct my_unpacked_struct} would need to
3218 struct my_unpacked_struct
3224 struct __attribute__ ((__packed__)) my_packed_struct
3228 struct my_unpacked_struct s;
3232 You may only specify this attribute on the definition of a @code{enum},
3233 @code{struct} or @code{union}, not on a @code{typedef} which does not
3234 also define the enumerated type, structure or union.
3236 @item transparent_union
3237 This attribute, attached to a @code{union} type definition, indicates
3238 that any function parameter having that union type causes calls to that
3239 function to be treated in a special way.
3241 First, the argument corresponding to a transparent union type can be of
3242 any type in the union; no cast is required. Also, if the union contains
3243 a pointer type, the corresponding argument can be a null pointer
3244 constant or a void pointer expression; and if the union contains a void
3245 pointer type, the corresponding argument can be any pointer expression.
3246 If the union member type is a pointer, qualifiers like @code{const} on
3247 the referenced type must be respected, just as with normal pointer
3250 Second, the argument is passed to the function using the calling
3251 conventions of the first member of the transparent union, not the calling
3252 conventions of the union itself. All members of the union must have the
3253 same machine representation; this is necessary for this argument passing
3256 Transparent unions are designed for library functions that have multiple
3257 interfaces for compatibility reasons. For example, suppose the
3258 @code{wait} function must accept either a value of type @code{int *} to
3259 comply with Posix, or a value of type @code{union wait *} to comply with
3260 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3261 @code{wait} would accept both kinds of arguments, but it would also
3262 accept any other pointer type and this would make argument type checking
3263 less useful. Instead, @code{<sys/wait.h>} might define the interface
3271 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3273 pid_t wait (wait_status_ptr_t);
3276 This interface allows either @code{int *} or @code{union wait *}
3277 arguments to be passed, using the @code{int *} calling convention.
3278 The program can call @code{wait} with arguments of either type:
3281 int w1 () @{ int w; return wait (&w); @}
3282 int w2 () @{ union wait w; return wait (&w); @}
3285 With this interface, @code{wait}'s implementation might look like this:
3288 pid_t wait (wait_status_ptr_t p)
3290 return waitpid (-1, p.__ip, 0);
3295 When attached to a type (including a @code{union} or a @code{struct}),
3296 this attribute means that variables of that type are meant to appear
3297 possibly unused. GCC will not produce a warning for any variables of
3298 that type, even if the variable appears to do nothing. This is often
3299 the case with lock or thread classes, which are usually defined and then
3300 not referenced, but contain constructors and destructors that have
3301 nontrivial bookkeeping functions.
3304 The @code{deprecated} attribute results in a warning if the type
3305 is used anywhere in the source file. This is useful when identifying
3306 types that are expected to be removed in a future version of a program.
3307 If possible, the warning also includes the location of the declaration
3308 of the deprecated type, to enable users to easily find further
3309 information about why the type is deprecated, or what they should do
3310 instead. Note that the warnings only occur for uses and then only
3311 if the type is being applied to an identifier that itself is not being
3312 declared as deprecated.
3315 typedef int T1 __attribute__ ((deprecated));
3319 typedef T1 T3 __attribute__ ((deprecated));
3320 T3 z __attribute__ ((deprecated));
3323 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3324 warning is issued for line 4 because T2 is not explicitly
3325 deprecated. Line 5 has no warning because T3 is explicitly
3326 deprecated. Similarly for line 6.
3328 The @code{deprecated} attribute can also be used for functions and
3329 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3332 Accesses to objects with types with this attribute are not subjected to
3333 type-based alias analysis, but are instead assumed to be able to alias
3334 any other type of objects, just like the @code{char} type. See
3335 @option{-fstrict-aliasing} for more information on aliasing issues.
3340 typedef short __attribute__((__may_alias__)) short_a;
3346 short_a *b = (short_a *) &a;
3350 if (a == 0x12345678)
3357 If you replaced @code{short_a} with @code{short} in the variable
3358 declaration, the above program would abort when compiled with
3359 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3360 above in recent GCC versions.
3362 @subsection ARM Type Attributes
3364 On those ARM targets that support @code{dllimport} (such as Symbian
3365 OS), you can use the @code{notshared} attribute to indicate that the
3366 virtual table and other similar data for a class should not be
3367 exported from a DLL@. For example:
3370 class __declspec(notshared) C @{
3372 __declspec(dllimport) C();
3376 __declspec(dllexport)
3380 In this code, @code{C::C} is exported from the current DLL, but the
3381 virtual table for @code{C} is not exported. (You can use
3382 @code{__attribute__} instead of @code{__declspec} if you prefer, but
3383 most Symbian OS code uses @code{__declspec}.)
3385 @subsection i386 Type Attributes
3387 Two attributes are currently defined for i386 configurations:
3388 @code{ms_struct} and @code{gcc_struct}
3392 @cindex @code{ms_struct}
3393 @cindex @code{gcc_struct}
3395 If @code{packed} is used on a structure, or if bit-fields are used
3396 it may be that the Microsoft ABI packs them differently
3397 than GCC would normally pack them. Particularly when moving packed
3398 data between functions compiled with GCC and the native Microsoft compiler
3399 (either via function call or as data in a file), it may be necessary to access
3402 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3403 compilers to match the native Microsoft compiler.
3406 To specify multiple attributes, separate them by commas within the
3407 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3411 @section An Inline Function is As Fast As a Macro
3412 @cindex inline functions
3413 @cindex integrating function code
3415 @cindex macros, inline alternative
3417 By declaring a function @code{inline}, you can direct GCC to
3418 integrate that function's code into the code for its callers. This
3419 makes execution faster by eliminating the function-call overhead; in
3420 addition, if any of the actual argument values are constant, their known
3421 values may permit simplifications at compile time so that not all of the
3422 inline function's code needs to be included. The effect on code size is
3423 less predictable; object code may be larger or smaller with function
3424 inlining, depending on the particular case. Inlining of functions is an
3425 optimization and it really ``works'' only in optimizing compilation. If
3426 you don't use @option{-O}, no function is really inline.
3428 Inline functions are included in the ISO C99 standard, but there are
3429 currently substantial differences between what GCC implements and what
3430 the ISO C99 standard requires.
3432 To declare a function inline, use the @code{inline} keyword in its
3433 declaration, like this:
3443 (If you are writing a header file to be included in ISO C programs, write
3444 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3445 You can also make all ``simple enough'' functions inline with the option
3446 @option{-finline-functions}.
3449 Note that certain usages in a function definition can make it unsuitable
3450 for inline substitution. Among these usages are: use of varargs, use of
3451 alloca, use of variable sized data types (@pxref{Variable Length}),
3452 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3453 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3454 will warn when a function marked @code{inline} could not be substituted,
3455 and will give the reason for the failure.
3457 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3458 does not affect the linkage of the function.
3460 @cindex automatic @code{inline} for C++ member fns
3461 @cindex @code{inline} automatic for C++ member fns
3462 @cindex member fns, automatically @code{inline}
3463 @cindex C++ member fns, automatically @code{inline}
3464 @opindex fno-default-inline
3465 GCC automatically inlines member functions defined within the class
3466 body of C++ programs even if they are not explicitly declared
3467 @code{inline}. (You can override this with @option{-fno-default-inline};
3468 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3470 @cindex inline functions, omission of
3471 @opindex fkeep-inline-functions
3472 When a function is both inline and @code{static}, if all calls to the
3473 function are integrated into the caller, and the function's address is
3474 never used, then the function's own assembler code is never referenced.
3475 In this case, GCC does not actually output assembler code for the
3476 function, unless you specify the option @option{-fkeep-inline-functions}.
3477 Some calls cannot be integrated for various reasons (in particular,
3478 calls that precede the function's definition cannot be integrated, and
3479 neither can recursive calls within the definition). If there is a
3480 nonintegrated call, then the function is compiled to assembler code as
3481 usual. The function must also be compiled as usual if the program
3482 refers to its address, because that can't be inlined.
3484 @cindex non-static inline function
3485 When an inline function is not @code{static}, then the compiler must assume
3486 that there may be calls from other source files; since a global symbol can
3487 be defined only once in any program, the function must not be defined in
3488 the other source files, so the calls therein cannot be integrated.
3489 Therefore, a non-@code{static} inline function is always compiled on its
3490 own in the usual fashion.
3492 If you specify both @code{inline} and @code{extern} in the function
3493 definition, then the definition is used only for inlining. In no case
3494 is the function compiled on its own, not even if you refer to its
3495 address explicitly. Such an address becomes an external reference, as
3496 if you had only declared the function, and had not defined it.
3498 This combination of @code{inline} and @code{extern} has almost the
3499 effect of a macro. The way to use it is to put a function definition in
3500 a header file with these keywords, and put another copy of the
3501 definition (lacking @code{inline} and @code{extern}) in a library file.
3502 The definition in the header file will cause most calls to the function
3503 to be inlined. If any uses of the function remain, they will refer to
3504 the single copy in the library.
3506 Since GCC eventually will implement ISO C99 semantics for
3507 inline functions, it is best to use @code{static inline} only
3508 to guarantee compatibility. (The
3509 existing semantics will remain available when @option{-std=gnu89} is
3510 specified, but eventually the default will be @option{-std=gnu99} and
3511 that will implement the C99 semantics, though it does not do so yet.)
3513 GCC does not inline any functions when not optimizing unless you specify
3514 the @samp{always_inline} attribute for the function, like this:
3517 /* @r{Prototype.} */
3518 inline void foo (const char) __attribute__((always_inline));
3522 @section Assembler Instructions with C Expression Operands
3523 @cindex extended @code{asm}
3524 @cindex @code{asm} expressions
3525 @cindex assembler instructions
3528 In an assembler instruction using @code{asm}, you can specify the
3529 operands of the instruction using C expressions. This means you need not
3530 guess which registers or memory locations will contain the data you want
3533 You must specify an assembler instruction template much like what
3534 appears in a machine description, plus an operand constraint string for
3537 For example, here is how to use the 68881's @code{fsinx} instruction:
3540 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3544 Here @code{angle} is the C expression for the input operand while
3545 @code{result} is that of the output operand. Each has @samp{"f"} as its
3546 operand constraint, saying that a floating point register is required.
3547 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3548 output operands' constraints must use @samp{=}. The constraints use the
3549 same language used in the machine description (@pxref{Constraints}).
3551 Each operand is described by an operand-constraint string followed by
3552 the C expression in parentheses. A colon separates the assembler
3553 template from the first output operand and another separates the last
3554 output operand from the first input, if any. Commas separate the
3555 operands within each group. The total number of operands is currently
3556 limited to 30; this limitation may be lifted in some future version of
3559 If there are no output operands but there are input operands, you must
3560 place two consecutive colons surrounding the place where the output
3563 As of GCC version 3.1, it is also possible to specify input and output
3564 operands using symbolic names which can be referenced within the
3565 assembler code. These names are specified inside square brackets
3566 preceding the constraint string, and can be referenced inside the
3567 assembler code using @code{%[@var{name}]} instead of a percentage sign
3568 followed by the operand number. Using named operands the above example
3572 asm ("fsinx %[angle],%[output]"
3573 : [output] "=f" (result)
3574 : [angle] "f" (angle));
3578 Note that the symbolic operand names have no relation whatsoever to
3579 other C identifiers. You may use any name you like, even those of
3580 existing C symbols, but you must ensure that no two operands within the same
3581 assembler construct use the same symbolic name.
3583 Output operand expressions must be lvalues; the compiler can check this.
3584 The input operands need not be lvalues. The compiler cannot check
3585 whether the operands have data types that are reasonable for the
3586 instruction being executed. It does not parse the assembler instruction
3587 template and does not know what it means or even whether it is valid
3588 assembler input. The extended @code{asm} feature is most often used for
3589 machine instructions the compiler itself does not know exist. If
3590 the output expression cannot be directly addressed (for example, it is a
3591 bit-field), your constraint must allow a register. In that case, GCC
3592 will use the register as the output of the @code{asm}, and then store
3593 that register into the output.
3595 The ordinary output operands must be write-only; GCC will assume that
3596 the values in these operands before the instruction are dead and need
3597 not be generated. Extended asm supports input-output or read-write
3598 operands. Use the constraint character @samp{+} to indicate such an
3599 operand and list it with the output operands. You should only use
3600 read-write operands when the constraints for the operand (or the
3601 operand in which only some of the bits are to be changed) allow a
3604 You may, as an alternative, logically split its function into two
3605 separate operands, one input operand and one write-only output
3606 operand. The connection between them is expressed by constraints
3607 which say they need to be in the same location when the instruction
3608 executes. You can use the same C expression for both operands, or
3609 different expressions. For example, here we write the (fictitious)
3610 @samp{combine} instruction with @code{bar} as its read-only source
3611 operand and @code{foo} as its read-write destination:
3614 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3618 The constraint @samp{"0"} for operand 1 says that it must occupy the
3619 same location as operand 0. A number in constraint is allowed only in
3620 an input operand and it must refer to an output operand.
3622 Only a number in the constraint can guarantee that one operand will be in
3623 the same place as another. The mere fact that @code{foo} is the value
3624 of both operands is not enough to guarantee that they will be in the
3625 same place in the generated assembler code. The following would not
3629 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3632 Various optimizations or reloading could cause operands 0 and 1 to be in
3633 different registers; GCC knows no reason not to do so. For example, the
3634 compiler might find a copy of the value of @code{foo} in one register and
3635 use it for operand 1, but generate the output operand 0 in a different
3636 register (copying it afterward to @code{foo}'s own address). Of course,
3637 since the register for operand 1 is not even mentioned in the assembler
3638 code, the result will not work, but GCC can't tell that.
3640 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3641 the operand number for a matching constraint. For example:
3644 asm ("cmoveq %1,%2,%[result]"
3645 : [result] "=r"(result)
3646 : "r" (test), "r"(new), "[result]"(old));
3649 Sometimes you need to make an @code{asm} operand be a specific register,
3650 but there's no matching constraint letter for that register @emph{by
3651 itself}. To force the operand into that register, use a local variable
3652 for the operand and specify the register in the variable declaration.
3653 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
3654 register constraint letter that matches the register:
3657 register int *p1 asm ("r0") = @dots{};
3658 register int *p2 asm ("r1") = @dots{};
3659 register int *result asm ("r0");
3660 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3663 @anchor{Example of asm with clobbered asm reg}
3664 In the above example, beware that a register that is call-clobbered by
3665 the target ABI will be overwritten by any function call in the
3666 assignment, including library calls for arithmetic operators.
3667 Assuming it is a call-clobbered register, this may happen to @code{r0}
3668 above by the assignment to @code{p2}. If you have to use such a
3669 register, use temporary variables for expressions between the register
3674 register int *p1 asm ("r0") = @dots{};
3675 register int *p2 asm ("r1") = t1;
3676 register int *result asm ("r0");
3677 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3680 Some instructions clobber specific hard registers. To describe this,
3681 write a third colon after the input operands, followed by the names of
3682 the clobbered hard registers (given as strings). Here is a realistic
3683 example for the VAX:
3686 asm volatile ("movc3 %0,%1,%2"
3687 : /* @r{no outputs} */
3688 : "g" (from), "g" (to), "g" (count)
3689 : "r0", "r1", "r2", "r3", "r4", "r5");
3692 You may not write a clobber description in a way that overlaps with an
3693 input or output operand. For example, you may not have an operand
3694 describing a register class with one member if you mention that register
3695 in the clobber list. Variables declared to live in specific registers
3696 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
3697 have no part mentioned in the clobber description.
3698 There is no way for you to specify that an input
3699 operand is modified without also specifying it as an output
3700 operand. Note that if all the output operands you specify are for this
3701 purpose (and hence unused), you will then also need to specify
3702 @code{volatile} for the @code{asm} construct, as described below, to
3703 prevent GCC from deleting the @code{asm} statement as unused.
3705 If you refer to a particular hardware register from the assembler code,
3706 you will probably have to list the register after the third colon to
3707 tell the compiler the register's value is modified. In some assemblers,
3708 the register names begin with @samp{%}; to produce one @samp{%} in the
3709 assembler code, you must write @samp{%%} in the input.
3711 If your assembler instruction can alter the condition code register, add
3712 @samp{cc} to the list of clobbered registers. GCC on some machines
3713 represents the condition codes as a specific hardware register;
3714 @samp{cc} serves to name this register. On other machines, the
3715 condition code is handled differently, and specifying @samp{cc} has no
3716 effect. But it is valid no matter what the machine.
3718 If your assembler instructions access memory in an unpredictable
3719 fashion, add @samp{memory} to the list of clobbered registers. This
3720 will cause GCC to not keep memory values cached in registers across the
3721 assembler instruction and not optimize stores or loads to that memory.
3722 You will also want to add the @code{volatile} keyword if the memory
3723 affected is not listed in the inputs or outputs of the @code{asm}, as
3724 the @samp{memory} clobber does not count as a side-effect of the
3725 @code{asm}. If you know how large the accessed memory is, you can add
3726 it as input or output but if this is not known, you should add
3727 @samp{memory}. As an example, if you access ten bytes of a string, you
3728 can use a memory input like:
3731 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
3734 Note that in the following example the memory input is necessary,
3735 otherwise GCC might optimize the store to @code{x} away:
3742 asm ("magic stuff accessing an 'int' pointed to by '%1'"
3743 "=&d" (r) : "a" (y), "m" (*y));
3748 You can put multiple assembler instructions together in a single
3749 @code{asm} template, separated by the characters normally used in assembly
3750 code for the system. A combination that works in most places is a newline
3751 to break the line, plus a tab character to move to the instruction field
3752 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
3753 assembler allows semicolons as a line-breaking character. Note that some
3754 assembler dialects use semicolons to start a comment.
3755 The input operands are guaranteed not to use any of the clobbered
3756 registers, and neither will the output operands' addresses, so you can
3757 read and write the clobbered registers as many times as you like. Here
3758 is an example of multiple instructions in a template; it assumes the
3759 subroutine @code{_foo} accepts arguments in registers 9 and 10:
3762 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
3764 : "g" (from), "g" (to)
3768 Unless an output operand has the @samp{&} constraint modifier, GCC
3769 may allocate it in the same register as an unrelated input operand, on
3770 the assumption the inputs are consumed before the outputs are produced.
3771 This assumption may be false if the assembler code actually consists of
3772 more than one instruction. In such a case, use @samp{&} for each output
3773 operand that may not overlap an input. @xref{Modifiers}.
3775 If you want to test the condition code produced by an assembler
3776 instruction, you must include a branch and a label in the @code{asm}
3777 construct, as follows:
3780 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
3786 This assumes your assembler supports local labels, as the GNU assembler
3787 and most Unix assemblers do.
3789 Speaking of labels, jumps from one @code{asm} to another are not
3790 supported. The compiler's optimizers do not know about these jumps, and
3791 therefore they cannot take account of them when deciding how to
3794 @cindex macros containing @code{asm}
3795 Usually the most convenient way to use these @code{asm} instructions is to
3796 encapsulate them in macros that look like functions. For example,
3800 (@{ double __value, __arg = (x); \
3801 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
3806 Here the variable @code{__arg} is used to make sure that the instruction
3807 operates on a proper @code{double} value, and to accept only those
3808 arguments @code{x} which can convert automatically to a @code{double}.
3810 Another way to make sure the instruction operates on the correct data
3811 type is to use a cast in the @code{asm}. This is different from using a
3812 variable @code{__arg} in that it converts more different types. For
3813 example, if the desired type were @code{int}, casting the argument to
3814 @code{int} would accept a pointer with no complaint, while assigning the
3815 argument to an @code{int} variable named @code{__arg} would warn about
3816 using a pointer unless the caller explicitly casts it.
3818 If an @code{asm} has output operands, GCC assumes for optimization
3819 purposes the instruction has no side effects except to change the output
3820 operands. This does not mean instructions with a side effect cannot be
3821 used, but you must be careful, because the compiler may eliminate them
3822 if the output operands aren't used, or move them out of loops, or
3823 replace two with one if they constitute a common subexpression. Also,
3824 if your instruction does have a side effect on a variable that otherwise
3825 appears not to change, the old value of the variable may be reused later
3826 if it happens to be found in a register.
3828 You can prevent an @code{asm} instruction from being deleted
3829 by writing the keyword @code{volatile} after
3830 the @code{asm}. For example:
3833 #define get_and_set_priority(new) \
3835 asm volatile ("get_and_set_priority %0, %1" \
3836 : "=g" (__old) : "g" (new)); \
3841 The @code{volatile} keyword indicates that the instruction has
3842 important side-effects. GCC will not delete a volatile @code{asm} if
3843 it is reachable. (The instruction can still be deleted if GCC can
3844 prove that control-flow will never reach the location of the
3845 instruction.) Note that even a volatile @code{asm} instruction
3846 can be moved relative to other code, including across jump
3847 instructions. For example, on many targets there is a system
3848 register which can be set to control the rounding mode of
3849 floating point operations. You might try
3850 setting it with a volatile @code{asm}, like this PowerPC example:
3853 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
3858 This will not work reliably, as the compiler may move the addition back
3859 before the volatile @code{asm}. To make it work you need to add an
3860 artificial dependency to the @code{asm} referencing a variable in the code
3861 you don't want moved, for example:
3864 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
3868 Similarly, you can't expect a
3869 sequence of volatile @code{asm} instructions to remain perfectly
3870 consecutive. If you want consecutive output, use a single @code{asm}.
3871 Also, GCC will perform some optimizations across a volatile @code{asm}
3872 instruction; GCC does not ``forget everything'' when it encounters
3873 a volatile @code{asm} instruction the way some other compilers do.
3875 An @code{asm} instruction without any output operands will be treated
3876 identically to a volatile @code{asm} instruction.
3878 It is a natural idea to look for a way to give access to the condition
3879 code left by the assembler instruction. However, when we attempted to
3880 implement this, we found no way to make it work reliably. The problem
3881 is that output operands might need reloading, which would result in
3882 additional following ``store'' instructions. On most machines, these
3883 instructions would alter the condition code before there was time to
3884 test it. This problem doesn't arise for ordinary ``test'' and
3885 ``compare'' instructions because they don't have any output operands.
3887 For reasons similar to those described above, it is not possible to give
3888 an assembler instruction access to the condition code left by previous
3891 If you are writing a header file that should be includable in ISO C
3892 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
3895 @subsection Size of an @code{asm}
3897 Some targets require that GCC track the size of each instruction used in
3898 order to generate correct code. Because the final length of an
3899 @code{asm} is only known by the assembler, GCC must make an estimate as
3900 to how big it will be. The estimate is formed by counting the number of
3901 statements in the pattern of the @code{asm} and multiplying that by the
3902 length of the longest instruction on that processor. Statements in the
3903 @code{asm} are identified by newline characters and whatever statement
3904 separator characters are supported by the assembler; on most processors
3905 this is the `@code{;}' character.
3907 Normally, GCC's estimate is perfectly adequate to ensure that correct
3908 code is generated, but it is possible to confuse the compiler if you use
3909 pseudo instructions or assembler macros that expand into multiple real
3910 instructions or if you use assembler directives that expand to more
3911 space in the object file than would be needed for a single instruction.
3912 If this happens then the assembler will produce a diagnostic saying that
3913 a label is unreachable.
3915 @subsection i386 floating point asm operands
3917 There are several rules on the usage of stack-like regs in
3918 asm_operands insns. These rules apply only to the operands that are
3923 Given a set of input regs that die in an asm_operands, it is
3924 necessary to know which are implicitly popped by the asm, and
3925 which must be explicitly popped by gcc.
3927 An input reg that is implicitly popped by the asm must be
3928 explicitly clobbered, unless it is constrained to match an
3932 For any input reg that is implicitly popped by an asm, it is
3933 necessary to know how to adjust the stack to compensate for the pop.
3934 If any non-popped input is closer to the top of the reg-stack than
3935 the implicitly popped reg, it would not be possible to know what the
3936 stack looked like---it's not clear how the rest of the stack ``slides
3939 All implicitly popped input regs must be closer to the top of
3940 the reg-stack than any input that is not implicitly popped.
3942 It is possible that if an input dies in an insn, reload might
3943 use the input reg for an output reload. Consider this example:
3946 asm ("foo" : "=t" (a) : "f" (b));
3949 This asm says that input B is not popped by the asm, and that
3950 the asm pushes a result onto the reg-stack, i.e., the stack is one
3951 deeper after the asm than it was before. But, it is possible that
3952 reload will think that it can use the same reg for both the input and
3953 the output, if input B dies in this insn.
3955 If any input operand uses the @code{f} constraint, all output reg
3956 constraints must use the @code{&} earlyclobber.
3958 The asm above would be written as
3961 asm ("foo" : "=&t" (a) : "f" (b));
3965 Some operands need to be in particular places on the stack. All
3966 output operands fall in this category---there is no other way to
3967 know which regs the outputs appear in unless the user indicates
3968 this in the constraints.
3970 Output operands must specifically indicate which reg an output
3971 appears in after an asm. @code{=f} is not allowed: the operand
3972 constraints must select a class with a single reg.
3975 Output operands may not be ``inserted'' between existing stack regs.
3976 Since no 387 opcode uses a read/write operand, all output operands
3977 are dead before the asm_operands, and are pushed by the asm_operands.
3978 It makes no sense to push anywhere but the top of the reg-stack.
3980 Output operands must start at the top of the reg-stack: output
3981 operands may not ``skip'' a reg.
3984 Some asm statements may need extra stack space for internal
3985 calculations. This can be guaranteed by clobbering stack registers
3986 unrelated to the inputs and outputs.
3990 Here are a couple of reasonable asms to want to write. This asm
3991 takes one input, which is internally popped, and produces two outputs.
3994 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
3997 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
3998 and replaces them with one output. The user must code the @code{st(1)}
3999 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4002 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4008 @section Controlling Names Used in Assembler Code
4009 @cindex assembler names for identifiers
4010 @cindex names used in assembler code
4011 @cindex identifiers, names in assembler code
4013 You can specify the name to be used in the assembler code for a C
4014 function or variable by writing the @code{asm} (or @code{__asm__})
4015 keyword after the declarator as follows:
4018 int foo asm ("myfoo") = 2;
4022 This specifies that the name to be used for the variable @code{foo} in
4023 the assembler code should be @samp{myfoo} rather than the usual
4026 On systems where an underscore is normally prepended to the name of a C
4027 function or variable, this feature allows you to define names for the
4028 linker that do not start with an underscore.
4030 It does not make sense to use this feature with a non-static local
4031 variable since such variables do not have assembler names. If you are
4032 trying to put the variable in a particular register, see @ref{Explicit
4033 Reg Vars}. GCC presently accepts such code with a warning, but will
4034 probably be changed to issue an error, rather than a warning, in the
4037 You cannot use @code{asm} in this way in a function @emph{definition}; but
4038 you can get the same effect by writing a declaration for the function
4039 before its definition and putting @code{asm} there, like this:
4042 extern func () asm ("FUNC");
4049 It is up to you to make sure that the assembler names you choose do not
4050 conflict with any other assembler symbols. Also, you must not use a
4051 register name; that would produce completely invalid assembler code. GCC
4052 does not as yet have the ability to store static variables in registers.
4053 Perhaps that will be added.
4055 @node Explicit Reg Vars
4056 @section Variables in Specified Registers
4057 @cindex explicit register variables
4058 @cindex variables in specified registers
4059 @cindex specified registers
4060 @cindex registers, global allocation
4062 GNU C allows you to put a few global variables into specified hardware
4063 registers. You can also specify the register in which an ordinary
4064 register variable should be allocated.
4068 Global register variables reserve registers throughout the program.
4069 This may be useful in programs such as programming language
4070 interpreters which have a couple of global variables that are accessed
4074 Local register variables in specific registers do not reserve the
4075 registers, except at the point where they are used as input or output
4076 operands in an @code{asm} statement and the @code{asm} statement itself is
4077 not deleted. The compiler's data flow analysis is capable of determining
4078 where the specified registers contain live values, and where they are
4079 available for other uses. Stores into local register variables may be deleted
4080 when they appear to be dead according to dataflow analysis. References
4081 to local register variables may be deleted or moved or simplified.
4083 These local variables are sometimes convenient for use with the extended
4084 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4085 output of the assembler instruction directly into a particular register.
4086 (This will work provided the register you specify fits the constraints
4087 specified for that operand in the @code{asm}.)
4095 @node Global Reg Vars
4096 @subsection Defining Global Register Variables
4097 @cindex global register variables
4098 @cindex registers, global variables in
4100 You can define a global register variable in GNU C like this:
4103 register int *foo asm ("a5");
4107 Here @code{a5} is the name of the register which should be used. Choose a
4108 register which is normally saved and restored by function calls on your
4109 machine, so that library routines will not clobber it.
4111 Naturally the register name is cpu-dependent, so you would need to
4112 conditionalize your program according to cpu type. The register
4113 @code{a5} would be a good choice on a 68000 for a variable of pointer
4114 type. On machines with register windows, be sure to choose a ``global''
4115 register that is not affected magically by the function call mechanism.
4117 In addition, operating systems on one type of cpu may differ in how they
4118 name the registers; then you would need additional conditionals. For
4119 example, some 68000 operating systems call this register @code{%a5}.
4121 Eventually there may be a way of asking the compiler to choose a register
4122 automatically, but first we need to figure out how it should choose and
4123 how to enable you to guide the choice. No solution is evident.
4125 Defining a global register variable in a certain register reserves that
4126 register entirely for this use, at least within the current compilation.
4127 The register will not be allocated for any other purpose in the functions
4128 in the current compilation. The register will not be saved and restored by
4129 these functions. Stores into this register are never deleted even if they
4130 would appear to be dead, but references may be deleted or moved or
4133 It is not safe to access the global register variables from signal
4134 handlers, or from more than one thread of control, because the system
4135 library routines may temporarily use the register for other things (unless
4136 you recompile them specially for the task at hand).
4138 @cindex @code{qsort}, and global register variables
4139 It is not safe for one function that uses a global register variable to
4140 call another such function @code{foo} by way of a third function
4141 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4142 different source file in which the variable wasn't declared). This is
4143 because @code{lose} might save the register and put some other value there.
4144 For example, you can't expect a global register variable to be available in
4145 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4146 might have put something else in that register. (If you are prepared to
4147 recompile @code{qsort} with the same global register variable, you can
4148 solve this problem.)
4150 If you want to recompile @code{qsort} or other source files which do not
4151 actually use your global register variable, so that they will not use that
4152 register for any other purpose, then it suffices to specify the compiler
4153 option @option{-ffixed-@var{reg}}. You need not actually add a global
4154 register declaration to their source code.
4156 A function which can alter the value of a global register variable cannot
4157 safely be called from a function compiled without this variable, because it
4158 could clobber the value the caller expects to find there on return.
4159 Therefore, the function which is the entry point into the part of the
4160 program that uses the global register variable must explicitly save and
4161 restore the value which belongs to its caller.
4163 @cindex register variable after @code{longjmp}
4164 @cindex global register after @code{longjmp}
4165 @cindex value after @code{longjmp}
4168 On most machines, @code{longjmp} will restore to each global register
4169 variable the value it had at the time of the @code{setjmp}. On some
4170 machines, however, @code{longjmp} will not change the value of global
4171 register variables. To be portable, the function that called @code{setjmp}
4172 should make other arrangements to save the values of the global register
4173 variables, and to restore them in a @code{longjmp}. This way, the same
4174 thing will happen regardless of what @code{longjmp} does.
4176 All global register variable declarations must precede all function
4177 definitions. If such a declaration could appear after function
4178 definitions, the declaration would be too late to prevent the register from
4179 being used for other purposes in the preceding functions.
4181 Global register variables may not have initial values, because an
4182 executable file has no means to supply initial contents for a register.
4184 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4185 registers, but certain library functions, such as @code{getwd}, as well
4186 as the subroutines for division and remainder, modify g3 and g4. g1 and
4187 g2 are local temporaries.
4189 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4190 Of course, it will not do to use more than a few of those.
4192 @node Local Reg Vars
4193 @subsection Specifying Registers for Local Variables
4194 @cindex local variables, specifying registers
4195 @cindex specifying registers for local variables
4196 @cindex registers for local variables
4198 You can define a local register variable with a specified register
4202 register int *foo asm ("a5");
4206 Here @code{a5} is the name of the register which should be used. Note
4207 that this is the same syntax used for defining global register
4208 variables, but for a local variable it would appear within a function.
4210 Naturally the register name is cpu-dependent, but this is not a
4211 problem, since specific registers are most often useful with explicit
4212 assembler instructions (@pxref{Extended Asm}). Both of these things
4213 generally require that you conditionalize your program according to
4216 In addition, operating systems on one type of cpu may differ in how they
4217 name the registers; then you would need additional conditionals. For
4218 example, some 68000 operating systems call this register @code{%a5}.
4220 Defining such a register variable does not reserve the register; it
4221 remains available for other uses in places where flow control determines
4222 the variable's value is not live.
4224 This option does not guarantee that GCC will generate code that has
4225 this variable in the register you specify at all times. You may not
4226 code an explicit reference to this register in the @emph{assembler
4227 instruction template} part of an @code{asm} statement and assume it will
4228 always refer to this variable. However, using the variable as an
4229 @code{asm} @emph{operand} guarantees that the specified register is used
4232 Stores into local register variables may be deleted when they appear to be dead
4233 according to dataflow analysis. References to local register variables may
4234 be deleted or moved or simplified.
4236 As for global register variables, it's recommended that you choose a
4237 register which is normally saved and restored by function calls on
4238 your machine, so that library routines will not clobber it. A common
4239 pitfall is to initialize multiple call-clobbered registers with
4240 arbitrary expressions, where a function call or library call for an
4241 arithmetic operator will overwrite a register value from a previous
4242 assignment, for example @code{r0} below:
4244 register int *p1 asm ("r0") = @dots{};
4245 register int *p2 asm ("r1") = @dots{};
4247 In those cases, a solution is to use a temporary variable for
4248 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
4250 @node Alternate Keywords
4251 @section Alternate Keywords
4252 @cindex alternate keywords
4253 @cindex keywords, alternate
4255 @option{-ansi} and the various @option{-std} options disable certain
4256 keywords. This causes trouble when you want to use GNU C extensions, or
4257 a general-purpose header file that should be usable by all programs,
4258 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4259 @code{inline} are not available in programs compiled with
4260 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4261 program compiled with @option{-std=c99}). The ISO C99 keyword
4262 @code{restrict} is only available when @option{-std=gnu99} (which will
4263 eventually be the default) or @option{-std=c99} (or the equivalent
4264 @option{-std=iso9899:1999}) is used.
4266 The way to solve these problems is to put @samp{__} at the beginning and
4267 end of each problematical keyword. For example, use @code{__asm__}
4268 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4270 Other C compilers won't accept these alternative keywords; if you want to
4271 compile with another compiler, you can define the alternate keywords as
4272 macros to replace them with the customary keywords. It looks like this:
4280 @findex __extension__
4282 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4284 prevent such warnings within one expression by writing
4285 @code{__extension__} before the expression. @code{__extension__} has no
4286 effect aside from this.
4288 @node Incomplete Enums
4289 @section Incomplete @code{enum} Types
4291 You can define an @code{enum} tag without specifying its possible values.
4292 This results in an incomplete type, much like what you get if you write
4293 @code{struct foo} without describing the elements. A later declaration
4294 which does specify the possible values completes the type.
4296 You can't allocate variables or storage using the type while it is
4297 incomplete. However, you can work with pointers to that type.
4299 This extension may not be very useful, but it makes the handling of
4300 @code{enum} more consistent with the way @code{struct} and @code{union}
4303 This extension is not supported by GNU C++.
4305 @node Function Names
4306 @section Function Names as Strings
4307 @cindex @code{__func__} identifier
4308 @cindex @code{__FUNCTION__} identifier
4309 @cindex @code{__PRETTY_FUNCTION__} identifier
4311 GCC provides three magic variables which hold the name of the current
4312 function, as a string. The first of these is @code{__func__}, which
4313 is part of the C99 standard:
4316 The identifier @code{__func__} is implicitly declared by the translator
4317 as if, immediately following the opening brace of each function
4318 definition, the declaration
4321 static const char __func__[] = "function-name";
4324 appeared, where function-name is the name of the lexically-enclosing
4325 function. This name is the unadorned name of the function.
4328 @code{__FUNCTION__} is another name for @code{__func__}. Older
4329 versions of GCC recognize only this name. However, it is not
4330 standardized. For maximum portability, we recommend you use
4331 @code{__func__}, but provide a fallback definition with the
4335 #if __STDC_VERSION__ < 199901L
4337 # define __func__ __FUNCTION__
4339 # define __func__ "<unknown>"
4344 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4345 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4346 the type signature of the function as well as its bare name. For
4347 example, this program:
4351 extern int printf (char *, ...);
4358 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4359 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4377 __PRETTY_FUNCTION__ = void a::sub(int)
4380 These identifiers are not preprocessor macros. In GCC 3.3 and
4381 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4382 were treated as string literals; they could be used to initialize
4383 @code{char} arrays, and they could be concatenated with other string
4384 literals. GCC 3.4 and later treat them as variables, like
4385 @code{__func__}. In C++, @code{__FUNCTION__} and
4386 @code{__PRETTY_FUNCTION__} have always been variables.
4388 @node Return Address
4389 @section Getting the Return or Frame Address of a Function
4391 These functions may be used to get information about the callers of a
4394 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4395 This function returns the return address of the current function, or of
4396 one of its callers. The @var{level} argument is number of frames to
4397 scan up the call stack. A value of @code{0} yields the return address
4398 of the current function, a value of @code{1} yields the return address
4399 of the caller of the current function, and so forth. When inlining
4400 the expected behavior is that the function will return the address of
4401 the function that will be returned to. To work around this behavior use
4402 the @code{noinline} function attribute.
4404 The @var{level} argument must be a constant integer.
4406 On some machines it may be impossible to determine the return address of
4407 any function other than the current one; in such cases, or when the top
4408 of the stack has been reached, this function will return @code{0} or a
4409 random value. In addition, @code{__builtin_frame_address} may be used
4410 to determine if the top of the stack has been reached.
4412 This function should only be used with a nonzero argument for debugging
4416 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4417 This function is similar to @code{__builtin_return_address}, but it
4418 returns the address of the function frame rather than the return address
4419 of the function. Calling @code{__builtin_frame_address} with a value of
4420 @code{0} yields the frame address of the current function, a value of
4421 @code{1} yields the frame address of the caller of the current function,
4424 The frame is the area on the stack which holds local variables and saved
4425 registers. The frame address is normally the address of the first word
4426 pushed on to the stack by the function. However, the exact definition
4427 depends upon the processor and the calling convention. If the processor
4428 has a dedicated frame pointer register, and the function has a frame,
4429 then @code{__builtin_frame_address} will return the value of the frame
4432 On some machines it may be impossible to determine the frame address of
4433 any function other than the current one; in such cases, or when the top
4434 of the stack has been reached, this function will return @code{0} if
4435 the first frame pointer is properly initialized by the startup code.
4437 This function should only be used with a nonzero argument for debugging
4441 @node Vector Extensions
4442 @section Using vector instructions through built-in functions
4444 On some targets, the instruction set contains SIMD vector instructions that
4445 operate on multiple values contained in one large register at the same time.
4446 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4449 The first step in using these extensions is to provide the necessary data
4450 types. This should be done using an appropriate @code{typedef}:
4453 typedef int v4si __attribute__ ((vector_size (16)));
4456 The @code{int} type specifies the base type, while the attribute specifies
4457 the vector size for the variable, measured in bytes. For example, the
4458 declaration above causes the compiler to set the mode for the @code{v4si}
4459 type to be 16 bytes wide and divided into @code{int} sized units. For
4460 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4461 corresponding mode of @code{foo} will be @acronym{V4SI}.
4463 The @code{vector_size} attribute is only applicable to integral and
4464 float scalars, although arrays, pointers, and function return values
4465 are allowed in conjunction with this construct.
4467 All the basic integer types can be used as base types, both as signed
4468 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4469 @code{long long}. In addition, @code{float} and @code{double} can be
4470 used to build floating-point vector types.
4472 Specifying a combination that is not valid for the current architecture
4473 will cause GCC to synthesize the instructions using a narrower mode.
4474 For example, if you specify a variable of type @code{V4SI} and your
4475 architecture does not allow for this specific SIMD type, GCC will
4476 produce code that uses 4 @code{SIs}.
4478 The types defined in this manner can be used with a subset of normal C
4479 operations. Currently, GCC will allow using the following operators
4480 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4482 The operations behave like C++ @code{valarrays}. Addition is defined as
4483 the addition of the corresponding elements of the operands. For
4484 example, in the code below, each of the 4 elements in @var{a} will be
4485 added to the corresponding 4 elements in @var{b} and the resulting
4486 vector will be stored in @var{c}.
4489 typedef int v4si __attribute__ ((vector_size (16)));
4496 Subtraction, multiplication, division, and the logical operations
4497 operate in a similar manner. Likewise, the result of using the unary
4498 minus or complement operators on a vector type is a vector whose
4499 elements are the negative or complemented values of the corresponding
4500 elements in the operand.
4502 You can declare variables and use them in function calls and returns, as
4503 well as in assignments and some casts. You can specify a vector type as
4504 a return type for a function. Vector types can also be used as function
4505 arguments. It is possible to cast from one vector type to another,
4506 provided they are of the same size (in fact, you can also cast vectors
4507 to and from other datatypes of the same size).
4509 You cannot operate between vectors of different lengths or different
4510 signedness without a cast.
4512 A port that supports hardware vector operations, usually provides a set
4513 of built-in functions that can be used to operate on vectors. For
4514 example, a function to add two vectors and multiply the result by a
4515 third could look like this:
4518 v4si f (v4si a, v4si b, v4si c)
4520 v4si tmp = __builtin_addv4si (a, b);
4521 return __builtin_mulv4si (tmp, c);
4528 @findex __builtin_offsetof
4530 GCC implements for both C and C++ a syntactic extension to implement
4531 the @code{offsetof} macro.
4535 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
4537 offsetof_member_designator:
4539 | offsetof_member_designator "." @code{identifier}
4540 | offsetof_member_designator "[" @code{expr} "]"
4543 This extension is sufficient such that
4546 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
4549 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
4550 may be dependent. In either case, @var{member} may consist of a single
4551 identifier, or a sequence of member accesses and array references.
4553 @node Other Builtins
4554 @section Other built-in functions provided by GCC
4555 @cindex built-in functions
4556 @findex __builtin_isgreater
4557 @findex __builtin_isgreaterequal
4558 @findex __builtin_isless
4559 @findex __builtin_islessequal
4560 @findex __builtin_islessgreater
4561 @findex __builtin_isunordered
4562 @findex __builtin_powi
4563 @findex __builtin_powif
4564 @findex __builtin_powil
4719 @findex fprintf_unlocked
4721 @findex fputs_unlocked
4831 @findex printf_unlocked
4860 @findex significandf
4861 @findex significandl
4928 GCC provides a large number of built-in functions other than the ones
4929 mentioned above. Some of these are for internal use in the processing
4930 of exceptions or variable-length argument lists and will not be
4931 documented here because they may change from time to time; we do not
4932 recommend general use of these functions.
4934 The remaining functions are provided for optimization purposes.
4936 @opindex fno-builtin
4937 GCC includes built-in versions of many of the functions in the standard
4938 C library. The versions prefixed with @code{__builtin_} will always be
4939 treated as having the same meaning as the C library function even if you
4940 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
4941 Many of these functions are only optimized in certain cases; if they are
4942 not optimized in a particular case, a call to the library function will
4947 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
4948 @option{-std=c99}), the functions
4949 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
4950 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
4951 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
4952 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
4953 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
4954 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
4955 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
4956 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
4957 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
4958 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
4959 @code{significandf}, @code{significandl}, @code{significand},
4960 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
4961 @code{strdup}, @code{strfmon}, @code{toascii}, @code{y0f}, @code{y0l},
4962 @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
4964 may be handled as built-in functions.
4965 All these functions have corresponding versions
4966 prefixed with @code{__builtin_}, which may be used even in strict C89
4969 The ISO C99 functions
4970 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
4971 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
4972 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
4973 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
4974 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
4975 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
4976 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
4977 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
4978 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
4979 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
4980 @code{cimagl}, @code{cimag}, @code{conjf}, @code{conjl}, @code{conj},
4981 @code{copysignf}, @code{copysignl}, @code{copysign}, @code{cpowf},
4982 @code{cpowl}, @code{cpow}, @code{cprojf}, @code{cprojl}, @code{cproj},
4983 @code{crealf}, @code{creall}, @code{creal}, @code{csinf}, @code{csinhf},
4984 @code{csinhl}, @code{csinh}, @code{csinl}, @code{csin}, @code{csqrtf},
4985 @code{csqrtl}, @code{csqrt}, @code{ctanf}, @code{ctanhf}, @code{ctanhl},
4986 @code{ctanh}, @code{ctanl}, @code{ctan}, @code{erfcf}, @code{erfcl},
4987 @code{erfc}, @code{erff}, @code{erfl}, @code{erf}, @code{exp2f},
4988 @code{exp2l}, @code{exp2}, @code{expm1f}, @code{expm1l}, @code{expm1},
4989 @code{fdimf}, @code{fdiml}, @code{fdim}, @code{fmaf}, @code{fmal},
4990 @code{fmaxf}, @code{fmaxl}, @code{fmax}, @code{fma}, @code{fminf},
4991 @code{fminl}, @code{fmin}, @code{hypotf}, @code{hypotl}, @code{hypot},
4992 @code{ilogbf}, @code{ilogbl}, @code{ilogb}, @code{imaxabs},
4993 @code{isblank}, @code{iswblank}, @code{lgammaf}, @code{lgammal},
4994 @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
4995 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
4996 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
4997 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
4998 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
4999 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
5000 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
5001 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
5002 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
5003 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
5004 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
5005 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
5006 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
5007 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
5008 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
5009 are handled as built-in functions
5010 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5012 There are also built-in versions of the ISO C99 functions
5013 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5014 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5015 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5016 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5017 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5018 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5019 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5020 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5021 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5022 that are recognized in any mode since ISO C90 reserves these names for
5023 the purpose to which ISO C99 puts them. All these functions have
5024 corresponding versions prefixed with @code{__builtin_}.
5026 The ISO C94 functions
5027 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
5028 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
5029 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
5031 are handled as built-in functions
5032 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5034 The ISO C90 functions
5035 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5036 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5037 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5038 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
5039 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
5040 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
5041 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
5042 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
5043 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
5044 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
5045 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
5046 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
5047 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
5048 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
5049 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
5050 @code{vprintf} and @code{vsprintf}
5051 are all recognized as built-in functions unless
5052 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5053 is specified for an individual function). All of these functions have
5054 corresponding versions prefixed with @code{__builtin_}.
5056 GCC provides built-in versions of the ISO C99 floating point comparison
5057 macros that avoid raising exceptions for unordered operands. They have
5058 the same names as the standard macros ( @code{isgreater},
5059 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5060 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5061 prefixed. We intend for a library implementor to be able to simply
5062 @code{#define} each standard macro to its built-in equivalent.
5064 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
5066 You can use the built-in function @code{__builtin_types_compatible_p} to
5067 determine whether two types are the same.
5069 This built-in function returns 1 if the unqualified versions of the
5070 types @var{type1} and @var{type2} (which are types, not expressions) are
5071 compatible, 0 otherwise. The result of this built-in function can be
5072 used in integer constant expressions.
5074 This built-in function ignores top level qualifiers (e.g., @code{const},
5075 @code{volatile}). For example, @code{int} is equivalent to @code{const
5078 The type @code{int[]} and @code{int[5]} are compatible. On the other
5079 hand, @code{int} and @code{char *} are not compatible, even if the size
5080 of their types, on the particular architecture are the same. Also, the
5081 amount of pointer indirection is taken into account when determining
5082 similarity. Consequently, @code{short *} is not similar to
5083 @code{short **}. Furthermore, two types that are typedefed are
5084 considered compatible if their underlying types are compatible.
5086 An @code{enum} type is not considered to be compatible with another
5087 @code{enum} type even if both are compatible with the same integer
5088 type; this is what the C standard specifies.
5089 For example, @code{enum @{foo, bar@}} is not similar to
5090 @code{enum @{hot, dog@}}.
5092 You would typically use this function in code whose execution varies
5093 depending on the arguments' types. For example:
5099 if (__builtin_types_compatible_p (typeof (x), long double)) \
5100 tmp = foo_long_double (tmp); \
5101 else if (__builtin_types_compatible_p (typeof (x), double)) \
5102 tmp = foo_double (tmp); \
5103 else if (__builtin_types_compatible_p (typeof (x), float)) \
5104 tmp = foo_float (tmp); \
5111 @emph{Note:} This construct is only available for C@.
5115 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
5117 You can use the built-in function @code{__builtin_choose_expr} to
5118 evaluate code depending on the value of a constant expression. This
5119 built-in function returns @var{exp1} if @var{const_exp}, which is a
5120 constant expression that must be able to be determined at compile time,
5121 is nonzero. Otherwise it returns 0.
5123 This built-in function is analogous to the @samp{? :} operator in C,
5124 except that the expression returned has its type unaltered by promotion
5125 rules. Also, the built-in function does not evaluate the expression
5126 that was not chosen. For example, if @var{const_exp} evaluates to true,
5127 @var{exp2} is not evaluated even if it has side-effects.
5129 This built-in function can return an lvalue if the chosen argument is an
5132 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5133 type. Similarly, if @var{exp2} is returned, its return type is the same
5140 __builtin_choose_expr ( \
5141 __builtin_types_compatible_p (typeof (x), double), \
5143 __builtin_choose_expr ( \
5144 __builtin_types_compatible_p (typeof (x), float), \
5146 /* @r{The void expression results in a compile-time error} \
5147 @r{when assigning the result to something.} */ \
5151 @emph{Note:} This construct is only available for C@. Furthermore, the
5152 unused expression (@var{exp1} or @var{exp2} depending on the value of
5153 @var{const_exp}) may still generate syntax errors. This may change in
5158 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5159 You can use the built-in function @code{__builtin_constant_p} to
5160 determine if a value is known to be constant at compile-time and hence
5161 that GCC can perform constant-folding on expressions involving that
5162 value. The argument of the function is the value to test. The function
5163 returns the integer 1 if the argument is known to be a compile-time
5164 constant and 0 if it is not known to be a compile-time constant. A
5165 return of 0 does not indicate that the value is @emph{not} a constant,
5166 but merely that GCC cannot prove it is a constant with the specified
5167 value of the @option{-O} option.
5169 You would typically use this function in an embedded application where
5170 memory was a critical resource. If you have some complex calculation,
5171 you may want it to be folded if it involves constants, but need to call
5172 a function if it does not. For example:
5175 #define Scale_Value(X) \
5176 (__builtin_constant_p (X) \
5177 ? ((X) * SCALE + OFFSET) : Scale (X))
5180 You may use this built-in function in either a macro or an inline
5181 function. However, if you use it in an inlined function and pass an
5182 argument of the function as the argument to the built-in, GCC will
5183 never return 1 when you call the inline function with a string constant
5184 or compound literal (@pxref{Compound Literals}) and will not return 1
5185 when you pass a constant numeric value to the inline function unless you
5186 specify the @option{-O} option.
5188 You may also use @code{__builtin_constant_p} in initializers for static
5189 data. For instance, you can write
5192 static const int table[] = @{
5193 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5199 This is an acceptable initializer even if @var{EXPRESSION} is not a
5200 constant expression. GCC must be more conservative about evaluating the
5201 built-in in this case, because it has no opportunity to perform
5204 Previous versions of GCC did not accept this built-in in data
5205 initializers. The earliest version where it is completely safe is
5209 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5210 @opindex fprofile-arcs
5211 You may use @code{__builtin_expect} to provide the compiler with
5212 branch prediction information. In general, you should prefer to
5213 use actual profile feedback for this (@option{-fprofile-arcs}), as
5214 programmers are notoriously bad at predicting how their programs
5215 actually perform. However, there are applications in which this
5216 data is hard to collect.
5218 The return value is the value of @var{exp}, which should be an
5219 integral expression. The value of @var{c} must be a compile-time
5220 constant. The semantics of the built-in are that it is expected
5221 that @var{exp} == @var{c}. For example:
5224 if (__builtin_expect (x, 0))
5229 would indicate that we do not expect to call @code{foo}, since
5230 we expect @code{x} to be zero. Since you are limited to integral
5231 expressions for @var{exp}, you should use constructions such as
5234 if (__builtin_expect (ptr != NULL, 1))
5239 when testing pointer or floating-point values.
5242 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5243 This function is used to minimize cache-miss latency by moving data into
5244 a cache before it is accessed.
5245 You can insert calls to @code{__builtin_prefetch} into code for which
5246 you know addresses of data in memory that is likely to be accessed soon.
5247 If the target supports them, data prefetch instructions will be generated.
5248 If the prefetch is done early enough before the access then the data will
5249 be in the cache by the time it is accessed.
5251 The value of @var{addr} is the address of the memory to prefetch.
5252 There are two optional arguments, @var{rw} and @var{locality}.
5253 The value of @var{rw} is a compile-time constant one or zero; one
5254 means that the prefetch is preparing for a write to the memory address
5255 and zero, the default, means that the prefetch is preparing for a read.
5256 The value @var{locality} must be a compile-time constant integer between
5257 zero and three. A value of zero means that the data has no temporal
5258 locality, so it need not be left in the cache after the access. A value
5259 of three means that the data has a high degree of temporal locality and
5260 should be left in all levels of cache possible. Values of one and two
5261 mean, respectively, a low or moderate degree of temporal locality. The
5265 for (i = 0; i < n; i++)
5268 __builtin_prefetch (&a[i+j], 1, 1);
5269 __builtin_prefetch (&b[i+j], 0, 1);
5274 Data prefetch does not generate faults if @var{addr} is invalid, but
5275 the address expression itself must be valid. For example, a prefetch
5276 of @code{p->next} will not fault if @code{p->next} is not a valid
5277 address, but evaluation will fault if @code{p} is not a valid address.
5279 If the target does not support data prefetch, the address expression
5280 is evaluated if it includes side effects but no other code is generated
5281 and GCC does not issue a warning.
5284 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5285 Returns a positive infinity, if supported by the floating-point format,
5286 else @code{DBL_MAX}. This function is suitable for implementing the
5287 ISO C macro @code{HUGE_VAL}.
5290 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5291 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5294 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5295 Similar to @code{__builtin_huge_val}, except the return
5296 type is @code{long double}.
5299 @deftypefn {Built-in Function} double __builtin_inf (void)
5300 Similar to @code{__builtin_huge_val}, except a warning is generated
5301 if the target floating-point format does not support infinities.
5304 @deftypefn {Built-in Function} float __builtin_inff (void)
5305 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5306 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5309 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5310 Similar to @code{__builtin_inf}, except the return
5311 type is @code{long double}.
5314 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5315 This is an implementation of the ISO C99 function @code{nan}.
5317 Since ISO C99 defines this function in terms of @code{strtod}, which we
5318 do not implement, a description of the parsing is in order. The string
5319 is parsed as by @code{strtol}; that is, the base is recognized by
5320 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5321 in the significand such that the least significant bit of the number
5322 is at the least significant bit of the significand. The number is
5323 truncated to fit the significand field provided. The significand is
5324 forced to be a quiet NaN@.
5326 This function, if given a string literal, is evaluated early enough
5327 that it is considered a compile-time constant.
5330 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
5331 Similar to @code{__builtin_nan}, except the return type is @code{float}.
5334 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
5335 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
5338 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
5339 Similar to @code{__builtin_nan}, except the significand is forced
5340 to be a signaling NaN@. The @code{nans} function is proposed by
5341 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
5344 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
5345 Similar to @code{__builtin_nans}, except the return type is @code{float}.
5348 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
5349 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
5352 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
5353 Returns one plus the index of the least significant 1-bit of @var{x}, or
5354 if @var{x} is zero, returns zero.
5357 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
5358 Returns the number of leading 0-bits in @var{x}, starting at the most
5359 significant bit position. If @var{x} is 0, the result is undefined.
5362 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
5363 Returns the number of trailing 0-bits in @var{x}, starting at the least
5364 significant bit position. If @var{x} is 0, the result is undefined.
5367 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
5368 Returns the number of 1-bits in @var{x}.
5371 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
5372 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
5376 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
5377 Similar to @code{__builtin_ffs}, except the argument type is
5378 @code{unsigned long}.
5381 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
5382 Similar to @code{__builtin_clz}, except the argument type is
5383 @code{unsigned long}.
5386 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
5387 Similar to @code{__builtin_ctz}, except the argument type is
5388 @code{unsigned long}.
5391 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
5392 Similar to @code{__builtin_popcount}, except the argument type is
5393 @code{unsigned long}.
5396 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
5397 Similar to @code{__builtin_parity}, except the argument type is
5398 @code{unsigned long}.
5401 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
5402 Similar to @code{__builtin_ffs}, except the argument type is
5403 @code{unsigned long long}.
5406 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
5407 Similar to @code{__builtin_clz}, except the argument type is
5408 @code{unsigned long long}.
5411 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
5412 Similar to @code{__builtin_ctz}, except the argument type is
5413 @code{unsigned long long}.
5416 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
5417 Similar to @code{__builtin_popcount}, except the argument type is
5418 @code{unsigned long long}.
5421 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
5422 Similar to @code{__builtin_parity}, except the argument type is
5423 @code{unsigned long long}.
5426 @deftypefn {Built-in Function} double __builtin_powi (double, int)
5427 Returns the first argument raised to the power of the second. Unlike the
5428 @code{pow} function no guarantees about precision and rounding are made.
5431 @deftypefn {Built-in Function} float __builtin_powif (float, int)
5432 Similar to @code{__builtin_powi}, except the argument and return types
5436 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
5437 Similar to @code{__builtin_powi}, except the argument and return types
5438 are @code{long double}.
5442 @node Target Builtins
5443 @section Built-in Functions Specific to Particular Target Machines
5445 On some target machines, GCC supports many built-in functions specific
5446 to those machines. Generally these generate calls to specific machine
5447 instructions, but allow the compiler to schedule those calls.
5450 * Alpha Built-in Functions::
5451 * ARM Built-in Functions::
5452 * Blackfin Built-in Functions::
5453 * FR-V Built-in Functions::
5454 * X86 Built-in Functions::
5455 * MIPS Paired-Single Support::
5456 * PowerPC AltiVec Built-in Functions::
5457 * SPARC VIS Built-in Functions::
5460 @node Alpha Built-in Functions
5461 @subsection Alpha Built-in Functions
5463 These built-in functions are available for the Alpha family of
5464 processors, depending on the command-line switches used.
5466 The following built-in functions are always available. They
5467 all generate the machine instruction that is part of the name.
5470 long __builtin_alpha_implver (void)
5471 long __builtin_alpha_rpcc (void)
5472 long __builtin_alpha_amask (long)
5473 long __builtin_alpha_cmpbge (long, long)
5474 long __builtin_alpha_extbl (long, long)
5475 long __builtin_alpha_extwl (long, long)
5476 long __builtin_alpha_extll (long, long)
5477 long __builtin_alpha_extql (long, long)
5478 long __builtin_alpha_extwh (long, long)
5479 long __builtin_alpha_extlh (long, long)
5480 long __builtin_alpha_extqh (long, long)
5481 long __builtin_alpha_insbl (long, long)
5482 long __builtin_alpha_inswl (long, long)
5483 long __builtin_alpha_insll (long, long)
5484 long __builtin_alpha_insql (long, long)
5485 long __builtin_alpha_inswh (long, long)
5486 long __builtin_alpha_inslh (long, long)
5487 long __builtin_alpha_insqh (long, long)
5488 long __builtin_alpha_mskbl (long, long)
5489 long __builtin_alpha_mskwl (long, long)
5490 long __builtin_alpha_mskll (long, long)
5491 long __builtin_alpha_mskql (long, long)
5492 long __builtin_alpha_mskwh (long, long)
5493 long __builtin_alpha_msklh (long, long)
5494 long __builtin_alpha_mskqh (long, long)
5495 long __builtin_alpha_umulh (long, long)
5496 long __builtin_alpha_zap (long, long)
5497 long __builtin_alpha_zapnot (long, long)
5500 The following built-in functions are always with @option{-mmax}
5501 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
5502 later. They all generate the machine instruction that is part
5506 long __builtin_alpha_pklb (long)
5507 long __builtin_alpha_pkwb (long)
5508 long __builtin_alpha_unpkbl (long)
5509 long __builtin_alpha_unpkbw (long)
5510 long __builtin_alpha_minub8 (long, long)
5511 long __builtin_alpha_minsb8 (long, long)
5512 long __builtin_alpha_minuw4 (long, long)
5513 long __builtin_alpha_minsw4 (long, long)
5514 long __builtin_alpha_maxub8 (long, long)
5515 long __builtin_alpha_maxsb8 (long, long)
5516 long __builtin_alpha_maxuw4 (long, long)
5517 long __builtin_alpha_maxsw4 (long, long)
5518 long __builtin_alpha_perr (long, long)
5521 The following built-in functions are always with @option{-mcix}
5522 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
5523 later. They all generate the machine instruction that is part
5527 long __builtin_alpha_cttz (long)
5528 long __builtin_alpha_ctlz (long)
5529 long __builtin_alpha_ctpop (long)
5532 The following builtins are available on systems that use the OSF/1
5533 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
5534 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
5535 @code{rdval} and @code{wrval}.
5538 void *__builtin_thread_pointer (void)
5539 void __builtin_set_thread_pointer (void *)
5542 @node ARM Built-in Functions
5543 @subsection ARM Built-in Functions
5545 These built-in functions are available for the ARM family of
5546 processors, when the @option{-mcpu=iwmmxt} switch is used:
5549 typedef int v2si __attribute__ ((vector_size (8)));
5550 typedef short v4hi __attribute__ ((vector_size (8)));
5551 typedef char v8qi __attribute__ ((vector_size (8)));
5553 int __builtin_arm_getwcx (int)
5554 void __builtin_arm_setwcx (int, int)
5555 int __builtin_arm_textrmsb (v8qi, int)
5556 int __builtin_arm_textrmsh (v4hi, int)
5557 int __builtin_arm_textrmsw (v2si, int)
5558 int __builtin_arm_textrmub (v8qi, int)
5559 int __builtin_arm_textrmuh (v4hi, int)
5560 int __builtin_arm_textrmuw (v2si, int)
5561 v8qi __builtin_arm_tinsrb (v8qi, int)
5562 v4hi __builtin_arm_tinsrh (v4hi, int)
5563 v2si __builtin_arm_tinsrw (v2si, int)
5564 long long __builtin_arm_tmia (long long, int, int)
5565 long long __builtin_arm_tmiabb (long long, int, int)
5566 long long __builtin_arm_tmiabt (long long, int, int)
5567 long long __builtin_arm_tmiaph (long long, int, int)
5568 long long __builtin_arm_tmiatb (long long, int, int)
5569 long long __builtin_arm_tmiatt (long long, int, int)
5570 int __builtin_arm_tmovmskb (v8qi)
5571 int __builtin_arm_tmovmskh (v4hi)
5572 int __builtin_arm_tmovmskw (v2si)
5573 long long __builtin_arm_waccb (v8qi)
5574 long long __builtin_arm_wacch (v4hi)
5575 long long __builtin_arm_waccw (v2si)
5576 v8qi __builtin_arm_waddb (v8qi, v8qi)
5577 v8qi __builtin_arm_waddbss (v8qi, v8qi)
5578 v8qi __builtin_arm_waddbus (v8qi, v8qi)
5579 v4hi __builtin_arm_waddh (v4hi, v4hi)
5580 v4hi __builtin_arm_waddhss (v4hi, v4hi)
5581 v4hi __builtin_arm_waddhus (v4hi, v4hi)
5582 v2si __builtin_arm_waddw (v2si, v2si)
5583 v2si __builtin_arm_waddwss (v2si, v2si)
5584 v2si __builtin_arm_waddwus (v2si, v2si)
5585 v8qi __builtin_arm_walign (v8qi, v8qi, int)
5586 long long __builtin_arm_wand(long long, long long)
5587 long long __builtin_arm_wandn (long long, long long)
5588 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
5589 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
5590 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
5591 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
5592 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
5593 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
5594 v2si __builtin_arm_wcmpeqw (v2si, v2si)
5595 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
5596 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
5597 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
5598 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
5599 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
5600 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
5601 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
5602 long long __builtin_arm_wmacsz (v4hi, v4hi)
5603 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
5604 long long __builtin_arm_wmacuz (v4hi, v4hi)
5605 v4hi __builtin_arm_wmadds (v4hi, v4hi)
5606 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
5607 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
5608 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
5609 v2si __builtin_arm_wmaxsw (v2si, v2si)
5610 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
5611 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
5612 v2si __builtin_arm_wmaxuw (v2si, v2si)
5613 v8qi __builtin_arm_wminsb (v8qi, v8qi)
5614 v4hi __builtin_arm_wminsh (v4hi, v4hi)
5615 v2si __builtin_arm_wminsw (v2si, v2si)
5616 v8qi __builtin_arm_wminub (v8qi, v8qi)
5617 v4hi __builtin_arm_wminuh (v4hi, v4hi)
5618 v2si __builtin_arm_wminuw (v2si, v2si)
5619 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
5620 v4hi __builtin_arm_wmulul (v4hi, v4hi)
5621 v4hi __builtin_arm_wmulum (v4hi, v4hi)
5622 long long __builtin_arm_wor (long long, long long)
5623 v2si __builtin_arm_wpackdss (long long, long long)
5624 v2si __builtin_arm_wpackdus (long long, long long)
5625 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
5626 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
5627 v4hi __builtin_arm_wpackwss (v2si, v2si)
5628 v4hi __builtin_arm_wpackwus (v2si, v2si)
5629 long long __builtin_arm_wrord (long long, long long)
5630 long long __builtin_arm_wrordi (long long, int)
5631 v4hi __builtin_arm_wrorh (v4hi, long long)
5632 v4hi __builtin_arm_wrorhi (v4hi, int)
5633 v2si __builtin_arm_wrorw (v2si, long long)
5634 v2si __builtin_arm_wrorwi (v2si, int)
5635 v2si __builtin_arm_wsadb (v8qi, v8qi)
5636 v2si __builtin_arm_wsadbz (v8qi, v8qi)
5637 v2si __builtin_arm_wsadh (v4hi, v4hi)
5638 v2si __builtin_arm_wsadhz (v4hi, v4hi)
5639 v4hi __builtin_arm_wshufh (v4hi, int)
5640 long long __builtin_arm_wslld (long long, long long)
5641 long long __builtin_arm_wslldi (long long, int)
5642 v4hi __builtin_arm_wsllh (v4hi, long long)
5643 v4hi __builtin_arm_wsllhi (v4hi, int)
5644 v2si __builtin_arm_wsllw (v2si, long long)
5645 v2si __builtin_arm_wsllwi (v2si, int)
5646 long long __builtin_arm_wsrad (long long, long long)
5647 long long __builtin_arm_wsradi (long long, int)
5648 v4hi __builtin_arm_wsrah (v4hi, long long)
5649 v4hi __builtin_arm_wsrahi (v4hi, int)
5650 v2si __builtin_arm_wsraw (v2si, long long)
5651 v2si __builtin_arm_wsrawi (v2si, int)
5652 long long __builtin_arm_wsrld (long long, long long)
5653 long long __builtin_arm_wsrldi (long long, int)
5654 v4hi __builtin_arm_wsrlh (v4hi, long long)
5655 v4hi __builtin_arm_wsrlhi (v4hi, int)
5656 v2si __builtin_arm_wsrlw (v2si, long long)
5657 v2si __builtin_arm_wsrlwi (v2si, int)
5658 v8qi __builtin_arm_wsubb (v8qi, v8qi)
5659 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
5660 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
5661 v4hi __builtin_arm_wsubh (v4hi, v4hi)
5662 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
5663 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
5664 v2si __builtin_arm_wsubw (v2si, v2si)
5665 v2si __builtin_arm_wsubwss (v2si, v2si)
5666 v2si __builtin_arm_wsubwus (v2si, v2si)
5667 v4hi __builtin_arm_wunpckehsb (v8qi)
5668 v2si __builtin_arm_wunpckehsh (v4hi)
5669 long long __builtin_arm_wunpckehsw (v2si)
5670 v4hi __builtin_arm_wunpckehub (v8qi)
5671 v2si __builtin_arm_wunpckehuh (v4hi)
5672 long long __builtin_arm_wunpckehuw (v2si)
5673 v4hi __builtin_arm_wunpckelsb (v8qi)
5674 v2si __builtin_arm_wunpckelsh (v4hi)
5675 long long __builtin_arm_wunpckelsw (v2si)
5676 v4hi __builtin_arm_wunpckelub (v8qi)
5677 v2si __builtin_arm_wunpckeluh (v4hi)
5678 long long __builtin_arm_wunpckeluw (v2si)
5679 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
5680 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
5681 v2si __builtin_arm_wunpckihw (v2si, v2si)
5682 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
5683 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
5684 v2si __builtin_arm_wunpckilw (v2si, v2si)
5685 long long __builtin_arm_wxor (long long, long long)
5686 long long __builtin_arm_wzero ()
5689 @node Blackfin Built-in Functions
5690 @subsection Blackfin Built-in Functions
5692 Currently, there are two Blackfin-specific built-in functions. These are
5693 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
5694 using inline assembly; by using these built-in functions the compiler can
5695 automatically add workarounds for hardware errata involving these
5696 instructions. These functions are named as follows:
5699 void __builtin_bfin_csync (void)
5700 void __builtin_bfin_ssync (void)
5703 @node FR-V Built-in Functions
5704 @subsection FR-V Built-in Functions
5706 GCC provides many FR-V-specific built-in functions. In general,
5707 these functions are intended to be compatible with those described
5708 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
5709 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
5710 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
5711 pointer rather than by value.
5713 Most of the functions are named after specific FR-V instructions.
5714 Such functions are said to be ``directly mapped'' and are summarized
5715 here in tabular form.
5719 * Directly-mapped Integer Functions::
5720 * Directly-mapped Media Functions::
5721 * Other Built-in Functions::
5724 @node Argument Types
5725 @subsubsection Argument Types
5727 The arguments to the built-in functions can be divided into three groups:
5728 register numbers, compile-time constants and run-time values. In order
5729 to make this classification clear at a glance, the arguments and return
5730 values are given the following pseudo types:
5732 @multitable @columnfractions .20 .30 .15 .35
5733 @item Pseudo type @tab Real C type @tab Constant? @tab Description
5734 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
5735 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
5736 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
5737 @item @code{uw2} @tab @code{unsigned long long} @tab No
5738 @tab an unsigned doubleword
5739 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
5740 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
5741 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
5742 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
5745 These pseudo types are not defined by GCC, they are simply a notational
5746 convenience used in this manual.
5748 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
5749 and @code{sw2} are evaluated at run time. They correspond to
5750 register operands in the underlying FR-V instructions.
5752 @code{const} arguments represent immediate operands in the underlying
5753 FR-V instructions. They must be compile-time constants.
5755 @code{acc} arguments are evaluated at compile time and specify the number
5756 of an accumulator register. For example, an @code{acc} argument of 2
5757 will select the ACC2 register.
5759 @code{iacc} arguments are similar to @code{acc} arguments but specify the
5760 number of an IACC register. See @pxref{Other Built-in Functions}
5763 @node Directly-mapped Integer Functions
5764 @subsubsection Directly-mapped Integer Functions
5766 The functions listed below map directly to FR-V I-type instructions.
5768 @multitable @columnfractions .45 .32 .23
5769 @item Function prototype @tab Example usage @tab Assembly output
5770 @item @code{sw1 __ADDSS (sw1, sw1)}
5771 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
5772 @tab @code{ADDSS @var{a},@var{b},@var{c}}
5773 @item @code{sw1 __SCAN (sw1, sw1)}
5774 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
5775 @tab @code{SCAN @var{a},@var{b},@var{c}}
5776 @item @code{sw1 __SCUTSS (sw1)}
5777 @tab @code{@var{b} = __SCUTSS (@var{a})}
5778 @tab @code{SCUTSS @var{a},@var{b}}
5779 @item @code{sw1 __SLASS (sw1, sw1)}
5780 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
5781 @tab @code{SLASS @var{a},@var{b},@var{c}}
5782 @item @code{void __SMASS (sw1, sw1)}
5783 @tab @code{__SMASS (@var{a}, @var{b})}
5784 @tab @code{SMASS @var{a},@var{b}}
5785 @item @code{void __SMSSS (sw1, sw1)}
5786 @tab @code{__SMSSS (@var{a}, @var{b})}
5787 @tab @code{SMSSS @var{a},@var{b}}
5788 @item @code{void __SMU (sw1, sw1)}
5789 @tab @code{__SMU (@var{a}, @var{b})}
5790 @tab @code{SMU @var{a},@var{b}}
5791 @item @code{sw2 __SMUL (sw1, sw1)}
5792 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
5793 @tab @code{SMUL @var{a},@var{b},@var{c}}
5794 @item @code{sw1 __SUBSS (sw1, sw1)}
5795 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
5796 @tab @code{SUBSS @var{a},@var{b},@var{c}}
5797 @item @code{uw2 __UMUL (uw1, uw1)}
5798 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
5799 @tab @code{UMUL @var{a},@var{b},@var{c}}
5802 @node Directly-mapped Media Functions
5803 @subsubsection Directly-mapped Media Functions
5805 The functions listed below map directly to FR-V M-type instructions.
5807 @multitable @columnfractions .45 .32 .23
5808 @item Function prototype @tab Example usage @tab Assembly output
5809 @item @code{uw1 __MABSHS (sw1)}
5810 @tab @code{@var{b} = __MABSHS (@var{a})}
5811 @tab @code{MABSHS @var{a},@var{b}}
5812 @item @code{void __MADDACCS (acc, acc)}
5813 @tab @code{__MADDACCS (@var{b}, @var{a})}
5814 @tab @code{MADDACCS @var{a},@var{b}}
5815 @item @code{sw1 __MADDHSS (sw1, sw1)}
5816 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
5817 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
5818 @item @code{uw1 __MADDHUS (uw1, uw1)}
5819 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
5820 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
5821 @item @code{uw1 __MAND (uw1, uw1)}
5822 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
5823 @tab @code{MAND @var{a},@var{b},@var{c}}
5824 @item @code{void __MASACCS (acc, acc)}
5825 @tab @code{__MASACCS (@var{b}, @var{a})}
5826 @tab @code{MASACCS @var{a},@var{b}}
5827 @item @code{uw1 __MAVEH (uw1, uw1)}
5828 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
5829 @tab @code{MAVEH @var{a},@var{b},@var{c}}
5830 @item @code{uw2 __MBTOH (uw1)}
5831 @tab @code{@var{b} = __MBTOH (@var{a})}
5832 @tab @code{MBTOH @var{a},@var{b}}
5833 @item @code{void __MBTOHE (uw1 *, uw1)}
5834 @tab @code{__MBTOHE (&@var{b}, @var{a})}
5835 @tab @code{MBTOHE @var{a},@var{b}}
5836 @item @code{void __MCLRACC (acc)}
5837 @tab @code{__MCLRACC (@var{a})}
5838 @tab @code{MCLRACC @var{a}}
5839 @item @code{void __MCLRACCA (void)}
5840 @tab @code{__MCLRACCA ()}
5841 @tab @code{MCLRACCA}
5842 @item @code{uw1 __Mcop1 (uw1, uw1)}
5843 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
5844 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
5845 @item @code{uw1 __Mcop2 (uw1, uw1)}
5846 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
5847 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
5848 @item @code{uw1 __MCPLHI (uw2, const)}
5849 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
5850 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
5851 @item @code{uw1 __MCPLI (uw2, const)}
5852 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
5853 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
5854 @item @code{void __MCPXIS (acc, sw1, sw1)}
5855 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
5856 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
5857 @item @code{void __MCPXIU (acc, uw1, uw1)}
5858 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
5859 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
5860 @item @code{void __MCPXRS (acc, sw1, sw1)}
5861 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
5862 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
5863 @item @code{void __MCPXRU (acc, uw1, uw1)}
5864 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
5865 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
5866 @item @code{uw1 __MCUT (acc, uw1)}
5867 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
5868 @tab @code{MCUT @var{a},@var{b},@var{c}}
5869 @item @code{uw1 __MCUTSS (acc, sw1)}
5870 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
5871 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
5872 @item @code{void __MDADDACCS (acc, acc)}
5873 @tab @code{__MDADDACCS (@var{b}, @var{a})}
5874 @tab @code{MDADDACCS @var{a},@var{b}}
5875 @item @code{void __MDASACCS (acc, acc)}
5876 @tab @code{__MDASACCS (@var{b}, @var{a})}
5877 @tab @code{MDASACCS @var{a},@var{b}}
5878 @item @code{uw2 __MDCUTSSI (acc, const)}
5879 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
5880 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
5881 @item @code{uw2 __MDPACKH (uw2, uw2)}
5882 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
5883 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
5884 @item @code{uw2 __MDROTLI (uw2, const)}
5885 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
5886 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
5887 @item @code{void __MDSUBACCS (acc, acc)}
5888 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
5889 @tab @code{MDSUBACCS @var{a},@var{b}}
5890 @item @code{void __MDUNPACKH (uw1 *, uw2)}
5891 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
5892 @tab @code{MDUNPACKH @var{a},@var{b}}
5893 @item @code{uw2 __MEXPDHD (uw1, const)}
5894 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
5895 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
5896 @item @code{uw1 __MEXPDHW (uw1, const)}
5897 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
5898 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
5899 @item @code{uw1 __MHDSETH (uw1, const)}
5900 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
5901 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
5902 @item @code{sw1 __MHDSETS (const)}
5903 @tab @code{@var{b} = __MHDSETS (@var{a})}
5904 @tab @code{MHDSETS #@var{a},@var{b}}
5905 @item @code{uw1 __MHSETHIH (uw1, const)}
5906 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
5907 @tab @code{MHSETHIH #@var{a},@var{b}}
5908 @item @code{sw1 __MHSETHIS (sw1, const)}
5909 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
5910 @tab @code{MHSETHIS #@var{a},@var{b}}
5911 @item @code{uw1 __MHSETLOH (uw1, const)}
5912 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
5913 @tab @code{MHSETLOH #@var{a},@var{b}}
5914 @item @code{sw1 __MHSETLOS (sw1, const)}
5915 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
5916 @tab @code{MHSETLOS #@var{a},@var{b}}
5917 @item @code{uw1 __MHTOB (uw2)}
5918 @tab @code{@var{b} = __MHTOB (@var{a})}
5919 @tab @code{MHTOB @var{a},@var{b}}
5920 @item @code{void __MMACHS (acc, sw1, sw1)}
5921 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
5922 @tab @code{MMACHS @var{a},@var{b},@var{c}}
5923 @item @code{void __MMACHU (acc, uw1, uw1)}
5924 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
5925 @tab @code{MMACHU @var{a},@var{b},@var{c}}
5926 @item @code{void __MMRDHS (acc, sw1, sw1)}
5927 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
5928 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
5929 @item @code{void __MMRDHU (acc, uw1, uw1)}
5930 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
5931 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
5932 @item @code{void __MMULHS (acc, sw1, sw1)}
5933 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
5934 @tab @code{MMULHS @var{a},@var{b},@var{c}}
5935 @item @code{void __MMULHU (acc, uw1, uw1)}
5936 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
5937 @tab @code{MMULHU @var{a},@var{b},@var{c}}
5938 @item @code{void __MMULXHS (acc, sw1, sw1)}
5939 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
5940 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
5941 @item @code{void __MMULXHU (acc, uw1, uw1)}
5942 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
5943 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
5944 @item @code{uw1 __MNOT (uw1)}
5945 @tab @code{@var{b} = __MNOT (@var{a})}
5946 @tab @code{MNOT @var{a},@var{b}}
5947 @item @code{uw1 __MOR (uw1, uw1)}
5948 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
5949 @tab @code{MOR @var{a},@var{b},@var{c}}
5950 @item @code{uw1 __MPACKH (uh, uh)}
5951 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
5952 @tab @code{MPACKH @var{a},@var{b},@var{c}}
5953 @item @code{sw2 __MQADDHSS (sw2, sw2)}
5954 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
5955 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
5956 @item @code{uw2 __MQADDHUS (uw2, uw2)}
5957 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
5958 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
5959 @item @code{void __MQCPXIS (acc, sw2, sw2)}
5960 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
5961 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
5962 @item @code{void __MQCPXIU (acc, uw2, uw2)}
5963 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
5964 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
5965 @item @code{void __MQCPXRS (acc, sw2, sw2)}
5966 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
5967 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
5968 @item @code{void __MQCPXRU (acc, uw2, uw2)}
5969 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
5970 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
5971 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
5972 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
5973 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
5974 @item @code{sw2 __MQLMTHS (sw2, sw2)}
5975 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
5976 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
5977 @item @code{void __MQMACHS (acc, sw2, sw2)}
5978 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
5979 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
5980 @item @code{void __MQMACHU (acc, uw2, uw2)}
5981 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
5982 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
5983 @item @code{void __MQMACXHS (acc, sw2, sw2)}
5984 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
5985 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
5986 @item @code{void __MQMULHS (acc, sw2, sw2)}
5987 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
5988 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
5989 @item @code{void __MQMULHU (acc, uw2, uw2)}
5990 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
5991 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
5992 @item @code{void __MQMULXHS (acc, sw2, sw2)}
5993 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
5994 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
5995 @item @code{void __MQMULXHU (acc, uw2, uw2)}
5996 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
5997 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
5998 @item @code{sw2 __MQSATHS (sw2, sw2)}
5999 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
6000 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
6001 @item @code{uw2 __MQSLLHI (uw2, int)}
6002 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
6003 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
6004 @item @code{sw2 __MQSRAHI (sw2, int)}
6005 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
6006 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
6007 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
6008 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
6009 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
6010 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
6011 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
6012 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
6013 @item @code{void __MQXMACHS (acc, sw2, sw2)}
6014 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
6015 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
6016 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
6017 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
6018 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
6019 @item @code{uw1 __MRDACC (acc)}
6020 @tab @code{@var{b} = __MRDACC (@var{a})}
6021 @tab @code{MRDACC @var{a},@var{b}}
6022 @item @code{uw1 __MRDACCG (acc)}
6023 @tab @code{@var{b} = __MRDACCG (@var{a})}
6024 @tab @code{MRDACCG @var{a},@var{b}}
6025 @item @code{uw1 __MROTLI (uw1, const)}
6026 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
6027 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
6028 @item @code{uw1 __MROTRI (uw1, const)}
6029 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
6030 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
6031 @item @code{sw1 __MSATHS (sw1, sw1)}
6032 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
6033 @tab @code{MSATHS @var{a},@var{b},@var{c}}
6034 @item @code{uw1 __MSATHU (uw1, uw1)}
6035 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
6036 @tab @code{MSATHU @var{a},@var{b},@var{c}}
6037 @item @code{uw1 __MSLLHI (uw1, const)}
6038 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
6039 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
6040 @item @code{sw1 __MSRAHI (sw1, const)}
6041 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
6042 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
6043 @item @code{uw1 __MSRLHI (uw1, const)}
6044 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
6045 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
6046 @item @code{void __MSUBACCS (acc, acc)}
6047 @tab @code{__MSUBACCS (@var{b}, @var{a})}
6048 @tab @code{MSUBACCS @var{a},@var{b}}
6049 @item @code{sw1 __MSUBHSS (sw1, sw1)}
6050 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
6051 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
6052 @item @code{uw1 __MSUBHUS (uw1, uw1)}
6053 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
6054 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
6055 @item @code{void __MTRAP (void)}
6056 @tab @code{__MTRAP ()}
6058 @item @code{uw2 __MUNPACKH (uw1)}
6059 @tab @code{@var{b} = __MUNPACKH (@var{a})}
6060 @tab @code{MUNPACKH @var{a},@var{b}}
6061 @item @code{uw1 __MWCUT (uw2, uw1)}
6062 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
6063 @tab @code{MWCUT @var{a},@var{b},@var{c}}
6064 @item @code{void __MWTACC (acc, uw1)}
6065 @tab @code{__MWTACC (@var{b}, @var{a})}
6066 @tab @code{MWTACC @var{a},@var{b}}
6067 @item @code{void __MWTACCG (acc, uw1)}
6068 @tab @code{__MWTACCG (@var{b}, @var{a})}
6069 @tab @code{MWTACCG @var{a},@var{b}}
6070 @item @code{uw1 __MXOR (uw1, uw1)}
6071 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
6072 @tab @code{MXOR @var{a},@var{b},@var{c}}
6075 @node Other Built-in Functions
6076 @subsubsection Other Built-in Functions
6078 This section describes built-in functions that are not named after
6079 a specific FR-V instruction.
6082 @item sw2 __IACCreadll (iacc @var{reg})
6083 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
6084 for future expansion and must be 0.
6086 @item sw1 __IACCreadl (iacc @var{reg})
6087 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
6088 Other values of @var{reg} are rejected as invalid.
6090 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
6091 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
6092 is reserved for future expansion and must be 0.
6094 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
6095 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
6096 is 1. Other values of @var{reg} are rejected as invalid.
6098 @item void __data_prefetch0 (const void *@var{x})
6099 Use the @code{dcpl} instruction to load the contents of address @var{x}
6100 into the data cache.
6102 @item void __data_prefetch (const void *@var{x})
6103 Use the @code{nldub} instruction to load the contents of address @var{x}
6104 into the data cache. The instruction will be issued in slot I1@.
6107 @node X86 Built-in Functions
6108 @subsection X86 Built-in Functions
6110 These built-in functions are available for the i386 and x86-64 family
6111 of computers, depending on the command-line switches used.
6113 The following machine modes are available for use with MMX built-in functions
6114 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
6115 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
6116 vector of eight 8-bit integers. Some of the built-in functions operate on
6117 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
6119 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
6120 of two 32-bit floating point values.
6122 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
6123 floating point values. Some instructions use a vector of four 32-bit
6124 integers, these use @code{V4SI}. Finally, some instructions operate on an
6125 entire vector register, interpreting it as a 128-bit integer, these use mode
6128 The following built-in functions are made available by @option{-mmmx}.
6129 All of them generate the machine instruction that is part of the name.
6132 v8qi __builtin_ia32_paddb (v8qi, v8qi)
6133 v4hi __builtin_ia32_paddw (v4hi, v4hi)
6134 v2si __builtin_ia32_paddd (v2si, v2si)
6135 v8qi __builtin_ia32_psubb (v8qi, v8qi)
6136 v4hi __builtin_ia32_psubw (v4hi, v4hi)
6137 v2si __builtin_ia32_psubd (v2si, v2si)
6138 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
6139 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
6140 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
6141 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
6142 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
6143 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
6144 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
6145 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
6146 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
6147 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
6148 di __builtin_ia32_pand (di, di)
6149 di __builtin_ia32_pandn (di,di)
6150 di __builtin_ia32_por (di, di)
6151 di __builtin_ia32_pxor (di, di)
6152 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
6153 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
6154 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
6155 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
6156 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
6157 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
6158 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
6159 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
6160 v2si __builtin_ia32_punpckhdq (v2si, v2si)
6161 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
6162 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
6163 v2si __builtin_ia32_punpckldq (v2si, v2si)
6164 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
6165 v4hi __builtin_ia32_packssdw (v2si, v2si)
6166 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
6169 The following built-in functions are made available either with
6170 @option{-msse}, or with a combination of @option{-m3dnow} and
6171 @option{-march=athlon}. All of them generate the machine
6172 instruction that is part of the name.
6175 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
6176 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
6177 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
6178 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
6179 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
6180 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
6181 v8qi __builtin_ia32_pminub (v8qi, v8qi)
6182 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
6183 int __builtin_ia32_pextrw (v4hi, int)
6184 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
6185 int __builtin_ia32_pmovmskb (v8qi)
6186 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
6187 void __builtin_ia32_movntq (di *, di)
6188 void __builtin_ia32_sfence (void)
6191 The following built-in functions are available when @option{-msse} is used.
6192 All of them generate the machine instruction that is part of the name.
6195 int __builtin_ia32_comieq (v4sf, v4sf)
6196 int __builtin_ia32_comineq (v4sf, v4sf)
6197 int __builtin_ia32_comilt (v4sf, v4sf)
6198 int __builtin_ia32_comile (v4sf, v4sf)
6199 int __builtin_ia32_comigt (v4sf, v4sf)
6200 int __builtin_ia32_comige (v4sf, v4sf)
6201 int __builtin_ia32_ucomieq (v4sf, v4sf)
6202 int __builtin_ia32_ucomineq (v4sf, v4sf)
6203 int __builtin_ia32_ucomilt (v4sf, v4sf)
6204 int __builtin_ia32_ucomile (v4sf, v4sf)
6205 int __builtin_ia32_ucomigt (v4sf, v4sf)
6206 int __builtin_ia32_ucomige (v4sf, v4sf)
6207 v4sf __builtin_ia32_addps (v4sf, v4sf)
6208 v4sf __builtin_ia32_subps (v4sf, v4sf)
6209 v4sf __builtin_ia32_mulps (v4sf, v4sf)
6210 v4sf __builtin_ia32_divps (v4sf, v4sf)
6211 v4sf __builtin_ia32_addss (v4sf, v4sf)
6212 v4sf __builtin_ia32_subss (v4sf, v4sf)
6213 v4sf __builtin_ia32_mulss (v4sf, v4sf)
6214 v4sf __builtin_ia32_divss (v4sf, v4sf)
6215 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
6216 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
6217 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
6218 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
6219 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
6220 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
6221 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
6222 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
6223 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
6224 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
6225 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
6226 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
6227 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
6228 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
6229 v4si __builtin_ia32_cmpless (v4sf, v4sf)
6230 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
6231 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
6232 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
6233 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
6234 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
6235 v4sf __builtin_ia32_maxps (v4sf, v4sf)
6236 v4sf __builtin_ia32_maxss (v4sf, v4sf)
6237 v4sf __builtin_ia32_minps (v4sf, v4sf)
6238 v4sf __builtin_ia32_minss (v4sf, v4sf)
6239 v4sf __builtin_ia32_andps (v4sf, v4sf)
6240 v4sf __builtin_ia32_andnps (v4sf, v4sf)
6241 v4sf __builtin_ia32_orps (v4sf, v4sf)
6242 v4sf __builtin_ia32_xorps (v4sf, v4sf)
6243 v4sf __builtin_ia32_movss (v4sf, v4sf)
6244 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
6245 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
6246 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
6247 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
6248 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
6249 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
6250 v2si __builtin_ia32_cvtps2pi (v4sf)
6251 int __builtin_ia32_cvtss2si (v4sf)
6252 v2si __builtin_ia32_cvttps2pi (v4sf)
6253 int __builtin_ia32_cvttss2si (v4sf)
6254 v4sf __builtin_ia32_rcpps (v4sf)
6255 v4sf __builtin_ia32_rsqrtps (v4sf)
6256 v4sf __builtin_ia32_sqrtps (v4sf)
6257 v4sf __builtin_ia32_rcpss (v4sf)
6258 v4sf __builtin_ia32_rsqrtss (v4sf)
6259 v4sf __builtin_ia32_sqrtss (v4sf)
6260 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
6261 void __builtin_ia32_movntps (float *, v4sf)
6262 int __builtin_ia32_movmskps (v4sf)
6265 The following built-in functions are available when @option{-msse} is used.
6268 @item v4sf __builtin_ia32_loadaps (float *)
6269 Generates the @code{movaps} machine instruction as a load from memory.
6270 @item void __builtin_ia32_storeaps (float *, v4sf)
6271 Generates the @code{movaps} machine instruction as a store to memory.
6272 @item v4sf __builtin_ia32_loadups (float *)
6273 Generates the @code{movups} machine instruction as a load from memory.
6274 @item void __builtin_ia32_storeups (float *, v4sf)
6275 Generates the @code{movups} machine instruction as a store to memory.
6276 @item v4sf __builtin_ia32_loadsss (float *)
6277 Generates the @code{movss} machine instruction as a load from memory.
6278 @item void __builtin_ia32_storess (float *, v4sf)
6279 Generates the @code{movss} machine instruction as a store to memory.
6280 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
6281 Generates the @code{movhps} machine instruction as a load from memory.
6282 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
6283 Generates the @code{movlps} machine instruction as a load from memory
6284 @item void __builtin_ia32_storehps (v4sf, v2si *)
6285 Generates the @code{movhps} machine instruction as a store to memory.
6286 @item void __builtin_ia32_storelps (v4sf, v2si *)
6287 Generates the @code{movlps} machine instruction as a store to memory.
6290 The following built-in functions are available when @option{-msse2} is used.
6291 All of them generate the machine instruction that is part of the name.
6294 int __builtin_ia32_comisdeq (v2df, v2df)
6295 int __builtin_ia32_comisdlt (v2df, v2df)
6296 int __builtin_ia32_comisdle (v2df, v2df)
6297 int __builtin_ia32_comisdgt (v2df, v2df)
6298 int __builtin_ia32_comisdge (v2df, v2df)
6299 int __builtin_ia32_comisdneq (v2df, v2df)
6300 int __builtin_ia32_ucomisdeq (v2df, v2df)
6301 int __builtin_ia32_ucomisdlt (v2df, v2df)
6302 int __builtin_ia32_ucomisdle (v2df, v2df)
6303 int __builtin_ia32_ucomisdgt (v2df, v2df)
6304 int __builtin_ia32_ucomisdge (v2df, v2df)
6305 int __builtin_ia32_ucomisdneq (v2df, v2df)
6306 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
6307 v2df __builtin_ia32_cmpltpd (v2df, v2df)
6308 v2df __builtin_ia32_cmplepd (v2df, v2df)
6309 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
6310 v2df __builtin_ia32_cmpgepd (v2df, v2df)
6311 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
6312 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
6313 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
6314 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
6315 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
6316 v2df __builtin_ia32_cmpngepd (v2df, v2df)
6317 v2df __builtin_ia32_cmpordpd (v2df, v2df)
6318 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
6319 v2df __builtin_ia32_cmpltsd (v2df, v2df)
6320 v2df __builtin_ia32_cmplesd (v2df, v2df)
6321 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
6322 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
6323 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
6324 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
6325 v2df __builtin_ia32_cmpordsd (v2df, v2df)
6326 v2di __builtin_ia32_paddq (v2di, v2di)
6327 v2di __builtin_ia32_psubq (v2di, v2di)
6328 v2df __builtin_ia32_addpd (v2df, v2df)
6329 v2df __builtin_ia32_subpd (v2df, v2df)
6330 v2df __builtin_ia32_mulpd (v2df, v2df)
6331 v2df __builtin_ia32_divpd (v2df, v2df)
6332 v2df __builtin_ia32_addsd (v2df, v2df)
6333 v2df __builtin_ia32_subsd (v2df, v2df)
6334 v2df __builtin_ia32_mulsd (v2df, v2df)
6335 v2df __builtin_ia32_divsd (v2df, v2df)
6336 v2df __builtin_ia32_minpd (v2df, v2df)
6337 v2df __builtin_ia32_maxpd (v2df, v2df)
6338 v2df __builtin_ia32_minsd (v2df, v2df)
6339 v2df __builtin_ia32_maxsd (v2df, v2df)
6340 v2df __builtin_ia32_andpd (v2df, v2df)
6341 v2df __builtin_ia32_andnpd (v2df, v2df)
6342 v2df __builtin_ia32_orpd (v2df, v2df)
6343 v2df __builtin_ia32_xorpd (v2df, v2df)
6344 v2df __builtin_ia32_movsd (v2df, v2df)
6345 v2df __builtin_ia32_unpckhpd (v2df, v2df)
6346 v2df __builtin_ia32_unpcklpd (v2df, v2df)
6347 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
6348 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
6349 v4si __builtin_ia32_paddd128 (v4si, v4si)
6350 v2di __builtin_ia32_paddq128 (v2di, v2di)
6351 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
6352 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
6353 v4si __builtin_ia32_psubd128 (v4si, v4si)
6354 v2di __builtin_ia32_psubq128 (v2di, v2di)
6355 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
6356 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
6357 v2di __builtin_ia32_pand128 (v2di, v2di)
6358 v2di __builtin_ia32_pandn128 (v2di, v2di)
6359 v2di __builtin_ia32_por128 (v2di, v2di)
6360 v2di __builtin_ia32_pxor128 (v2di, v2di)
6361 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
6362 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
6363 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
6364 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
6365 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
6366 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
6367 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
6368 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
6369 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
6370 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
6371 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
6372 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
6373 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
6374 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
6375 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
6376 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
6377 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
6378 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
6379 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
6380 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
6381 v16qi __builtin_ia32_packsswb128 (v16qi, v16qi)
6382 v8hi __builtin_ia32_packssdw128 (v8hi, v8hi)
6383 v16qi __builtin_ia32_packuswb128 (v16qi, v16qi)
6384 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
6385 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
6386 v2df __builtin_ia32_loadupd (double *)
6387 void __builtin_ia32_storeupd (double *, v2df)
6388 v2df __builtin_ia32_loadhpd (v2df, double *)
6389 v2df __builtin_ia32_loadlpd (v2df, double *)
6390 int __builtin_ia32_movmskpd (v2df)
6391 int __builtin_ia32_pmovmskb128 (v16qi)
6392 void __builtin_ia32_movnti (int *, int)
6393 void __builtin_ia32_movntpd (double *, v2df)
6394 void __builtin_ia32_movntdq (v2df *, v2df)
6395 v4si __builtin_ia32_pshufd (v4si, int)
6396 v8hi __builtin_ia32_pshuflw (v8hi, int)
6397 v8hi __builtin_ia32_pshufhw (v8hi, int)
6398 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
6399 v2df __builtin_ia32_sqrtpd (v2df)
6400 v2df __builtin_ia32_sqrtsd (v2df)
6401 v2df __builtin_ia32_shufpd (v2df, v2df, int)
6402 v2df __builtin_ia32_cvtdq2pd (v4si)
6403 v4sf __builtin_ia32_cvtdq2ps (v4si)
6404 v4si __builtin_ia32_cvtpd2dq (v2df)
6405 v2si __builtin_ia32_cvtpd2pi (v2df)
6406 v4sf __builtin_ia32_cvtpd2ps (v2df)
6407 v4si __builtin_ia32_cvttpd2dq (v2df)
6408 v2si __builtin_ia32_cvttpd2pi (v2df)
6409 v2df __builtin_ia32_cvtpi2pd (v2si)
6410 int __builtin_ia32_cvtsd2si (v2df)
6411 int __builtin_ia32_cvttsd2si (v2df)
6412 long long __builtin_ia32_cvtsd2si64 (v2df)
6413 long long __builtin_ia32_cvttsd2si64 (v2df)
6414 v4si __builtin_ia32_cvtps2dq (v4sf)
6415 v2df __builtin_ia32_cvtps2pd (v4sf)
6416 v4si __builtin_ia32_cvttps2dq (v4sf)
6417 v2df __builtin_ia32_cvtsi2sd (v2df, int)
6418 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
6419 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
6420 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
6421 void __builtin_ia32_clflush (const void *)
6422 void __builtin_ia32_lfence (void)
6423 void __builtin_ia32_mfence (void)
6424 v16qi __builtin_ia32_loaddqu (const char *)
6425 void __builtin_ia32_storedqu (char *, v16qi)
6426 unsigned long long __builtin_ia32_pmuludq (v2si, v2si)
6427 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
6428 v8hi __builtin_ia32_psllw128 (v8hi, v2di)
6429 v4si __builtin_ia32_pslld128 (v4si, v2di)
6430 v2di __builtin_ia32_psllq128 (v4si, v2di)
6431 v8hi __builtin_ia32_psrlw128 (v8hi, v2di)
6432 v4si __builtin_ia32_psrld128 (v4si, v2di)
6433 v2di __builtin_ia32_psrlq128 (v2di, v2di)
6434 v8hi __builtin_ia32_psraw128 (v8hi, v2di)
6435 v4si __builtin_ia32_psrad128 (v4si, v2di)
6436 v2di __builtin_ia32_pslldqi128 (v2di, int)
6437 v8hi __builtin_ia32_psllwi128 (v8hi, int)
6438 v4si __builtin_ia32_pslldi128 (v4si, int)
6439 v2di __builtin_ia32_psllqi128 (v2di, int)
6440 v2di __builtin_ia32_psrldqi128 (v2di, int)
6441 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
6442 v4si __builtin_ia32_psrldi128 (v4si, int)
6443 v2di __builtin_ia32_psrlqi128 (v2di, int)
6444 v8hi __builtin_ia32_psrawi128 (v8hi, int)
6445 v4si __builtin_ia32_psradi128 (v4si, int)
6446 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
6449 The following built-in functions are available when @option{-msse3} is used.
6450 All of them generate the machine instruction that is part of the name.
6453 v2df __builtin_ia32_addsubpd (v2df, v2df)
6454 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
6455 v2df __builtin_ia32_haddpd (v2df, v2df)
6456 v4sf __builtin_ia32_haddps (v4sf, v4sf)
6457 v2df __builtin_ia32_hsubpd (v2df, v2df)
6458 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
6459 v16qi __builtin_ia32_lddqu (char const *)
6460 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
6461 v2df __builtin_ia32_movddup (v2df)
6462 v4sf __builtin_ia32_movshdup (v4sf)
6463 v4sf __builtin_ia32_movsldup (v4sf)
6464 void __builtin_ia32_mwait (unsigned int, unsigned int)
6467 The following built-in functions are available when @option{-msse3} is used.
6470 @item v2df __builtin_ia32_loadddup (double const *)
6471 Generates the @code{movddup} machine instruction as a load from memory.
6474 The following built-in functions are available when @option{-m3dnow} is used.
6475 All of them generate the machine instruction that is part of the name.
6478 void __builtin_ia32_femms (void)
6479 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
6480 v2si __builtin_ia32_pf2id (v2sf)
6481 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
6482 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
6483 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
6484 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
6485 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
6486 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
6487 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
6488 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
6489 v2sf __builtin_ia32_pfrcp (v2sf)
6490 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
6491 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
6492 v2sf __builtin_ia32_pfrsqrt (v2sf)
6493 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
6494 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
6495 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
6496 v2sf __builtin_ia32_pi2fd (v2si)
6497 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
6500 The following built-in functions are available when both @option{-m3dnow}
6501 and @option{-march=athlon} are used. All of them generate the machine
6502 instruction that is part of the name.
6505 v2si __builtin_ia32_pf2iw (v2sf)
6506 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
6507 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
6508 v2sf __builtin_ia32_pi2fw (v2si)
6509 v2sf __builtin_ia32_pswapdsf (v2sf)
6510 v2si __builtin_ia32_pswapdsi (v2si)
6513 @node MIPS Paired-Single Support
6514 @subsection MIPS Paired-Single Support
6516 The MIPS64 architecture includes a number of instructions that
6517 operate on pairs of single-precision floating-point values.
6518 Each pair is packed into a 64-bit floating-point register,
6519 with one element being designated the ``upper half'' and
6520 the other being designated the ``lower half''.
6522 GCC supports paired-single operations using both the generic
6523 vector extensions (@pxref{Vector Extensions}) and a collection of
6524 MIPS-specific built-in functions. Both kinds of support are
6525 enabled by the @option{-mpaired-single} command-line option.
6527 The vector type associated with paired-single values is usually
6528 called @code{v2sf}. It can be defined in C as follows:
6531 typedef float v2sf __attribute__ ((vector_size (8)));
6534 @code{v2sf} values are initialized in the same way as aggregates.
6538 v2sf a = @{1.5, 9.1@};
6541 b = (v2sf) @{e, f@};
6544 @emph{Note:} The CPU's endianness determines which value is stored in
6545 the upper half of a register and which value is stored in the lower half.
6546 On little-endian targets, the first value is the lower one and the second
6547 value is the upper one. The opposite order applies to big-endian targets.
6548 For example, the code above will set the lower half of @code{a} to
6549 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
6552 * Paired-Single Arithmetic::
6553 * Paired-Single Built-in Functions::
6554 * MIPS-3D Built-in Functions::
6557 @node Paired-Single Arithmetic
6558 @subsubsection Paired-Single Arithmetic
6560 The table below lists the @code{v2sf} operations for which hardware
6561 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
6562 values and @code{x} is an integral value.
6564 @multitable @columnfractions .50 .50
6565 @item C code @tab MIPS instruction
6566 @item @code{a + b} @tab @code{add.ps}
6567 @item @code{a - b} @tab @code{sub.ps}
6568 @item @code{-a} @tab @code{neg.ps}
6569 @item @code{a * b} @tab @code{mul.ps}
6570 @item @code{a * b + c} @tab @code{madd.ps}
6571 @item @code{a * b - c} @tab @code{msub.ps}
6572 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
6573 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
6574 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
6577 Note that the multiply-accumulate instructions can be disabled
6578 using the command-line option @code{-mno-fused-madd}.
6580 @node Paired-Single Built-in Functions
6581 @subsubsection Paired-Single Built-in Functions
6583 The following paired-single functions map directly to a particular
6584 MIPS instruction. Please refer to the architecture specification
6585 for details on what each instruction does.
6588 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
6589 Pair lower lower (@code{pll.ps}).
6591 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
6592 Pair upper lower (@code{pul.ps}).
6594 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
6595 Pair lower upper (@code{plu.ps}).
6597 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
6598 Pair upper upper (@code{puu.ps}).
6600 @item v2sf __builtin_mips_cvt_ps_s (float, float)
6601 Convert pair to paired single (@code{cvt.ps.s}).
6603 @item float __builtin_mips_cvt_s_pl (v2sf)
6604 Convert pair lower to single (@code{cvt.s.pl}).
6606 @item float __builtin_mips_cvt_s_pu (v2sf)
6607 Convert pair upper to single (@code{cvt.s.pu}).
6609 @item v2sf __builtin_mips_abs_ps (v2sf)
6610 Absolute value (@code{abs.ps}).
6612 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
6613 Align variable (@code{alnv.ps}).
6615 @emph{Note:} The value of the third parameter must be 0 or 4
6616 modulo 8, otherwise the result will be unpredictable. Please read the
6617 instruction description for details.
6620 The following multi-instruction functions are also available.
6621 In each case, @var{cond} can be any of the 16 floating-point conditions:
6622 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
6623 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
6624 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
6627 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6628 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6629 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
6630 @code{movt.ps}/@code{movf.ps}).
6632 The @code{movt} functions return the value @var{x} computed by:
6635 c.@var{cond}.ps @var{cc},@var{a},@var{b}
6636 mov.ps @var{x},@var{c}
6637 movt.ps @var{x},@var{d},@var{cc}
6640 The @code{movf} functions are similar but use @code{movf.ps} instead
6643 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6644 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6645 Comparison of two paired-single values (@code{c.@var{cond}.ps},
6646 @code{bc1t}/@code{bc1f}).
6648 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
6649 and return either the upper or lower half of the result. For example:
6653 if (__builtin_mips_upper_c_eq_ps (a, b))
6654 upper_halves_are_equal ();
6656 upper_halves_are_unequal ();
6658 if (__builtin_mips_lower_c_eq_ps (a, b))
6659 lower_halves_are_equal ();
6661 lower_halves_are_unequal ();
6665 @node MIPS-3D Built-in Functions
6666 @subsubsection MIPS-3D Built-in Functions
6668 The MIPS-3D Application-Specific Extension (ASE) includes additional
6669 paired-single instructions that are designed to improve the performance
6670 of 3D graphics operations. Support for these instructions is controlled
6671 by the @option{-mips3d} command-line option.
6673 The functions listed below map directly to a particular MIPS-3D
6674 instruction. Please refer to the architecture specification for
6675 more details on what each instruction does.
6678 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
6679 Reduction add (@code{addr.ps}).
6681 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
6682 Reduction multiply (@code{mulr.ps}).
6684 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
6685 Convert paired single to paired word (@code{cvt.pw.ps}).
6687 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
6688 Convert paired word to paired single (@code{cvt.ps.pw}).
6690 @item float __builtin_mips_recip1_s (float)
6691 @itemx double __builtin_mips_recip1_d (double)
6692 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
6693 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
6695 @item float __builtin_mips_recip2_s (float, float)
6696 @itemx double __builtin_mips_recip2_d (double, double)
6697 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
6698 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
6700 @item float __builtin_mips_rsqrt1_s (float)
6701 @itemx double __builtin_mips_rsqrt1_d (double)
6702 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
6703 Reduced precision reciprocal square root (sequence step 1)
6704 (@code{rsqrt1.@var{fmt}}).
6706 @item float __builtin_mips_rsqrt2_s (float, float)
6707 @itemx double __builtin_mips_rsqrt2_d (double, double)
6708 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
6709 Reduced precision reciprocal square root (sequence step 2)
6710 (@code{rsqrt2.@var{fmt}}).
6713 The following multi-instruction functions are also available.
6714 In each case, @var{cond} can be any of the 16 floating-point conditions:
6715 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
6716 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
6717 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
6720 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
6721 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
6722 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
6723 @code{bc1t}/@code{bc1f}).
6725 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
6726 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
6731 if (__builtin_mips_cabs_eq_s (a, b))
6737 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6738 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6739 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
6740 @code{bc1t}/@code{bc1f}).
6742 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
6743 and return either the upper or lower half of the result. For example:
6747 if (__builtin_mips_upper_cabs_eq_ps (a, b))
6748 upper_halves_are_equal ();
6750 upper_halves_are_unequal ();
6752 if (__builtin_mips_lower_cabs_eq_ps (a, b))
6753 lower_halves_are_equal ();
6755 lower_halves_are_unequal ();
6758 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6759 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6760 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
6761 @code{movt.ps}/@code{movf.ps}).
6763 The @code{movt} functions return the value @var{x} computed by:
6766 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
6767 mov.ps @var{x},@var{c}
6768 movt.ps @var{x},@var{d},@var{cc}
6771 The @code{movf} functions are similar but use @code{movf.ps} instead
6774 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6775 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6776 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6777 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6778 Comparison of two paired-single values
6779 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
6780 @code{bc1any2t}/@code{bc1any2f}).
6782 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
6783 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
6784 result is true and the @code{all} forms return true if both results are true.
6789 if (__builtin_mips_any_c_eq_ps (a, b))
6794 if (__builtin_mips_all_c_eq_ps (a, b))
6800 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6801 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6802 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6803 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6804 Comparison of four paired-single values
6805 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
6806 @code{bc1any4t}/@code{bc1any4f}).
6808 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
6809 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
6810 The @code{any} forms return true if any of the four results are true
6811 and the @code{all} forms return true if all four results are true.
6816 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
6821 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
6828 @node PowerPC AltiVec Built-in Functions
6829 @subsection PowerPC AltiVec Built-in Functions
6831 GCC provides an interface for the PowerPC family of processors to access
6832 the AltiVec operations described in Motorola's AltiVec Programming
6833 Interface Manual. The interface is made available by including
6834 @code{<altivec.h>} and using @option{-maltivec} and
6835 @option{-mabi=altivec}. The interface supports the following vector
6839 vector unsigned char
6843 vector unsigned short
6854 GCC's implementation of the high-level language interface available from
6855 C and C++ code differs from Motorola's documentation in several ways.
6860 A vector constant is a list of constant expressions within curly braces.
6863 A vector initializer requires no cast if the vector constant is of the
6864 same type as the variable it is initializing.
6867 If @code{signed} or @code{unsigned} is omitted, the signedness of the
6868 vector type is the default signedness of the base type. The default
6869 varies depending on the operating system, so a portable program should
6870 always specify the signedness.
6873 Compiling with @option{-maltivec} adds keywords @code{__vector},
6874 @code{__pixel}, and @code{__bool}. Macros @option{vector},
6875 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
6879 GCC allows using a @code{typedef} name as the type specifier for a
6883 For C, overloaded functions are implemented with macros so the following
6887 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
6890 Since @code{vec_add} is a macro, the vector constant in the example
6891 is treated as four separate arguments. Wrap the entire argument in
6892 parentheses for this to work.
6895 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
6896 Internally, GCC uses built-in functions to achieve the functionality in
6897 the aforementioned header file, but they are not supported and are
6898 subject to change without notice.
6900 The following interfaces are supported for the generic and specific
6901 AltiVec operations and the AltiVec predicates. In cases where there
6902 is a direct mapping between generic and specific operations, only the
6903 generic names are shown here, although the specific operations can also
6906 Arguments that are documented as @code{const int} require literal
6907 integral values within the range required for that operation.
6910 vector signed char vec_abs (vector signed char);
6911 vector signed short vec_abs (vector signed short);
6912 vector signed int vec_abs (vector signed int);
6913 vector float vec_abs (vector float);
6915 vector signed char vec_abss (vector signed char);
6916 vector signed short vec_abss (vector signed short);
6917 vector signed int vec_abss (vector signed int);
6919 vector signed char vec_add (vector bool char, vector signed char);
6920 vector signed char vec_add (vector signed char, vector bool char);
6921 vector signed char vec_add (vector signed char, vector signed char);
6922 vector unsigned char vec_add (vector bool char, vector unsigned char);
6923 vector unsigned char vec_add (vector unsigned char, vector bool char);
6924 vector unsigned char vec_add (vector unsigned char,
6925 vector unsigned char);
6926 vector signed short vec_add (vector bool short, vector signed short);
6927 vector signed short vec_add (vector signed short, vector bool short);
6928 vector signed short vec_add (vector signed short, vector signed short);
6929 vector unsigned short vec_add (vector bool short,
6930 vector unsigned short);
6931 vector unsigned short vec_add (vector unsigned short,
6933 vector unsigned short vec_add (vector unsigned short,
6934 vector unsigned short);
6935 vector signed int vec_add (vector bool int, vector signed int);
6936 vector signed int vec_add (vector signed int, vector bool int);
6937 vector signed int vec_add (vector signed int, vector signed int);
6938 vector unsigned int vec_add (vector bool int, vector unsigned int);
6939 vector unsigned int vec_add (vector unsigned int, vector bool int);
6940 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
6941 vector float vec_add (vector float, vector float);
6943 vector float vec_vaddfp (vector float, vector float);
6945 vector signed int vec_vadduwm (vector bool int, vector signed int);
6946 vector signed int vec_vadduwm (vector signed int, vector bool int);
6947 vector signed int vec_vadduwm (vector signed int, vector signed int);
6948 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
6949 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
6950 vector unsigned int vec_vadduwm (vector unsigned int,
6951 vector unsigned int);
6953 vector signed short vec_vadduhm (vector bool short,
6954 vector signed short);
6955 vector signed short vec_vadduhm (vector signed short,
6957 vector signed short vec_vadduhm (vector signed short,
6958 vector signed short);
6959 vector unsigned short vec_vadduhm (vector bool short,
6960 vector unsigned short);
6961 vector unsigned short vec_vadduhm (vector unsigned short,
6963 vector unsigned short vec_vadduhm (vector unsigned short,
6964 vector unsigned short);
6966 vector signed char vec_vaddubm (vector bool char, vector signed char);
6967 vector signed char vec_vaddubm (vector signed char, vector bool char);
6968 vector signed char vec_vaddubm (vector signed char, vector signed char);
6969 vector unsigned char vec_vaddubm (vector bool char,
6970 vector unsigned char);
6971 vector unsigned char vec_vaddubm (vector unsigned char,
6973 vector unsigned char vec_vaddubm (vector unsigned char,
6974 vector unsigned char);
6976 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
6978 vector unsigned char vec_adds (vector bool char, vector unsigned char);
6979 vector unsigned char vec_adds (vector unsigned char, vector bool char);
6980 vector unsigned char vec_adds (vector unsigned char,
6981 vector unsigned char);
6982 vector signed char vec_adds (vector bool char, vector signed char);
6983 vector signed char vec_adds (vector signed char, vector bool char);
6984 vector signed char vec_adds (vector signed char, vector signed char);
6985 vector unsigned short vec_adds (vector bool short,
6986 vector unsigned short);
6987 vector unsigned short vec_adds (vector unsigned short,
6989 vector unsigned short vec_adds (vector unsigned short,
6990 vector unsigned short);
6991 vector signed short vec_adds (vector bool short, vector signed short);
6992 vector signed short vec_adds (vector signed short, vector bool short);
6993 vector signed short vec_adds (vector signed short, vector signed short);
6994 vector unsigned int vec_adds (vector bool int, vector unsigned int);
6995 vector unsigned int vec_adds (vector unsigned int, vector bool int);
6996 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
6997 vector signed int vec_adds (vector bool int, vector signed int);
6998 vector signed int vec_adds (vector signed int, vector bool int);
6999 vector signed int vec_adds (vector signed int, vector signed int);
7001 vector signed int vec_vaddsws (vector bool int, vector signed int);
7002 vector signed int vec_vaddsws (vector signed int, vector bool int);
7003 vector signed int vec_vaddsws (vector signed int, vector signed int);
7005 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
7006 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
7007 vector unsigned int vec_vadduws (vector unsigned int,
7008 vector unsigned int);
7010 vector signed short vec_vaddshs (vector bool short,
7011 vector signed short);
7012 vector signed short vec_vaddshs (vector signed short,
7014 vector signed short vec_vaddshs (vector signed short,
7015 vector signed short);
7017 vector unsigned short vec_vadduhs (vector bool short,
7018 vector unsigned short);
7019 vector unsigned short vec_vadduhs (vector unsigned short,
7021 vector unsigned short vec_vadduhs (vector unsigned short,
7022 vector unsigned short);
7024 vector signed char vec_vaddsbs (vector bool char, vector signed char);
7025 vector signed char vec_vaddsbs (vector signed char, vector bool char);
7026 vector signed char vec_vaddsbs (vector signed char, vector signed char);
7028 vector unsigned char vec_vaddubs (vector bool char,
7029 vector unsigned char);
7030 vector unsigned char vec_vaddubs (vector unsigned char,
7032 vector unsigned char vec_vaddubs (vector unsigned char,
7033 vector unsigned char);
7035 vector float vec_and (vector float, vector float);
7036 vector float vec_and (vector float, vector bool int);
7037 vector float vec_and (vector bool int, vector float);
7038 vector bool int vec_and (vector bool int, vector bool int);
7039 vector signed int vec_and (vector bool int, vector signed int);
7040 vector signed int vec_and (vector signed int, vector bool int);
7041 vector signed int vec_and (vector signed int, vector signed int);
7042 vector unsigned int vec_and (vector bool int, vector unsigned int);
7043 vector unsigned int vec_and (vector unsigned int, vector bool int);
7044 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
7045 vector bool short vec_and (vector bool short, vector bool short);
7046 vector signed short vec_and (vector bool short, vector signed short);
7047 vector signed short vec_and (vector signed short, vector bool short);
7048 vector signed short vec_and (vector signed short, vector signed short);
7049 vector unsigned short vec_and (vector bool short,
7050 vector unsigned short);
7051 vector unsigned short vec_and (vector unsigned short,
7053 vector unsigned short vec_and (vector unsigned short,
7054 vector unsigned short);
7055 vector signed char vec_and (vector bool char, vector signed char);
7056 vector bool char vec_and (vector bool char, vector bool char);
7057 vector signed char vec_and (vector signed char, vector bool char);
7058 vector signed char vec_and (vector signed char, vector signed char);
7059 vector unsigned char vec_and (vector bool char, vector unsigned char);
7060 vector unsigned char vec_and (vector unsigned char, vector bool char);
7061 vector unsigned char vec_and (vector unsigned char,
7062 vector unsigned char);
7064 vector float vec_andc (vector float, vector float);
7065 vector float vec_andc (vector float, vector bool int);
7066 vector float vec_andc (vector bool int, vector float);
7067 vector bool int vec_andc (vector bool int, vector bool int);
7068 vector signed int vec_andc (vector bool int, vector signed int);
7069 vector signed int vec_andc (vector signed int, vector bool int);
7070 vector signed int vec_andc (vector signed int, vector signed int);
7071 vector unsigned int vec_andc (vector bool int, vector unsigned int);
7072 vector unsigned int vec_andc (vector unsigned int, vector bool int);
7073 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
7074 vector bool short vec_andc (vector bool short, vector bool short);
7075 vector signed short vec_andc (vector bool short, vector signed short);
7076 vector signed short vec_andc (vector signed short, vector bool short);
7077 vector signed short vec_andc (vector signed short, vector signed short);
7078 vector unsigned short vec_andc (vector bool short,
7079 vector unsigned short);
7080 vector unsigned short vec_andc (vector unsigned short,
7082 vector unsigned short vec_andc (vector unsigned short,
7083 vector unsigned short);
7084 vector signed char vec_andc (vector bool char, vector signed char);
7085 vector bool char vec_andc (vector bool char, vector bool char);
7086 vector signed char vec_andc (vector signed char, vector bool char);
7087 vector signed char vec_andc (vector signed char, vector signed char);
7088 vector unsigned char vec_andc (vector bool char, vector unsigned char);
7089 vector unsigned char vec_andc (vector unsigned char, vector bool char);
7090 vector unsigned char vec_andc (vector unsigned char,
7091 vector unsigned char);
7093 vector unsigned char vec_avg (vector unsigned char,
7094 vector unsigned char);
7095 vector signed char vec_avg (vector signed char, vector signed char);
7096 vector unsigned short vec_avg (vector unsigned short,
7097 vector unsigned short);
7098 vector signed short vec_avg (vector signed short, vector signed short);
7099 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
7100 vector signed int vec_avg (vector signed int, vector signed int);
7102 vector signed int vec_vavgsw (vector signed int, vector signed int);
7104 vector unsigned int vec_vavguw (vector unsigned int,
7105 vector unsigned int);
7107 vector signed short vec_vavgsh (vector signed short,
7108 vector signed short);
7110 vector unsigned short vec_vavguh (vector unsigned short,
7111 vector unsigned short);
7113 vector signed char vec_vavgsb (vector signed char, vector signed char);
7115 vector unsigned char vec_vavgub (vector unsigned char,
7116 vector unsigned char);
7118 vector float vec_ceil (vector float);
7120 vector signed int vec_cmpb (vector float, vector float);
7122 vector bool char vec_cmpeq (vector signed char, vector signed char);
7123 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
7124 vector bool short vec_cmpeq (vector signed short, vector signed short);
7125 vector bool short vec_cmpeq (vector unsigned short,
7126 vector unsigned short);
7127 vector bool int vec_cmpeq (vector signed int, vector signed int);
7128 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
7129 vector bool int vec_cmpeq (vector float, vector float);
7131 vector bool int vec_vcmpeqfp (vector float, vector float);
7133 vector bool int vec_vcmpequw (vector signed int, vector signed int);
7134 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
7136 vector bool short vec_vcmpequh (vector signed short,
7137 vector signed short);
7138 vector bool short vec_vcmpequh (vector unsigned short,
7139 vector unsigned short);
7141 vector bool char vec_vcmpequb (vector signed char, vector signed char);
7142 vector bool char vec_vcmpequb (vector unsigned char,
7143 vector unsigned char);
7145 vector bool int vec_cmpge (vector float, vector float);
7147 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
7148 vector bool char vec_cmpgt (vector signed char, vector signed char);
7149 vector bool short vec_cmpgt (vector unsigned short,
7150 vector unsigned short);
7151 vector bool short vec_cmpgt (vector signed short, vector signed short);
7152 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
7153 vector bool int vec_cmpgt (vector signed int, vector signed int);
7154 vector bool int vec_cmpgt (vector float, vector float);
7156 vector bool int vec_vcmpgtfp (vector float, vector float);
7158 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
7160 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
7162 vector bool short vec_vcmpgtsh (vector signed short,
7163 vector signed short);
7165 vector bool short vec_vcmpgtuh (vector unsigned short,
7166 vector unsigned short);
7168 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
7170 vector bool char vec_vcmpgtub (vector unsigned char,
7171 vector unsigned char);
7173 vector bool int vec_cmple (vector float, vector float);
7175 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
7176 vector bool char vec_cmplt (vector signed char, vector signed char);
7177 vector bool short vec_cmplt (vector unsigned short,
7178 vector unsigned short);
7179 vector bool short vec_cmplt (vector signed short, vector signed short);
7180 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
7181 vector bool int vec_cmplt (vector signed int, vector signed int);
7182 vector bool int vec_cmplt (vector float, vector float);
7184 vector float vec_ctf (vector unsigned int, const int);
7185 vector float vec_ctf (vector signed int, const int);
7187 vector float vec_vcfsx (vector signed int, const int);
7189 vector float vec_vcfux (vector unsigned int, const int);
7191 vector signed int vec_cts (vector float, const int);
7193 vector unsigned int vec_ctu (vector float, const int);
7195 void vec_dss (const int);
7197 void vec_dssall (void);
7199 void vec_dst (const vector unsigned char *, int, const int);
7200 void vec_dst (const vector signed char *, int, const int);
7201 void vec_dst (const vector bool char *, int, const int);
7202 void vec_dst (const vector unsigned short *, int, const int);
7203 void vec_dst (const vector signed short *, int, const int);
7204 void vec_dst (const vector bool short *, int, const int);
7205 void vec_dst (const vector pixel *, int, const int);
7206 void vec_dst (const vector unsigned int *, int, const int);
7207 void vec_dst (const vector signed int *, int, const int);
7208 void vec_dst (const vector bool int *, int, const int);
7209 void vec_dst (const vector float *, int, const int);
7210 void vec_dst (const unsigned char *, int, const int);
7211 void vec_dst (const signed char *, int, const int);
7212 void vec_dst (const unsigned short *, int, const int);
7213 void vec_dst (const short *, int, const int);
7214 void vec_dst (const unsigned int *, int, const int);
7215 void vec_dst (const int *, int, const int);
7216 void vec_dst (const unsigned long *, int, const int);
7217 void vec_dst (const long *, int, const int);
7218 void vec_dst (const float *, int, const int);
7220 void vec_dstst (const vector unsigned char *, int, const int);
7221 void vec_dstst (const vector signed char *, int, const int);
7222 void vec_dstst (const vector bool char *, int, const int);
7223 void vec_dstst (const vector unsigned short *, int, const int);
7224 void vec_dstst (const vector signed short *, int, const int);
7225 void vec_dstst (const vector bool short *, int, const int);
7226 void vec_dstst (const vector pixel *, int, const int);
7227 void vec_dstst (const vector unsigned int *, int, const int);
7228 void vec_dstst (const vector signed int *, int, const int);
7229 void vec_dstst (const vector bool int *, int, const int);
7230 void vec_dstst (const vector float *, int, const int);
7231 void vec_dstst (const unsigned char *, int, const int);
7232 void vec_dstst (const signed char *, int, const int);
7233 void vec_dstst (const unsigned short *, int, const int);
7234 void vec_dstst (const short *, int, const int);
7235 void vec_dstst (const unsigned int *, int, const int);
7236 void vec_dstst (const int *, int, const int);
7237 void vec_dstst (const unsigned long *, int, const int);
7238 void vec_dstst (const long *, int, const int);
7239 void vec_dstst (const float *, int, const int);
7241 void vec_dststt (const vector unsigned char *, int, const int);
7242 void vec_dststt (const vector signed char *, int, const int);
7243 void vec_dststt (const vector bool char *, int, const int);
7244 void vec_dststt (const vector unsigned short *, int, const int);
7245 void vec_dststt (const vector signed short *, int, const int);
7246 void vec_dststt (const vector bool short *, int, const int);
7247 void vec_dststt (const vector pixel *, int, const int);
7248 void vec_dststt (const vector unsigned int *, int, const int);
7249 void vec_dststt (const vector signed int *, int, const int);
7250 void vec_dststt (const vector bool int *, int, const int);
7251 void vec_dststt (const vector float *, int, const int);
7252 void vec_dststt (const unsigned char *, int, const int);
7253 void vec_dststt (const signed char *, int, const int);
7254 void vec_dststt (const unsigned short *, int, const int);
7255 void vec_dststt (const short *, int, const int);
7256 void vec_dststt (const unsigned int *, int, const int);
7257 void vec_dststt (const int *, int, const int);
7258 void vec_dststt (const unsigned long *, int, const int);
7259 void vec_dststt (const long *, int, const int);
7260 void vec_dststt (const float *, int, const int);
7262 void vec_dstt (const vector unsigned char *, int, const int);
7263 void vec_dstt (const vector signed char *, int, const int);
7264 void vec_dstt (const vector bool char *, int, const int);
7265 void vec_dstt (const vector unsigned short *, int, const int);
7266 void vec_dstt (const vector signed short *, int, const int);
7267 void vec_dstt (const vector bool short *, int, const int);
7268 void vec_dstt (const vector pixel *, int, const int);
7269 void vec_dstt (const vector unsigned int *, int, const int);
7270 void vec_dstt (const vector signed int *, int, const int);
7271 void vec_dstt (const vector bool int *, int, const int);
7272 void vec_dstt (const vector float *, int, const int);
7273 void vec_dstt (const unsigned char *, int, const int);
7274 void vec_dstt (const signed char *, int, const int);
7275 void vec_dstt (const unsigned short *, int, const int);
7276 void vec_dstt (const short *, int, const int);
7277 void vec_dstt (const unsigned int *, int, const int);
7278 void vec_dstt (const int *, int, const int);
7279 void vec_dstt (const unsigned long *, int, const int);
7280 void vec_dstt (const long *, int, const int);
7281 void vec_dstt (const float *, int, const int);
7283 vector float vec_expte (vector float);
7285 vector float vec_floor (vector float);
7287 vector float vec_ld (int, const vector float *);
7288 vector float vec_ld (int, const float *);
7289 vector bool int vec_ld (int, const vector bool int *);
7290 vector signed int vec_ld (int, const vector signed int *);
7291 vector signed int vec_ld (int, const int *);
7292 vector signed int vec_ld (int, const long *);
7293 vector unsigned int vec_ld (int, const vector unsigned int *);
7294 vector unsigned int vec_ld (int, const unsigned int *);
7295 vector unsigned int vec_ld (int, const unsigned long *);
7296 vector bool short vec_ld (int, const vector bool short *);
7297 vector pixel vec_ld (int, const vector pixel *);
7298 vector signed short vec_ld (int, const vector signed short *);
7299 vector signed short vec_ld (int, const short *);
7300 vector unsigned short vec_ld (int, const vector unsigned short *);
7301 vector unsigned short vec_ld (int, const unsigned short *);
7302 vector bool char vec_ld (int, const vector bool char *);
7303 vector signed char vec_ld (int, const vector signed char *);
7304 vector signed char vec_ld (int, const signed char *);
7305 vector unsigned char vec_ld (int, const vector unsigned char *);
7306 vector unsigned char vec_ld (int, const unsigned char *);
7308 vector signed char vec_lde (int, const signed char *);
7309 vector unsigned char vec_lde (int, const unsigned char *);
7310 vector signed short vec_lde (int, const short *);
7311 vector unsigned short vec_lde (int, const unsigned short *);
7312 vector float vec_lde (int, const float *);
7313 vector signed int vec_lde (int, const int *);
7314 vector unsigned int vec_lde (int, const unsigned int *);
7315 vector signed int vec_lde (int, const long *);
7316 vector unsigned int vec_lde (int, const unsigned long *);
7318 vector float vec_lvewx (int, float *);
7319 vector signed int vec_lvewx (int, int *);
7320 vector unsigned int vec_lvewx (int, unsigned int *);
7321 vector signed int vec_lvewx (int, long *);
7322 vector unsigned int vec_lvewx (int, unsigned long *);
7324 vector signed short vec_lvehx (int, short *);
7325 vector unsigned short vec_lvehx (int, unsigned short *);
7327 vector signed char vec_lvebx (int, char *);
7328 vector unsigned char vec_lvebx (int, unsigned char *);
7330 vector float vec_ldl (int, const vector float *);
7331 vector float vec_ldl (int, const float *);
7332 vector bool int vec_ldl (int, const vector bool int *);
7333 vector signed int vec_ldl (int, const vector signed int *);
7334 vector signed int vec_ldl (int, const int *);
7335 vector signed int vec_ldl (int, const long *);
7336 vector unsigned int vec_ldl (int, const vector unsigned int *);
7337 vector unsigned int vec_ldl (int, const unsigned int *);
7338 vector unsigned int vec_ldl (int, const unsigned long *);
7339 vector bool short vec_ldl (int, const vector bool short *);
7340 vector pixel vec_ldl (int, const vector pixel *);
7341 vector signed short vec_ldl (int, const vector signed short *);
7342 vector signed short vec_ldl (int, const short *);
7343 vector unsigned short vec_ldl (int, const vector unsigned short *);
7344 vector unsigned short vec_ldl (int, const unsigned short *);
7345 vector bool char vec_ldl (int, const vector bool char *);
7346 vector signed char vec_ldl (int, const vector signed char *);
7347 vector signed char vec_ldl (int, const signed char *);
7348 vector unsigned char vec_ldl (int, const vector unsigned char *);
7349 vector unsigned char vec_ldl (int, const unsigned char *);
7351 vector float vec_loge (vector float);
7353 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
7354 vector unsigned char vec_lvsl (int, const volatile signed char *);
7355 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
7356 vector unsigned char vec_lvsl (int, const volatile short *);
7357 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
7358 vector unsigned char vec_lvsl (int, const volatile int *);
7359 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
7360 vector unsigned char vec_lvsl (int, const volatile long *);
7361 vector unsigned char vec_lvsl (int, const volatile float *);
7363 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
7364 vector unsigned char vec_lvsr (int, const volatile signed char *);
7365 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
7366 vector unsigned char vec_lvsr (int, const volatile short *);
7367 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
7368 vector unsigned char vec_lvsr (int, const volatile int *);
7369 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
7370 vector unsigned char vec_lvsr (int, const volatile long *);
7371 vector unsigned char vec_lvsr (int, const volatile float *);
7373 vector float vec_madd (vector float, vector float, vector float);
7375 vector signed short vec_madds (vector signed short,
7376 vector signed short,
7377 vector signed short);
7379 vector unsigned char vec_max (vector bool char, vector unsigned char);
7380 vector unsigned char vec_max (vector unsigned char, vector bool char);
7381 vector unsigned char vec_max (vector unsigned char,
7382 vector unsigned char);
7383 vector signed char vec_max (vector bool char, vector signed char);
7384 vector signed char vec_max (vector signed char, vector bool char);
7385 vector signed char vec_max (vector signed char, vector signed char);
7386 vector unsigned short vec_max (vector bool short,
7387 vector unsigned short);
7388 vector unsigned short vec_max (vector unsigned short,
7390 vector unsigned short vec_max (vector unsigned short,
7391 vector unsigned short);
7392 vector signed short vec_max (vector bool short, vector signed short);
7393 vector signed short vec_max (vector signed short, vector bool short);
7394 vector signed short vec_max (vector signed short, vector signed short);
7395 vector unsigned int vec_max (vector bool int, vector unsigned int);
7396 vector unsigned int vec_max (vector unsigned int, vector bool int);
7397 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
7398 vector signed int vec_max (vector bool int, vector signed int);
7399 vector signed int vec_max (vector signed int, vector bool int);
7400 vector signed int vec_max (vector signed int, vector signed int);
7401 vector float vec_max (vector float, vector float);
7403 vector float vec_vmaxfp (vector float, vector float);
7405 vector signed int vec_vmaxsw (vector bool int, vector signed int);
7406 vector signed int vec_vmaxsw (vector signed int, vector bool int);
7407 vector signed int vec_vmaxsw (vector signed int, vector signed int);
7409 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
7410 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
7411 vector unsigned int vec_vmaxuw (vector unsigned int,
7412 vector unsigned int);
7414 vector signed short vec_vmaxsh (vector bool short, vector signed short);
7415 vector signed short vec_vmaxsh (vector signed short, vector bool short);
7416 vector signed short vec_vmaxsh (vector signed short,
7417 vector signed short);
7419 vector unsigned short vec_vmaxuh (vector bool short,
7420 vector unsigned short);
7421 vector unsigned short vec_vmaxuh (vector unsigned short,
7423 vector unsigned short vec_vmaxuh (vector unsigned short,
7424 vector unsigned short);
7426 vector signed char vec_vmaxsb (vector bool char, vector signed char);
7427 vector signed char vec_vmaxsb (vector signed char, vector bool char);
7428 vector signed char vec_vmaxsb (vector signed char, vector signed char);
7430 vector unsigned char vec_vmaxub (vector bool char,
7431 vector unsigned char);
7432 vector unsigned char vec_vmaxub (vector unsigned char,
7434 vector unsigned char vec_vmaxub (vector unsigned char,
7435 vector unsigned char);
7437 vector bool char vec_mergeh (vector bool char, vector bool char);
7438 vector signed char vec_mergeh (vector signed char, vector signed char);
7439 vector unsigned char vec_mergeh (vector unsigned char,
7440 vector unsigned char);
7441 vector bool short vec_mergeh (vector bool short, vector bool short);
7442 vector pixel vec_mergeh (vector pixel, vector pixel);
7443 vector signed short vec_mergeh (vector signed short,
7444 vector signed short);
7445 vector unsigned short vec_mergeh (vector unsigned short,
7446 vector unsigned short);
7447 vector float vec_mergeh (vector float, vector float);
7448 vector bool int vec_mergeh (vector bool int, vector bool int);
7449 vector signed int vec_mergeh (vector signed int, vector signed int);
7450 vector unsigned int vec_mergeh (vector unsigned int,
7451 vector unsigned int);
7453 vector float vec_vmrghw (vector float, vector float);
7454 vector bool int vec_vmrghw (vector bool int, vector bool int);
7455 vector signed int vec_vmrghw (vector signed int, vector signed int);
7456 vector unsigned int vec_vmrghw (vector unsigned int,
7457 vector unsigned int);
7459 vector bool short vec_vmrghh (vector bool short, vector bool short);
7460 vector signed short vec_vmrghh (vector signed short,
7461 vector signed short);
7462 vector unsigned short vec_vmrghh (vector unsigned short,
7463 vector unsigned short);
7464 vector pixel vec_vmrghh (vector pixel, vector pixel);
7466 vector bool char vec_vmrghb (vector bool char, vector bool char);
7467 vector signed char vec_vmrghb (vector signed char, vector signed char);
7468 vector unsigned char vec_vmrghb (vector unsigned char,
7469 vector unsigned char);
7471 vector bool char vec_mergel (vector bool char, vector bool char);
7472 vector signed char vec_mergel (vector signed char, vector signed char);
7473 vector unsigned char vec_mergel (vector unsigned char,
7474 vector unsigned char);
7475 vector bool short vec_mergel (vector bool short, vector bool short);
7476 vector pixel vec_mergel (vector pixel, vector pixel);
7477 vector signed short vec_mergel (vector signed short,
7478 vector signed short);
7479 vector unsigned short vec_mergel (vector unsigned short,
7480 vector unsigned short);
7481 vector float vec_mergel (vector float, vector float);
7482 vector bool int vec_mergel (vector bool int, vector bool int);
7483 vector signed int vec_mergel (vector signed int, vector signed int);
7484 vector unsigned int vec_mergel (vector unsigned int,
7485 vector unsigned int);
7487 vector float vec_vmrglw (vector float, vector float);
7488 vector signed int vec_vmrglw (vector signed int, vector signed int);
7489 vector unsigned int vec_vmrglw (vector unsigned int,
7490 vector unsigned int);
7491 vector bool int vec_vmrglw (vector bool int, vector bool int);
7493 vector bool short vec_vmrglh (vector bool short, vector bool short);
7494 vector signed short vec_vmrglh (vector signed short,
7495 vector signed short);
7496 vector unsigned short vec_vmrglh (vector unsigned short,
7497 vector unsigned short);
7498 vector pixel vec_vmrglh (vector pixel, vector pixel);
7500 vector bool char vec_vmrglb (vector bool char, vector bool char);
7501 vector signed char vec_vmrglb (vector signed char, vector signed char);
7502 vector unsigned char vec_vmrglb (vector unsigned char,
7503 vector unsigned char);
7505 vector unsigned short vec_mfvscr (void);
7507 vector unsigned char vec_min (vector bool char, vector unsigned char);
7508 vector unsigned char vec_min (vector unsigned char, vector bool char);
7509 vector unsigned char vec_min (vector unsigned char,
7510 vector unsigned char);
7511 vector signed char vec_min (vector bool char, vector signed char);
7512 vector signed char vec_min (vector signed char, vector bool char);
7513 vector signed char vec_min (vector signed char, vector signed char);
7514 vector unsigned short vec_min (vector bool short,
7515 vector unsigned short);
7516 vector unsigned short vec_min (vector unsigned short,
7518 vector unsigned short vec_min (vector unsigned short,
7519 vector unsigned short);
7520 vector signed short vec_min (vector bool short, vector signed short);
7521 vector signed short vec_min (vector signed short, vector bool short);
7522 vector signed short vec_min (vector signed short, vector signed short);
7523 vector unsigned int vec_min (vector bool int, vector unsigned int);
7524 vector unsigned int vec_min (vector unsigned int, vector bool int);
7525 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
7526 vector signed int vec_min (vector bool int, vector signed int);
7527 vector signed int vec_min (vector signed int, vector bool int);
7528 vector signed int vec_min (vector signed int, vector signed int);
7529 vector float vec_min (vector float, vector float);
7531 vector float vec_vminfp (vector float, vector float);
7533 vector signed int vec_vminsw (vector bool int, vector signed int);
7534 vector signed int vec_vminsw (vector signed int, vector bool int);
7535 vector signed int vec_vminsw (vector signed int, vector signed int);
7537 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
7538 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
7539 vector unsigned int vec_vminuw (vector unsigned int,
7540 vector unsigned int);
7542 vector signed short vec_vminsh (vector bool short, vector signed short);
7543 vector signed short vec_vminsh (vector signed short, vector bool short);
7544 vector signed short vec_vminsh (vector signed short,
7545 vector signed short);
7547 vector unsigned short vec_vminuh (vector bool short,
7548 vector unsigned short);
7549 vector unsigned short vec_vminuh (vector unsigned short,
7551 vector unsigned short vec_vminuh (vector unsigned short,
7552 vector unsigned short);
7554 vector signed char vec_vminsb (vector bool char, vector signed char);
7555 vector signed char vec_vminsb (vector signed char, vector bool char);
7556 vector signed char vec_vminsb (vector signed char, vector signed char);
7558 vector unsigned char vec_vminub (vector bool char,
7559 vector unsigned char);
7560 vector unsigned char vec_vminub (vector unsigned char,
7562 vector unsigned char vec_vminub (vector unsigned char,
7563 vector unsigned char);
7565 vector signed short vec_mladd (vector signed short,
7566 vector signed short,
7567 vector signed short);
7568 vector signed short vec_mladd (vector signed short,
7569 vector unsigned short,
7570 vector unsigned short);
7571 vector signed short vec_mladd (vector unsigned short,
7572 vector signed short,
7573 vector signed short);
7574 vector unsigned short vec_mladd (vector unsigned short,
7575 vector unsigned short,
7576 vector unsigned short);
7578 vector signed short vec_mradds (vector signed short,
7579 vector signed short,
7580 vector signed short);
7582 vector unsigned int vec_msum (vector unsigned char,
7583 vector unsigned char,
7584 vector unsigned int);
7585 vector signed int vec_msum (vector signed char,
7586 vector unsigned char,
7588 vector unsigned int vec_msum (vector unsigned short,
7589 vector unsigned short,
7590 vector unsigned int);
7591 vector signed int vec_msum (vector signed short,
7592 vector signed short,
7595 vector signed int vec_vmsumshm (vector signed short,
7596 vector signed short,
7599 vector unsigned int vec_vmsumuhm (vector unsigned short,
7600 vector unsigned short,
7601 vector unsigned int);
7603 vector signed int vec_vmsummbm (vector signed char,
7604 vector unsigned char,
7607 vector unsigned int vec_vmsumubm (vector unsigned char,
7608 vector unsigned char,
7609 vector unsigned int);
7611 vector unsigned int vec_msums (vector unsigned short,
7612 vector unsigned short,
7613 vector unsigned int);
7614 vector signed int vec_msums (vector signed short,
7615 vector signed short,
7618 vector signed int vec_vmsumshs (vector signed short,
7619 vector signed short,
7622 vector unsigned int vec_vmsumuhs (vector unsigned short,
7623 vector unsigned short,
7624 vector unsigned int);
7626 void vec_mtvscr (vector signed int);
7627 void vec_mtvscr (vector unsigned int);
7628 void vec_mtvscr (vector bool int);
7629 void vec_mtvscr (vector signed short);
7630 void vec_mtvscr (vector unsigned short);
7631 void vec_mtvscr (vector bool short);
7632 void vec_mtvscr (vector pixel);
7633 void vec_mtvscr (vector signed char);
7634 void vec_mtvscr (vector unsigned char);
7635 void vec_mtvscr (vector bool char);
7637 vector unsigned short vec_mule (vector unsigned char,
7638 vector unsigned char);
7639 vector signed short vec_mule (vector signed char,
7640 vector signed char);
7641 vector unsigned int vec_mule (vector unsigned short,
7642 vector unsigned short);
7643 vector signed int vec_mule (vector signed short, vector signed short);
7645 vector signed int vec_vmulesh (vector signed short,
7646 vector signed short);
7648 vector unsigned int vec_vmuleuh (vector unsigned short,
7649 vector unsigned short);
7651 vector signed short vec_vmulesb (vector signed char,
7652 vector signed char);
7654 vector unsigned short vec_vmuleub (vector unsigned char,
7655 vector unsigned char);
7657 vector unsigned short vec_mulo (vector unsigned char,
7658 vector unsigned char);
7659 vector signed short vec_mulo (vector signed char, vector signed char);
7660 vector unsigned int vec_mulo (vector unsigned short,
7661 vector unsigned short);
7662 vector signed int vec_mulo (vector signed short, vector signed short);
7664 vector signed int vec_vmulosh (vector signed short,
7665 vector signed short);
7667 vector unsigned int vec_vmulouh (vector unsigned short,
7668 vector unsigned short);
7670 vector signed short vec_vmulosb (vector signed char,
7671 vector signed char);
7673 vector unsigned short vec_vmuloub (vector unsigned char,
7674 vector unsigned char);
7676 vector float vec_nmsub (vector float, vector float, vector float);
7678 vector float vec_nor (vector float, vector float);
7679 vector signed int vec_nor (vector signed int, vector signed int);
7680 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
7681 vector bool int vec_nor (vector bool int, vector bool int);
7682 vector signed short vec_nor (vector signed short, vector signed short);
7683 vector unsigned short vec_nor (vector unsigned short,
7684 vector unsigned short);
7685 vector bool short vec_nor (vector bool short, vector bool short);
7686 vector signed char vec_nor (vector signed char, vector signed char);
7687 vector unsigned char vec_nor (vector unsigned char,
7688 vector unsigned char);
7689 vector bool char vec_nor (vector bool char, vector bool char);
7691 vector float vec_or (vector float, vector float);
7692 vector float vec_or (vector float, vector bool int);
7693 vector float vec_or (vector bool int, vector float);
7694 vector bool int vec_or (vector bool int, vector bool int);
7695 vector signed int vec_or (vector bool int, vector signed int);
7696 vector signed int vec_or (vector signed int, vector bool int);
7697 vector signed int vec_or (vector signed int, vector signed int);
7698 vector unsigned int vec_or (vector bool int, vector unsigned int);
7699 vector unsigned int vec_or (vector unsigned int, vector bool int);
7700 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
7701 vector bool short vec_or (vector bool short, vector bool short);
7702 vector signed short vec_or (vector bool short, vector signed short);
7703 vector signed short vec_or (vector signed short, vector bool short);
7704 vector signed short vec_or (vector signed short, vector signed short);
7705 vector unsigned short vec_or (vector bool short, vector unsigned short);
7706 vector unsigned short vec_or (vector unsigned short, vector bool short);
7707 vector unsigned short vec_or (vector unsigned short,
7708 vector unsigned short);
7709 vector signed char vec_or (vector bool char, vector signed char);
7710 vector bool char vec_or (vector bool char, vector bool char);
7711 vector signed char vec_or (vector signed char, vector bool char);
7712 vector signed char vec_or (vector signed char, vector signed char);
7713 vector unsigned char vec_or (vector bool char, vector unsigned char);
7714 vector unsigned char vec_or (vector unsigned char, vector bool char);
7715 vector unsigned char vec_or (vector unsigned char,
7716 vector unsigned char);
7718 vector signed char vec_pack (vector signed short, vector signed short);
7719 vector unsigned char vec_pack (vector unsigned short,
7720 vector unsigned short);
7721 vector bool char vec_pack (vector bool short, vector bool short);
7722 vector signed short vec_pack (vector signed int, vector signed int);
7723 vector unsigned short vec_pack (vector unsigned int,
7724 vector unsigned int);
7725 vector bool short vec_pack (vector bool int, vector bool int);
7727 vector bool short vec_vpkuwum (vector bool int, vector bool int);
7728 vector signed short vec_vpkuwum (vector signed int, vector signed int);
7729 vector unsigned short vec_vpkuwum (vector unsigned int,
7730 vector unsigned int);
7732 vector bool char vec_vpkuhum (vector bool short, vector bool short);
7733 vector signed char vec_vpkuhum (vector signed short,
7734 vector signed short);
7735 vector unsigned char vec_vpkuhum (vector unsigned short,
7736 vector unsigned short);
7738 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
7740 vector unsigned char vec_packs (vector unsigned short,
7741 vector unsigned short);
7742 vector signed char vec_packs (vector signed short, vector signed short);
7743 vector unsigned short vec_packs (vector unsigned int,
7744 vector unsigned int);
7745 vector signed short vec_packs (vector signed int, vector signed int);
7747 vector signed short vec_vpkswss (vector signed int, vector signed int);
7749 vector unsigned short vec_vpkuwus (vector unsigned int,
7750 vector unsigned int);
7752 vector signed char vec_vpkshss (vector signed short,
7753 vector signed short);
7755 vector unsigned char vec_vpkuhus (vector unsigned short,
7756 vector unsigned short);
7758 vector unsigned char vec_packsu (vector unsigned short,
7759 vector unsigned short);
7760 vector unsigned char vec_packsu (vector signed short,
7761 vector signed short);
7762 vector unsigned short vec_packsu (vector unsigned int,
7763 vector unsigned int);
7764 vector unsigned short vec_packsu (vector signed int, vector signed int);
7766 vector unsigned short vec_vpkswus (vector signed int,
7769 vector unsigned char vec_vpkshus (vector signed short,
7770 vector signed short);
7772 vector float vec_perm (vector float,
7774 vector unsigned char);
7775 vector signed int vec_perm (vector signed int,
7777 vector unsigned char);
7778 vector unsigned int vec_perm (vector unsigned int,
7779 vector unsigned int,
7780 vector unsigned char);
7781 vector bool int vec_perm (vector bool int,
7783 vector unsigned char);
7784 vector signed short vec_perm (vector signed short,
7785 vector signed short,
7786 vector unsigned char);
7787 vector unsigned short vec_perm (vector unsigned short,
7788 vector unsigned short,
7789 vector unsigned char);
7790 vector bool short vec_perm (vector bool short,
7792 vector unsigned char);
7793 vector pixel vec_perm (vector pixel,
7795 vector unsigned char);
7796 vector signed char vec_perm (vector signed char,
7798 vector unsigned char);
7799 vector unsigned char vec_perm (vector unsigned char,
7800 vector unsigned char,
7801 vector unsigned char);
7802 vector bool char vec_perm (vector bool char,
7804 vector unsigned char);
7806 vector float vec_re (vector float);
7808 vector signed char vec_rl (vector signed char,
7809 vector unsigned char);
7810 vector unsigned char vec_rl (vector unsigned char,
7811 vector unsigned char);
7812 vector signed short vec_rl (vector signed short, vector unsigned short);
7813 vector unsigned short vec_rl (vector unsigned short,
7814 vector unsigned short);
7815 vector signed int vec_rl (vector signed int, vector unsigned int);
7816 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
7818 vector signed int vec_vrlw (vector signed int, vector unsigned int);
7819 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
7821 vector signed short vec_vrlh (vector signed short,
7822 vector unsigned short);
7823 vector unsigned short vec_vrlh (vector unsigned short,
7824 vector unsigned short);
7826 vector signed char vec_vrlb (vector signed char, vector unsigned char);
7827 vector unsigned char vec_vrlb (vector unsigned char,
7828 vector unsigned char);
7830 vector float vec_round (vector float);
7832 vector float vec_rsqrte (vector float);
7834 vector float vec_sel (vector float, vector float, vector bool int);
7835 vector float vec_sel (vector float, vector float, vector unsigned int);
7836 vector signed int vec_sel (vector signed int,
7839 vector signed int vec_sel (vector signed int,
7841 vector unsigned int);
7842 vector unsigned int vec_sel (vector unsigned int,
7843 vector unsigned int,
7845 vector unsigned int vec_sel (vector unsigned int,
7846 vector unsigned int,
7847 vector unsigned int);
7848 vector bool int vec_sel (vector bool int,
7851 vector bool int vec_sel (vector bool int,
7853 vector unsigned int);
7854 vector signed short vec_sel (vector signed short,
7855 vector signed short,
7857 vector signed short vec_sel (vector signed short,
7858 vector signed short,
7859 vector unsigned short);
7860 vector unsigned short vec_sel (vector unsigned short,
7861 vector unsigned short,
7863 vector unsigned short vec_sel (vector unsigned short,
7864 vector unsigned short,
7865 vector unsigned short);
7866 vector bool short vec_sel (vector bool short,
7869 vector bool short vec_sel (vector bool short,
7871 vector unsigned short);
7872 vector signed char vec_sel (vector signed char,
7875 vector signed char vec_sel (vector signed char,
7877 vector unsigned char);
7878 vector unsigned char vec_sel (vector unsigned char,
7879 vector unsigned char,
7881 vector unsigned char vec_sel (vector unsigned char,
7882 vector unsigned char,
7883 vector unsigned char);
7884 vector bool char vec_sel (vector bool char,
7887 vector bool char vec_sel (vector bool char,
7889 vector unsigned char);
7891 vector signed char vec_sl (vector signed char,
7892 vector unsigned char);
7893 vector unsigned char vec_sl (vector unsigned char,
7894 vector unsigned char);
7895 vector signed short vec_sl (vector signed short, vector unsigned short);
7896 vector unsigned short vec_sl (vector unsigned short,
7897 vector unsigned short);
7898 vector signed int vec_sl (vector signed int, vector unsigned int);
7899 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
7901 vector signed int vec_vslw (vector signed int, vector unsigned int);
7902 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
7904 vector signed short vec_vslh (vector signed short,
7905 vector unsigned short);
7906 vector unsigned short vec_vslh (vector unsigned short,
7907 vector unsigned short);
7909 vector signed char vec_vslb (vector signed char, vector unsigned char);
7910 vector unsigned char vec_vslb (vector unsigned char,
7911 vector unsigned char);
7913 vector float vec_sld (vector float, vector float, const int);
7914 vector signed int vec_sld (vector signed int,
7917 vector unsigned int vec_sld (vector unsigned int,
7918 vector unsigned int,
7920 vector bool int vec_sld (vector bool int,
7923 vector signed short vec_sld (vector signed short,
7924 vector signed short,
7926 vector unsigned short vec_sld (vector unsigned short,
7927 vector unsigned short,
7929 vector bool short vec_sld (vector bool short,
7932 vector pixel vec_sld (vector pixel,
7935 vector signed char vec_sld (vector signed char,
7938 vector unsigned char vec_sld (vector unsigned char,
7939 vector unsigned char,
7941 vector bool char vec_sld (vector bool char,
7945 vector signed int vec_sll (vector signed int,
7946 vector unsigned int);
7947 vector signed int vec_sll (vector signed int,
7948 vector unsigned short);
7949 vector signed int vec_sll (vector signed int,
7950 vector unsigned char);
7951 vector unsigned int vec_sll (vector unsigned int,
7952 vector unsigned int);
7953 vector unsigned int vec_sll (vector unsigned int,
7954 vector unsigned short);
7955 vector unsigned int vec_sll (vector unsigned int,
7956 vector unsigned char);
7957 vector bool int vec_sll (vector bool int,
7958 vector unsigned int);
7959 vector bool int vec_sll (vector bool int,
7960 vector unsigned short);
7961 vector bool int vec_sll (vector bool int,
7962 vector unsigned char);
7963 vector signed short vec_sll (vector signed short,
7964 vector unsigned int);
7965 vector signed short vec_sll (vector signed short,
7966 vector unsigned short);
7967 vector signed short vec_sll (vector signed short,
7968 vector unsigned char);
7969 vector unsigned short vec_sll (vector unsigned short,
7970 vector unsigned int);
7971 vector unsigned short vec_sll (vector unsigned short,
7972 vector unsigned short);
7973 vector unsigned short vec_sll (vector unsigned short,
7974 vector unsigned char);
7975 vector bool short vec_sll (vector bool short, vector unsigned int);
7976 vector bool short vec_sll (vector bool short, vector unsigned short);
7977 vector bool short vec_sll (vector bool short, vector unsigned char);
7978 vector pixel vec_sll (vector pixel, vector unsigned int);
7979 vector pixel vec_sll (vector pixel, vector unsigned short);
7980 vector pixel vec_sll (vector pixel, vector unsigned char);
7981 vector signed char vec_sll (vector signed char, vector unsigned int);
7982 vector signed char vec_sll (vector signed char, vector unsigned short);
7983 vector signed char vec_sll (vector signed char, vector unsigned char);
7984 vector unsigned char vec_sll (vector unsigned char,
7985 vector unsigned int);
7986 vector unsigned char vec_sll (vector unsigned char,
7987 vector unsigned short);
7988 vector unsigned char vec_sll (vector unsigned char,
7989 vector unsigned char);
7990 vector bool char vec_sll (vector bool char, vector unsigned int);
7991 vector bool char vec_sll (vector bool char, vector unsigned short);
7992 vector bool char vec_sll (vector bool char, vector unsigned char);
7994 vector float vec_slo (vector float, vector signed char);
7995 vector float vec_slo (vector float, vector unsigned char);
7996 vector signed int vec_slo (vector signed int, vector signed char);
7997 vector signed int vec_slo (vector signed int, vector unsigned char);
7998 vector unsigned int vec_slo (vector unsigned int, vector signed char);
7999 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
8000 vector signed short vec_slo (vector signed short, vector signed char);
8001 vector signed short vec_slo (vector signed short, vector unsigned char);
8002 vector unsigned short vec_slo (vector unsigned short,
8003 vector signed char);
8004 vector unsigned short vec_slo (vector unsigned short,
8005 vector unsigned char);
8006 vector pixel vec_slo (vector pixel, vector signed char);
8007 vector pixel vec_slo (vector pixel, vector unsigned char);
8008 vector signed char vec_slo (vector signed char, vector signed char);
8009 vector signed char vec_slo (vector signed char, vector unsigned char);
8010 vector unsigned char vec_slo (vector unsigned char, vector signed char);
8011 vector unsigned char vec_slo (vector unsigned char,
8012 vector unsigned char);
8014 vector signed char vec_splat (vector signed char, const int);
8015 vector unsigned char vec_splat (vector unsigned char, const int);
8016 vector bool char vec_splat (vector bool char, const int);
8017 vector signed short vec_splat (vector signed short, const int);
8018 vector unsigned short vec_splat (vector unsigned short, const int);
8019 vector bool short vec_splat (vector bool short, const int);
8020 vector pixel vec_splat (vector pixel, const int);
8021 vector float vec_splat (vector float, const int);
8022 vector signed int vec_splat (vector signed int, const int);
8023 vector unsigned int vec_splat (vector unsigned int, const int);
8024 vector bool int vec_splat (vector bool int, const int);
8026 vector float vec_vspltw (vector float, const int);
8027 vector signed int vec_vspltw (vector signed int, const int);
8028 vector unsigned int vec_vspltw (vector unsigned int, const int);
8029 vector bool int vec_vspltw (vector bool int, const int);
8031 vector bool short vec_vsplth (vector bool short, const int);
8032 vector signed short vec_vsplth (vector signed short, const int);
8033 vector unsigned short vec_vsplth (vector unsigned short, const int);
8034 vector pixel vec_vsplth (vector pixel, const int);
8036 vector signed char vec_vspltb (vector signed char, const int);
8037 vector unsigned char vec_vspltb (vector unsigned char, const int);
8038 vector bool char vec_vspltb (vector bool char, const int);
8040 vector signed char vec_splat_s8 (const int);
8042 vector signed short vec_splat_s16 (const int);
8044 vector signed int vec_splat_s32 (const int);
8046 vector unsigned char vec_splat_u8 (const int);
8048 vector unsigned short vec_splat_u16 (const int);
8050 vector unsigned int vec_splat_u32 (const int);
8052 vector signed char vec_sr (vector signed char, vector unsigned char);
8053 vector unsigned char vec_sr (vector unsigned char,
8054 vector unsigned char);
8055 vector signed short vec_sr (vector signed short,
8056 vector unsigned short);
8057 vector unsigned short vec_sr (vector unsigned short,
8058 vector unsigned short);
8059 vector signed int vec_sr (vector signed int, vector unsigned int);
8060 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
8062 vector signed int vec_vsrw (vector signed int, vector unsigned int);
8063 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
8065 vector signed short vec_vsrh (vector signed short,
8066 vector unsigned short);
8067 vector unsigned short vec_vsrh (vector unsigned short,
8068 vector unsigned short);
8070 vector signed char vec_vsrb (vector signed char, vector unsigned char);
8071 vector unsigned char vec_vsrb (vector unsigned char,
8072 vector unsigned char);
8074 vector signed char vec_sra (vector signed char, vector unsigned char);
8075 vector unsigned char vec_sra (vector unsigned char,
8076 vector unsigned char);
8077 vector signed short vec_sra (vector signed short,
8078 vector unsigned short);
8079 vector unsigned short vec_sra (vector unsigned short,
8080 vector unsigned short);
8081 vector signed int vec_sra (vector signed int, vector unsigned int);
8082 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
8084 vector signed int vec_vsraw (vector signed int, vector unsigned int);
8085 vector unsigned int vec_vsraw (vector unsigned int,
8086 vector unsigned int);
8088 vector signed short vec_vsrah (vector signed short,
8089 vector unsigned short);
8090 vector unsigned short vec_vsrah (vector unsigned short,
8091 vector unsigned short);
8093 vector signed char vec_vsrab (vector signed char, vector unsigned char);
8094 vector unsigned char vec_vsrab (vector unsigned char,
8095 vector unsigned char);
8097 vector signed int vec_srl (vector signed int, vector unsigned int);
8098 vector signed int vec_srl (vector signed int, vector unsigned short);
8099 vector signed int vec_srl (vector signed int, vector unsigned char);
8100 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
8101 vector unsigned int vec_srl (vector unsigned int,
8102 vector unsigned short);
8103 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
8104 vector bool int vec_srl (vector bool int, vector unsigned int);
8105 vector bool int vec_srl (vector bool int, vector unsigned short);
8106 vector bool int vec_srl (vector bool int, vector unsigned char);
8107 vector signed short vec_srl (vector signed short, vector unsigned int);
8108 vector signed short vec_srl (vector signed short,
8109 vector unsigned short);
8110 vector signed short vec_srl (vector signed short, vector unsigned char);
8111 vector unsigned short vec_srl (vector unsigned short,
8112 vector unsigned int);
8113 vector unsigned short vec_srl (vector unsigned short,
8114 vector unsigned short);
8115 vector unsigned short vec_srl (vector unsigned short,
8116 vector unsigned char);
8117 vector bool short vec_srl (vector bool short, vector unsigned int);
8118 vector bool short vec_srl (vector bool short, vector unsigned short);
8119 vector bool short vec_srl (vector bool short, vector unsigned char);
8120 vector pixel vec_srl (vector pixel, vector unsigned int);
8121 vector pixel vec_srl (vector pixel, vector unsigned short);
8122 vector pixel vec_srl (vector pixel, vector unsigned char);
8123 vector signed char vec_srl (vector signed char, vector unsigned int);
8124 vector signed char vec_srl (vector signed char, vector unsigned short);
8125 vector signed char vec_srl (vector signed char, vector unsigned char);
8126 vector unsigned char vec_srl (vector unsigned char,
8127 vector unsigned int);
8128 vector unsigned char vec_srl (vector unsigned char,
8129 vector unsigned short);
8130 vector unsigned char vec_srl (vector unsigned char,
8131 vector unsigned char);
8132 vector bool char vec_srl (vector bool char, vector unsigned int);
8133 vector bool char vec_srl (vector bool char, vector unsigned short);
8134 vector bool char vec_srl (vector bool char, vector unsigned char);
8136 vector float vec_sro (vector float, vector signed char);
8137 vector float vec_sro (vector float, vector unsigned char);
8138 vector signed int vec_sro (vector signed int, vector signed char);
8139 vector signed int vec_sro (vector signed int, vector unsigned char);
8140 vector unsigned int vec_sro (vector unsigned int, vector signed char);
8141 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
8142 vector signed short vec_sro (vector signed short, vector signed char);
8143 vector signed short vec_sro (vector signed short, vector unsigned char);
8144 vector unsigned short vec_sro (vector unsigned short,
8145 vector signed char);
8146 vector unsigned short vec_sro (vector unsigned short,
8147 vector unsigned char);
8148 vector pixel vec_sro (vector pixel, vector signed char);
8149 vector pixel vec_sro (vector pixel, vector unsigned char);
8150 vector signed char vec_sro (vector signed char, vector signed char);
8151 vector signed char vec_sro (vector signed char, vector unsigned char);
8152 vector unsigned char vec_sro (vector unsigned char, vector signed char);
8153 vector unsigned char vec_sro (vector unsigned char,
8154 vector unsigned char);
8156 void vec_st (vector float, int, vector float *);
8157 void vec_st (vector float, int, float *);
8158 void vec_st (vector signed int, int, vector signed int *);
8159 void vec_st (vector signed int, int, int *);
8160 void vec_st (vector unsigned int, int, vector unsigned int *);
8161 void vec_st (vector unsigned int, int, unsigned int *);
8162 void vec_st (vector bool int, int, vector bool int *);
8163 void vec_st (vector bool int, int, unsigned int *);
8164 void vec_st (vector bool int, int, int *);
8165 void vec_st (vector signed short, int, vector signed short *);
8166 void vec_st (vector signed short, int, short *);
8167 void vec_st (vector unsigned short, int, vector unsigned short *);
8168 void vec_st (vector unsigned short, int, unsigned short *);
8169 void vec_st (vector bool short, int, vector bool short *);
8170 void vec_st (vector bool short, int, unsigned short *);
8171 void vec_st (vector pixel, int, vector pixel *);
8172 void vec_st (vector pixel, int, unsigned short *);
8173 void vec_st (vector pixel, int, short *);
8174 void vec_st (vector bool short, int, short *);
8175 void vec_st (vector signed char, int, vector signed char *);
8176 void vec_st (vector signed char, int, signed char *);
8177 void vec_st (vector unsigned char, int, vector unsigned char *);
8178 void vec_st (vector unsigned char, int, unsigned char *);
8179 void vec_st (vector bool char, int, vector bool char *);
8180 void vec_st (vector bool char, int, unsigned char *);
8181 void vec_st (vector bool char, int, signed char *);
8183 void vec_ste (vector signed char, int, signed char *);
8184 void vec_ste (vector unsigned char, int, unsigned char *);
8185 void vec_ste (vector bool char, int, signed char *);
8186 void vec_ste (vector bool char, int, unsigned char *);
8187 void vec_ste (vector signed short, int, short *);
8188 void vec_ste (vector unsigned short, int, unsigned short *);
8189 void vec_ste (vector bool short, int, short *);
8190 void vec_ste (vector bool short, int, unsigned short *);
8191 void vec_ste (vector pixel, int, short *);
8192 void vec_ste (vector pixel, int, unsigned short *);
8193 void vec_ste (vector float, int, float *);
8194 void vec_ste (vector signed int, int, int *);
8195 void vec_ste (vector unsigned int, int, unsigned int *);
8196 void vec_ste (vector bool int, int, int *);
8197 void vec_ste (vector bool int, int, unsigned int *);
8199 void vec_stvewx (vector float, int, float *);
8200 void vec_stvewx (vector signed int, int, int *);
8201 void vec_stvewx (vector unsigned int, int, unsigned int *);
8202 void vec_stvewx (vector bool int, int, int *);
8203 void vec_stvewx (vector bool int, int, unsigned int *);
8205 void vec_stvehx (vector signed short, int, short *);
8206 void vec_stvehx (vector unsigned short, int, unsigned short *);
8207 void vec_stvehx (vector bool short, int, short *);
8208 void vec_stvehx (vector bool short, int, unsigned short *);
8209 void vec_stvehx (vector pixel, int, short *);
8210 void vec_stvehx (vector pixel, int, unsigned short *);
8212 void vec_stvebx (vector signed char, int, signed char *);
8213 void vec_stvebx (vector unsigned char, int, unsigned char *);
8214 void vec_stvebx (vector bool char, int, signed char *);
8215 void vec_stvebx (vector bool char, int, unsigned char *);
8217 void vec_stl (vector float, int, vector float *);
8218 void vec_stl (vector float, int, float *);
8219 void vec_stl (vector signed int, int, vector signed int *);
8220 void vec_stl (vector signed int, int, int *);
8221 void vec_stl (vector unsigned int, int, vector unsigned int *);
8222 void vec_stl (vector unsigned int, int, unsigned int *);
8223 void vec_stl (vector bool int, int, vector bool int *);
8224 void vec_stl (vector bool int, int, unsigned int *);
8225 void vec_stl (vector bool int, int, int *);
8226 void vec_stl (vector signed short, int, vector signed short *);
8227 void vec_stl (vector signed short, int, short *);
8228 void vec_stl (vector unsigned short, int, vector unsigned short *);
8229 void vec_stl (vector unsigned short, int, unsigned short *);
8230 void vec_stl (vector bool short, int, vector bool short *);
8231 void vec_stl (vector bool short, int, unsigned short *);
8232 void vec_stl (vector bool short, int, short *);
8233 void vec_stl (vector pixel, int, vector pixel *);
8234 void vec_stl (vector pixel, int, unsigned short *);
8235 void vec_stl (vector pixel, int, short *);
8236 void vec_stl (vector signed char, int, vector signed char *);
8237 void vec_stl (vector signed char, int, signed char *);
8238 void vec_stl (vector unsigned char, int, vector unsigned char *);
8239 void vec_stl (vector unsigned char, int, unsigned char *);
8240 void vec_stl (vector bool char, int, vector bool char *);
8241 void vec_stl (vector bool char, int, unsigned char *);
8242 void vec_stl (vector bool char, int, signed char *);
8244 vector signed char vec_sub (vector bool char, vector signed char);
8245 vector signed char vec_sub (vector signed char, vector bool char);
8246 vector signed char vec_sub (vector signed char, vector signed char);
8247 vector unsigned char vec_sub (vector bool char, vector unsigned char);
8248 vector unsigned char vec_sub (vector unsigned char, vector bool char);
8249 vector unsigned char vec_sub (vector unsigned char,
8250 vector unsigned char);
8251 vector signed short vec_sub (vector bool short, vector signed short);
8252 vector signed short vec_sub (vector signed short, vector bool short);
8253 vector signed short vec_sub (vector signed short, vector signed short);
8254 vector unsigned short vec_sub (vector bool short,
8255 vector unsigned short);
8256 vector unsigned short vec_sub (vector unsigned short,
8258 vector unsigned short vec_sub (vector unsigned short,
8259 vector unsigned short);
8260 vector signed int vec_sub (vector bool int, vector signed int);
8261 vector signed int vec_sub (vector signed int, vector bool int);
8262 vector signed int vec_sub (vector signed int, vector signed int);
8263 vector unsigned int vec_sub (vector bool int, vector unsigned int);
8264 vector unsigned int vec_sub (vector unsigned int, vector bool int);
8265 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
8266 vector float vec_sub (vector float, vector float);
8268 vector float vec_vsubfp (vector float, vector float);
8270 vector signed int vec_vsubuwm (vector bool int, vector signed int);
8271 vector signed int vec_vsubuwm (vector signed int, vector bool int);
8272 vector signed int vec_vsubuwm (vector signed int, vector signed int);
8273 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
8274 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
8275 vector unsigned int vec_vsubuwm (vector unsigned int,
8276 vector unsigned int);
8278 vector signed short vec_vsubuhm (vector bool short,
8279 vector signed short);
8280 vector signed short vec_vsubuhm (vector signed short,
8282 vector signed short vec_vsubuhm (vector signed short,
8283 vector signed short);
8284 vector unsigned short vec_vsubuhm (vector bool short,
8285 vector unsigned short);
8286 vector unsigned short vec_vsubuhm (vector unsigned short,
8288 vector unsigned short vec_vsubuhm (vector unsigned short,
8289 vector unsigned short);
8291 vector signed char vec_vsububm (vector bool char, vector signed char);
8292 vector signed char vec_vsububm (vector signed char, vector bool char);
8293 vector signed char vec_vsububm (vector signed char, vector signed char);
8294 vector unsigned char vec_vsububm (vector bool char,
8295 vector unsigned char);
8296 vector unsigned char vec_vsububm (vector unsigned char,
8298 vector unsigned char vec_vsububm (vector unsigned char,
8299 vector unsigned char);
8301 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
8303 vector unsigned char vec_subs (vector bool char, vector unsigned char);
8304 vector unsigned char vec_subs (vector unsigned char, vector bool char);
8305 vector unsigned char vec_subs (vector unsigned char,
8306 vector unsigned char);
8307 vector signed char vec_subs (vector bool char, vector signed char);
8308 vector signed char vec_subs (vector signed char, vector bool char);
8309 vector signed char vec_subs (vector signed char, vector signed char);
8310 vector unsigned short vec_subs (vector bool short,
8311 vector unsigned short);
8312 vector unsigned short vec_subs (vector unsigned short,
8314 vector unsigned short vec_subs (vector unsigned short,
8315 vector unsigned short);
8316 vector signed short vec_subs (vector bool short, vector signed short);
8317 vector signed short vec_subs (vector signed short, vector bool short);
8318 vector signed short vec_subs (vector signed short, vector signed short);
8319 vector unsigned int vec_subs (vector bool int, vector unsigned int);
8320 vector unsigned int vec_subs (vector unsigned int, vector bool int);
8321 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
8322 vector signed int vec_subs (vector bool int, vector signed int);
8323 vector signed int vec_subs (vector signed int, vector bool int);
8324 vector signed int vec_subs (vector signed int, vector signed int);
8326 vector signed int vec_vsubsws (vector bool int, vector signed int);
8327 vector signed int vec_vsubsws (vector signed int, vector bool int);
8328 vector signed int vec_vsubsws (vector signed int, vector signed int);
8330 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
8331 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
8332 vector unsigned int vec_vsubuws (vector unsigned int,
8333 vector unsigned int);
8335 vector signed short vec_vsubshs (vector bool short,
8336 vector signed short);
8337 vector signed short vec_vsubshs (vector signed short,
8339 vector signed short vec_vsubshs (vector signed short,
8340 vector signed short);
8342 vector unsigned short vec_vsubuhs (vector bool short,
8343 vector unsigned short);
8344 vector unsigned short vec_vsubuhs (vector unsigned short,
8346 vector unsigned short vec_vsubuhs (vector unsigned short,
8347 vector unsigned short);
8349 vector signed char vec_vsubsbs (vector bool char, vector signed char);
8350 vector signed char vec_vsubsbs (vector signed char, vector bool char);
8351 vector signed char vec_vsubsbs (vector signed char, vector signed char);
8353 vector unsigned char vec_vsububs (vector bool char,
8354 vector unsigned char);
8355 vector unsigned char vec_vsububs (vector unsigned char,
8357 vector unsigned char vec_vsububs (vector unsigned char,
8358 vector unsigned char);
8360 vector unsigned int vec_sum4s (vector unsigned char,
8361 vector unsigned int);
8362 vector signed int vec_sum4s (vector signed char, vector signed int);
8363 vector signed int vec_sum4s (vector signed short, vector signed int);
8365 vector signed int vec_vsum4shs (vector signed short, vector signed int);
8367 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
8369 vector unsigned int vec_vsum4ubs (vector unsigned char,
8370 vector unsigned int);
8372 vector signed int vec_sum2s (vector signed int, vector signed int);
8374 vector signed int vec_sums (vector signed int, vector signed int);
8376 vector float vec_trunc (vector float);
8378 vector signed short vec_unpackh (vector signed char);
8379 vector bool short vec_unpackh (vector bool char);
8380 vector signed int vec_unpackh (vector signed short);
8381 vector bool int vec_unpackh (vector bool short);
8382 vector unsigned int vec_unpackh (vector pixel);
8384 vector bool int vec_vupkhsh (vector bool short);
8385 vector signed int vec_vupkhsh (vector signed short);
8387 vector unsigned int vec_vupkhpx (vector pixel);
8389 vector bool short vec_vupkhsb (vector bool char);
8390 vector signed short vec_vupkhsb (vector signed char);
8392 vector signed short vec_unpackl (vector signed char);
8393 vector bool short vec_unpackl (vector bool char);
8394 vector unsigned int vec_unpackl (vector pixel);
8395 vector signed int vec_unpackl (vector signed short);
8396 vector bool int vec_unpackl (vector bool short);
8398 vector unsigned int vec_vupklpx (vector pixel);
8400 vector bool int vec_vupklsh (vector bool short);
8401 vector signed int vec_vupklsh (vector signed short);
8403 vector bool short vec_vupklsb (vector bool char);
8404 vector signed short vec_vupklsb (vector signed char);
8406 vector float vec_xor (vector float, vector float);
8407 vector float vec_xor (vector float, vector bool int);
8408 vector float vec_xor (vector bool int, vector float);
8409 vector bool int vec_xor (vector bool int, vector bool int);
8410 vector signed int vec_xor (vector bool int, vector signed int);
8411 vector signed int vec_xor (vector signed int, vector bool int);
8412 vector signed int vec_xor (vector signed int, vector signed int);
8413 vector unsigned int vec_xor (vector bool int, vector unsigned int);
8414 vector unsigned int vec_xor (vector unsigned int, vector bool int);
8415 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
8416 vector bool short vec_xor (vector bool short, vector bool short);
8417 vector signed short vec_xor (vector bool short, vector signed short);
8418 vector signed short vec_xor (vector signed short, vector bool short);
8419 vector signed short vec_xor (vector signed short, vector signed short);
8420 vector unsigned short vec_xor (vector bool short,
8421 vector unsigned short);
8422 vector unsigned short vec_xor (vector unsigned short,
8424 vector unsigned short vec_xor (vector unsigned short,
8425 vector unsigned short);
8426 vector signed char vec_xor (vector bool char, vector signed char);
8427 vector bool char vec_xor (vector bool char, vector bool char);
8428 vector signed char vec_xor (vector signed char, vector bool char);
8429 vector signed char vec_xor (vector signed char, vector signed char);
8430 vector unsigned char vec_xor (vector bool char, vector unsigned char);
8431 vector unsigned char vec_xor (vector unsigned char, vector bool char);
8432 vector unsigned char vec_xor (vector unsigned char,
8433 vector unsigned char);
8435 int vec_all_eq (vector signed char, vector bool char);
8436 int vec_all_eq (vector signed char, vector signed char);
8437 int vec_all_eq (vector unsigned char, vector bool char);
8438 int vec_all_eq (vector unsigned char, vector unsigned char);
8439 int vec_all_eq (vector bool char, vector bool char);
8440 int vec_all_eq (vector bool char, vector unsigned char);
8441 int vec_all_eq (vector bool char, vector signed char);
8442 int vec_all_eq (vector signed short, vector bool short);
8443 int vec_all_eq (vector signed short, vector signed short);
8444 int vec_all_eq (vector unsigned short, vector bool short);
8445 int vec_all_eq (vector unsigned short, vector unsigned short);
8446 int vec_all_eq (vector bool short, vector bool short);
8447 int vec_all_eq (vector bool short, vector unsigned short);
8448 int vec_all_eq (vector bool short, vector signed short);
8449 int vec_all_eq (vector pixel, vector pixel);
8450 int vec_all_eq (vector signed int, vector bool int);
8451 int vec_all_eq (vector signed int, vector signed int);
8452 int vec_all_eq (vector unsigned int, vector bool int);
8453 int vec_all_eq (vector unsigned int, vector unsigned int);
8454 int vec_all_eq (vector bool int, vector bool int);
8455 int vec_all_eq (vector bool int, vector unsigned int);
8456 int vec_all_eq (vector bool int, vector signed int);
8457 int vec_all_eq (vector float, vector float);
8459 int vec_all_ge (vector bool char, vector unsigned char);
8460 int vec_all_ge (vector unsigned char, vector bool char);
8461 int vec_all_ge (vector unsigned char, vector unsigned char);
8462 int vec_all_ge (vector bool char, vector signed char);
8463 int vec_all_ge (vector signed char, vector bool char);
8464 int vec_all_ge (vector signed char, vector signed char);
8465 int vec_all_ge (vector bool short, vector unsigned short);
8466 int vec_all_ge (vector unsigned short, vector bool short);
8467 int vec_all_ge (vector unsigned short, vector unsigned short);
8468 int vec_all_ge (vector signed short, vector signed short);
8469 int vec_all_ge (vector bool short, vector signed short);
8470 int vec_all_ge (vector signed short, vector bool short);
8471 int vec_all_ge (vector bool int, vector unsigned int);
8472 int vec_all_ge (vector unsigned int, vector bool int);
8473 int vec_all_ge (vector unsigned int, vector unsigned int);
8474 int vec_all_ge (vector bool int, vector signed int);
8475 int vec_all_ge (vector signed int, vector bool int);
8476 int vec_all_ge (vector signed int, vector signed int);
8477 int vec_all_ge (vector float, vector float);
8479 int vec_all_gt (vector bool char, vector unsigned char);
8480 int vec_all_gt (vector unsigned char, vector bool char);
8481 int vec_all_gt (vector unsigned char, vector unsigned char);
8482 int vec_all_gt (vector bool char, vector signed char);
8483 int vec_all_gt (vector signed char, vector bool char);
8484 int vec_all_gt (vector signed char, vector signed char);
8485 int vec_all_gt (vector bool short, vector unsigned short);
8486 int vec_all_gt (vector unsigned short, vector bool short);
8487 int vec_all_gt (vector unsigned short, vector unsigned short);
8488 int vec_all_gt (vector bool short, vector signed short);
8489 int vec_all_gt (vector signed short, vector bool short);
8490 int vec_all_gt (vector signed short, vector signed short);
8491 int vec_all_gt (vector bool int, vector unsigned int);
8492 int vec_all_gt (vector unsigned int, vector bool int);
8493 int vec_all_gt (vector unsigned int, vector unsigned int);
8494 int vec_all_gt (vector bool int, vector signed int);
8495 int vec_all_gt (vector signed int, vector bool int);
8496 int vec_all_gt (vector signed int, vector signed int);
8497 int vec_all_gt (vector float, vector float);
8499 int vec_all_in (vector float, vector float);
8501 int vec_all_le (vector bool char, vector unsigned char);
8502 int vec_all_le (vector unsigned char, vector bool char);
8503 int vec_all_le (vector unsigned char, vector unsigned char);
8504 int vec_all_le (vector bool char, vector signed char);
8505 int vec_all_le (vector signed char, vector bool char);
8506 int vec_all_le (vector signed char, vector signed char);
8507 int vec_all_le (vector bool short, vector unsigned short);
8508 int vec_all_le (vector unsigned short, vector bool short);
8509 int vec_all_le (vector unsigned short, vector unsigned short);
8510 int vec_all_le (vector bool short, vector signed short);
8511 int vec_all_le (vector signed short, vector bool short);
8512 int vec_all_le (vector signed short, vector signed short);
8513 int vec_all_le (vector bool int, vector unsigned int);
8514 int vec_all_le (vector unsigned int, vector bool int);
8515 int vec_all_le (vector unsigned int, vector unsigned int);
8516 int vec_all_le (vector bool int, vector signed int);
8517 int vec_all_le (vector signed int, vector bool int);
8518 int vec_all_le (vector signed int, vector signed int);
8519 int vec_all_le (vector float, vector float);
8521 int vec_all_lt (vector bool char, vector unsigned char);
8522 int vec_all_lt (vector unsigned char, vector bool char);
8523 int vec_all_lt (vector unsigned char, vector unsigned char);
8524 int vec_all_lt (vector bool char, vector signed char);
8525 int vec_all_lt (vector signed char, vector bool char);
8526 int vec_all_lt (vector signed char, vector signed char);
8527 int vec_all_lt (vector bool short, vector unsigned short);
8528 int vec_all_lt (vector unsigned short, vector bool short);
8529 int vec_all_lt (vector unsigned short, vector unsigned short);
8530 int vec_all_lt (vector bool short, vector signed short);
8531 int vec_all_lt (vector signed short, vector bool short);
8532 int vec_all_lt (vector signed short, vector signed short);
8533 int vec_all_lt (vector bool int, vector unsigned int);
8534 int vec_all_lt (vector unsigned int, vector bool int);
8535 int vec_all_lt (vector unsigned int, vector unsigned int);
8536 int vec_all_lt (vector bool int, vector signed int);
8537 int vec_all_lt (vector signed int, vector bool int);
8538 int vec_all_lt (vector signed int, vector signed int);
8539 int vec_all_lt (vector float, vector float);
8541 int vec_all_nan (vector float);
8543 int vec_all_ne (vector signed char, vector bool char);
8544 int vec_all_ne (vector signed char, vector signed char);
8545 int vec_all_ne (vector unsigned char, vector bool char);
8546 int vec_all_ne (vector unsigned char, vector unsigned char);
8547 int vec_all_ne (vector bool char, vector bool char);
8548 int vec_all_ne (vector bool char, vector unsigned char);
8549 int vec_all_ne (vector bool char, vector signed char);
8550 int vec_all_ne (vector signed short, vector bool short);
8551 int vec_all_ne (vector signed short, vector signed short);
8552 int vec_all_ne (vector unsigned short, vector bool short);
8553 int vec_all_ne (vector unsigned short, vector unsigned short);
8554 int vec_all_ne (vector bool short, vector bool short);
8555 int vec_all_ne (vector bool short, vector unsigned short);
8556 int vec_all_ne (vector bool short, vector signed short);
8557 int vec_all_ne (vector pixel, vector pixel);
8558 int vec_all_ne (vector signed int, vector bool int);
8559 int vec_all_ne (vector signed int, vector signed int);
8560 int vec_all_ne (vector unsigned int, vector bool int);
8561 int vec_all_ne (vector unsigned int, vector unsigned int);
8562 int vec_all_ne (vector bool int, vector bool int);
8563 int vec_all_ne (vector bool int, vector unsigned int);
8564 int vec_all_ne (vector bool int, vector signed int);
8565 int vec_all_ne (vector float, vector float);
8567 int vec_all_nge (vector float, vector float);
8569 int vec_all_ngt (vector float, vector float);
8571 int vec_all_nle (vector float, vector float);
8573 int vec_all_nlt (vector float, vector float);
8575 int vec_all_numeric (vector float);
8577 int vec_any_eq (vector signed char, vector bool char);
8578 int vec_any_eq (vector signed char, vector signed char);
8579 int vec_any_eq (vector unsigned char, vector bool char);
8580 int vec_any_eq (vector unsigned char, vector unsigned char);
8581 int vec_any_eq (vector bool char, vector bool char);
8582 int vec_any_eq (vector bool char, vector unsigned char);
8583 int vec_any_eq (vector bool char, vector signed char);
8584 int vec_any_eq (vector signed short, vector bool short);
8585 int vec_any_eq (vector signed short, vector signed short);
8586 int vec_any_eq (vector unsigned short, vector bool short);
8587 int vec_any_eq (vector unsigned short, vector unsigned short);
8588 int vec_any_eq (vector bool short, vector bool short);
8589 int vec_any_eq (vector bool short, vector unsigned short);
8590 int vec_any_eq (vector bool short, vector signed short);
8591 int vec_any_eq (vector pixel, vector pixel);
8592 int vec_any_eq (vector signed int, vector bool int);
8593 int vec_any_eq (vector signed int, vector signed int);
8594 int vec_any_eq (vector unsigned int, vector bool int);
8595 int vec_any_eq (vector unsigned int, vector unsigned int);
8596 int vec_any_eq (vector bool int, vector bool int);
8597 int vec_any_eq (vector bool int, vector unsigned int);
8598 int vec_any_eq (vector bool int, vector signed int);
8599 int vec_any_eq (vector float, vector float);
8601 int vec_any_ge (vector signed char, vector bool char);
8602 int vec_any_ge (vector unsigned char, vector bool char);
8603 int vec_any_ge (vector unsigned char, vector unsigned char);
8604 int vec_any_ge (vector signed char, vector signed char);
8605 int vec_any_ge (vector bool char, vector unsigned char);
8606 int vec_any_ge (vector bool char, vector signed char);
8607 int vec_any_ge (vector unsigned short, vector bool short);
8608 int vec_any_ge (vector unsigned short, vector unsigned short);
8609 int vec_any_ge (vector signed short, vector signed short);
8610 int vec_any_ge (vector signed short, vector bool short);
8611 int vec_any_ge (vector bool short, vector unsigned short);
8612 int vec_any_ge (vector bool short, vector signed short);
8613 int vec_any_ge (vector signed int, vector bool int);
8614 int vec_any_ge (vector unsigned int, vector bool int);
8615 int vec_any_ge (vector unsigned int, vector unsigned int);
8616 int vec_any_ge (vector signed int, vector signed int);
8617 int vec_any_ge (vector bool int, vector unsigned int);
8618 int vec_any_ge (vector bool int, vector signed int);
8619 int vec_any_ge (vector float, vector float);
8621 int vec_any_gt (vector bool char, vector unsigned char);
8622 int vec_any_gt (vector unsigned char, vector bool char);
8623 int vec_any_gt (vector unsigned char, vector unsigned char);
8624 int vec_any_gt (vector bool char, vector signed char);
8625 int vec_any_gt (vector signed char, vector bool char);
8626 int vec_any_gt (vector signed char, vector signed char);
8627 int vec_any_gt (vector bool short, vector unsigned short);
8628 int vec_any_gt (vector unsigned short, vector bool short);
8629 int vec_any_gt (vector unsigned short, vector unsigned short);
8630 int vec_any_gt (vector bool short, vector signed short);
8631 int vec_any_gt (vector signed short, vector bool short);
8632 int vec_any_gt (vector signed short, vector signed short);
8633 int vec_any_gt (vector bool int, vector unsigned int);
8634 int vec_any_gt (vector unsigned int, vector bool int);
8635 int vec_any_gt (vector unsigned int, vector unsigned int);
8636 int vec_any_gt (vector bool int, vector signed int);
8637 int vec_any_gt (vector signed int, vector bool int);
8638 int vec_any_gt (vector signed int, vector signed int);
8639 int vec_any_gt (vector float, vector float);
8641 int vec_any_le (vector bool char, vector unsigned char);
8642 int vec_any_le (vector unsigned char, vector bool char);
8643 int vec_any_le (vector unsigned char, vector unsigned char);
8644 int vec_any_le (vector bool char, vector signed char);
8645 int vec_any_le (vector signed char, vector bool char);
8646 int vec_any_le (vector signed char, vector signed char);
8647 int vec_any_le (vector bool short, vector unsigned short);
8648 int vec_any_le (vector unsigned short, vector bool short);
8649 int vec_any_le (vector unsigned short, vector unsigned short);
8650 int vec_any_le (vector bool short, vector signed short);
8651 int vec_any_le (vector signed short, vector bool short);
8652 int vec_any_le (vector signed short, vector signed short);
8653 int vec_any_le (vector bool int, vector unsigned int);
8654 int vec_any_le (vector unsigned int, vector bool int);
8655 int vec_any_le (vector unsigned int, vector unsigned int);
8656 int vec_any_le (vector bool int, vector signed int);
8657 int vec_any_le (vector signed int, vector bool int);
8658 int vec_any_le (vector signed int, vector signed int);
8659 int vec_any_le (vector float, vector float);
8661 int vec_any_lt (vector bool char, vector unsigned char);
8662 int vec_any_lt (vector unsigned char, vector bool char);
8663 int vec_any_lt (vector unsigned char, vector unsigned char);
8664 int vec_any_lt (vector bool char, vector signed char);
8665 int vec_any_lt (vector signed char, vector bool char);
8666 int vec_any_lt (vector signed char, vector signed char);
8667 int vec_any_lt (vector bool short, vector unsigned short);
8668 int vec_any_lt (vector unsigned short, vector bool short);
8669 int vec_any_lt (vector unsigned short, vector unsigned short);
8670 int vec_any_lt (vector bool short, vector signed short);
8671 int vec_any_lt (vector signed short, vector bool short);
8672 int vec_any_lt (vector signed short, vector signed short);
8673 int vec_any_lt (vector bool int, vector unsigned int);
8674 int vec_any_lt (vector unsigned int, vector bool int);
8675 int vec_any_lt (vector unsigned int, vector unsigned int);
8676 int vec_any_lt (vector bool int, vector signed int);
8677 int vec_any_lt (vector signed int, vector bool int);
8678 int vec_any_lt (vector signed int, vector signed int);
8679 int vec_any_lt (vector float, vector float);
8681 int vec_any_nan (vector float);
8683 int vec_any_ne (vector signed char, vector bool char);
8684 int vec_any_ne (vector signed char, vector signed char);
8685 int vec_any_ne (vector unsigned char, vector bool char);
8686 int vec_any_ne (vector unsigned char, vector unsigned char);
8687 int vec_any_ne (vector bool char, vector bool char);
8688 int vec_any_ne (vector bool char, vector unsigned char);
8689 int vec_any_ne (vector bool char, vector signed char);
8690 int vec_any_ne (vector signed short, vector bool short);
8691 int vec_any_ne (vector signed short, vector signed short);
8692 int vec_any_ne (vector unsigned short, vector bool short);
8693 int vec_any_ne (vector unsigned short, vector unsigned short);
8694 int vec_any_ne (vector bool short, vector bool short);
8695 int vec_any_ne (vector bool short, vector unsigned short);
8696 int vec_any_ne (vector bool short, vector signed short);
8697 int vec_any_ne (vector pixel, vector pixel);
8698 int vec_any_ne (vector signed int, vector bool int);
8699 int vec_any_ne (vector signed int, vector signed int);
8700 int vec_any_ne (vector unsigned int, vector bool int);
8701 int vec_any_ne (vector unsigned int, vector unsigned int);
8702 int vec_any_ne (vector bool int, vector bool int);
8703 int vec_any_ne (vector bool int, vector unsigned int);
8704 int vec_any_ne (vector bool int, vector signed int);
8705 int vec_any_ne (vector float, vector float);
8707 int vec_any_nge (vector float, vector float);
8709 int vec_any_ngt (vector float, vector float);
8711 int vec_any_nle (vector float, vector float);
8713 int vec_any_nlt (vector float, vector float);
8715 int vec_any_numeric (vector float);
8717 int vec_any_out (vector float, vector float);
8720 @node SPARC VIS Built-in Functions
8721 @subsection SPARC VIS Built-in Functions
8723 GCC supports SIMD operations on the SPARC using both the generic vector
8724 extensions (@pxref{Vector Extensions}) as well as built-in functions for
8725 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
8726 switch, the VIS extension is exposed as the following built-in functions:
8729 typedef int v2si __attribute__ ((vector_size (8)));
8730 typedef short v4hi __attribute__ ((vector_size (8)));
8731 typedef short v2hi __attribute__ ((vector_size (4)));
8732 typedef char v8qi __attribute__ ((vector_size (8)));
8733 typedef char v4qi __attribute__ ((vector_size (4)));
8735 void * __builtin_vis_alignaddr (void *, long);
8736 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
8737 v2si __builtin_vis_faligndatav2si (v2si, v2si);
8738 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
8739 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
8741 v4hi __builtin_vis_fexpand (v4qi);
8743 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
8744 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
8745 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
8746 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
8747 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
8748 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
8749 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
8751 v4qi __builtin_vis_fpack16 (v4hi);
8752 v8qi __builtin_vis_fpack32 (v2si, v2si);
8753 v2hi __builtin_vis_fpackfix (v2si);
8754 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
8756 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
8759 @node Target Format Checks
8760 @section Format Checks Specific to Particular Target Machines
8762 For some target machines, GCC supports additional options to the
8764 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
8767 * Solaris Format Checks::
8770 @node Solaris Format Checks
8771 @subsection Solaris Format Checks
8773 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
8774 check. @code{cmn_err} accepts a subset of the standard @code{printf}
8775 conversions, and the two-argument @code{%b} conversion for displaying
8776 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
8779 @section Pragmas Accepted by GCC
8783 GCC supports several types of pragmas, primarily in order to compile
8784 code originally written for other compilers. Note that in general
8785 we do not recommend the use of pragmas; @xref{Function Attributes},
8786 for further explanation.
8790 * RS/6000 and PowerPC Pragmas::
8793 * Symbol-Renaming Pragmas::
8794 * Structure-Packing Pragmas::
8799 @subsection ARM Pragmas
8801 The ARM target defines pragmas for controlling the default addition of
8802 @code{long_call} and @code{short_call} attributes to functions.
8803 @xref{Function Attributes}, for information about the effects of these
8808 @cindex pragma, long_calls
8809 Set all subsequent functions to have the @code{long_call} attribute.
8812 @cindex pragma, no_long_calls
8813 Set all subsequent functions to have the @code{short_call} attribute.
8815 @item long_calls_off
8816 @cindex pragma, long_calls_off
8817 Do not affect the @code{long_call} or @code{short_call} attributes of
8818 subsequent functions.
8821 @node RS/6000 and PowerPC Pragmas
8822 @subsection RS/6000 and PowerPC Pragmas
8824 The RS/6000 and PowerPC targets define one pragma for controlling
8825 whether or not the @code{longcall} attribute is added to function
8826 declarations by default. This pragma overrides the @option{-mlongcall}
8827 option, but not the @code{longcall} and @code{shortcall} attributes.
8828 @xref{RS/6000 and PowerPC Options}, for more information about when long
8829 calls are and are not necessary.
8833 @cindex pragma, longcall
8834 Apply the @code{longcall} attribute to all subsequent function
8838 Do not apply the @code{longcall} attribute to subsequent function
8842 @c Describe c4x pragmas here.
8843 @c Describe h8300 pragmas here.
8844 @c Describe sh pragmas here.
8845 @c Describe v850 pragmas here.
8847 @node Darwin Pragmas
8848 @subsection Darwin Pragmas
8850 The following pragmas are available for all architectures running the
8851 Darwin operating system. These are useful for compatibility with other
8855 @item mark @var{tokens}@dots{}
8856 @cindex pragma, mark
8857 This pragma is accepted, but has no effect.
8859 @item options align=@var{alignment}
8860 @cindex pragma, options align
8861 This pragma sets the alignment of fields in structures. The values of
8862 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
8863 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
8864 properly; to restore the previous setting, use @code{reset} for the
8867 @item segment @var{tokens}@dots{}
8868 @cindex pragma, segment
8869 This pragma is accepted, but has no effect.
8871 @item unused (@var{var} [, @var{var}]@dots{})
8872 @cindex pragma, unused
8873 This pragma declares variables to be possibly unused. GCC will not
8874 produce warnings for the listed variables. The effect is similar to
8875 that of the @code{unused} attribute, except that this pragma may appear
8876 anywhere within the variables' scopes.
8879 @node Solaris Pragmas
8880 @subsection Solaris Pragmas
8882 The Solaris target supports @code{#pragma redefine_extname}
8883 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
8884 @code{#pragma} directives for compatibility with the system compiler.
8887 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
8888 @cindex pragma, align
8890 Increase the minimum alignment of each @var{variable} to @var{alignment}.
8891 This is the same as GCC's @code{aligned} attribute @pxref{Variable
8892 Attributes}). Macro expansion occurs on the arguments to this pragma
8893 when compiling C and Objective-C. It does not currently occur when
8894 compiling C++, but this is a bug which may be fixed in a future
8897 @item fini (@var{function} [, @var{function}]...)
8898 @cindex pragma, fini
8900 This pragma causes each listed @var{function} to be called after
8901 main, or during shared module unloading, by adding a call to the
8902 @code{.fini} section.
8904 @item init (@var{function} [, @var{function}]...)
8905 @cindex pragma, init
8907 This pragma causes each listed @var{function} to be called during
8908 initialization (before @code{main}) or during shared module loading, by
8909 adding a call to the @code{.init} section.
8913 @node Symbol-Renaming Pragmas
8914 @subsection Symbol-Renaming Pragmas
8916 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
8917 supports two @code{#pragma} directives which change the name used in
8918 assembly for a given declaration. These pragmas are only available on
8919 platforms whose system headers need them. To get this effect on all
8920 platforms supported by GCC, use the asm labels extension (@pxref{Asm
8924 @item redefine_extname @var{oldname} @var{newname}
8925 @cindex pragma, redefine_extname
8927 This pragma gives the C function @var{oldname} the assembly symbol
8928 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
8929 will be defined if this pragma is available (currently only on
8932 @item extern_prefix @var{string}
8933 @cindex pragma, extern_prefix
8935 This pragma causes all subsequent external function and variable
8936 declarations to have @var{string} prepended to their assembly symbols.
8937 This effect may be terminated with another @code{extern_prefix} pragma
8938 whose argument is an empty string. The preprocessor macro
8939 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
8940 available (currently only on Tru64 UNIX)@.
8943 These pragmas and the asm labels extension interact in a complicated
8944 manner. Here are some corner cases you may want to be aware of.
8947 @item Both pragmas silently apply only to declarations with external
8948 linkage. Asm labels do not have this restriction.
8950 @item In C++, both pragmas silently apply only to declarations with
8951 ``C'' linkage. Again, asm labels do not have this restriction.
8953 @item If any of the three ways of changing the assembly name of a
8954 declaration is applied to a declaration whose assembly name has
8955 already been determined (either by a previous use of one of these
8956 features, or because the compiler needed the assembly name in order to
8957 generate code), and the new name is different, a warning issues and
8958 the name does not change.
8960 @item The @var{oldname} used by @code{#pragma redefine_extname} is
8961 always the C-language name.
8963 @item If @code{#pragma extern_prefix} is in effect, and a declaration
8964 occurs with an asm label attached, the prefix is silently ignored for
8967 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
8968 apply to the same declaration, whichever triggered first wins, and a
8969 warning issues if they contradict each other. (We would like to have
8970 @code{#pragma redefine_extname} always win, for consistency with asm
8971 labels, but if @code{#pragma extern_prefix} triggers first we have no
8972 way of knowing that that happened.)
8975 @node Structure-Packing Pragmas
8976 @subsection Structure-Packing Pragmas
8978 For compatibility with Win32, GCC supports a set of @code{#pragma}
8979 directives which change the maximum alignment of members of structures,
8980 unions, and classes subsequently defined. The @var{n} value below always
8981 is required to be a small power of two and specifies the new alignment
8985 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
8986 @item @code{#pragma pack()} sets the alignment to the one that was in
8987 effect when compilation started (see also command line option
8988 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
8989 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
8990 setting on an internal stack and then optionally sets the new alignment.
8991 @item @code{#pragma pack(pop)} restores the alignment setting to the one
8992 saved at the top of the internal stack (and removes that stack entry).
8993 Note that @code{#pragma pack([@var{n}])} does not influence this internal
8994 stack; thus it is possible to have @code{#pragma pack(push)} followed by
8995 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
8996 @code{#pragma pack(pop)}.
9000 @subsection Weak Pragmas
9002 For compatibility with SVR4, GCC supports a set of @code{#pragma}
9003 directives for declaring symbols to be weak, and defining weak
9007 @item #pragma weak @var{symbol}
9008 @cindex pragma, weak
9009 This pragma declares @var{symbol} to be weak, as if the declaration
9010 had the attribute of the same name. The pragma may appear before
9011 or after the declaration of @var{symbol}, but must appear before
9012 either its first use or its definition. It is not an error for
9013 @var{symbol} to never be defined at all.
9015 @item #pragma weak @var{symbol1} = @var{symbol2}
9016 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
9017 It is an error if @var{symbol2} is not defined in the current
9021 @node Unnamed Fields
9022 @section Unnamed struct/union fields within structs/unions
9026 For compatibility with other compilers, GCC allows you to define
9027 a structure or union that contains, as fields, structures and unions
9028 without names. For example:
9041 In this example, the user would be able to access members of the unnamed
9042 union with code like @samp{foo.b}. Note that only unnamed structs and
9043 unions are allowed, you may not have, for example, an unnamed
9046 You must never create such structures that cause ambiguous field definitions.
9047 For example, this structure:
9058 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
9059 Such constructs are not supported and must be avoided. In the future,
9060 such constructs may be detected and treated as compilation errors.
9062 @opindex fms-extensions
9063 Unless @option{-fms-extensions} is used, the unnamed field must be a
9064 structure or union definition without a tag (for example, @samp{struct
9065 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
9066 also be a definition with a tag such as @samp{struct foo @{ int a;
9067 @};}, a reference to a previously defined structure or union such as
9068 @samp{struct foo;}, or a reference to a @code{typedef} name for a
9069 previously defined structure or union type.
9072 @section Thread-Local Storage
9073 @cindex Thread-Local Storage
9074 @cindex @acronym{TLS}
9077 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
9078 are allocated such that there is one instance of the variable per extant
9079 thread. The run-time model GCC uses to implement this originates
9080 in the IA-64 processor-specific ABI, but has since been migrated
9081 to other processors as well. It requires significant support from
9082 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
9083 system libraries (@file{libc.so} and @file{libpthread.so}), so it
9084 is not available everywhere.
9086 At the user level, the extension is visible with a new storage
9087 class keyword: @code{__thread}. For example:
9091 extern __thread struct state s;
9092 static __thread char *p;
9095 The @code{__thread} specifier may be used alone, with the @code{extern}
9096 or @code{static} specifiers, but with no other storage class specifier.
9097 When used with @code{extern} or @code{static}, @code{__thread} must appear
9098 immediately after the other storage class specifier.
9100 The @code{__thread} specifier may be applied to any global, file-scoped
9101 static, function-scoped static, or static data member of a class. It may
9102 not be applied to block-scoped automatic or non-static data member.
9104 When the address-of operator is applied to a thread-local variable, it is
9105 evaluated at run-time and returns the address of the current thread's
9106 instance of that variable. An address so obtained may be used by any
9107 thread. When a thread terminates, any pointers to thread-local variables
9108 in that thread become invalid.
9110 No static initialization may refer to the address of a thread-local variable.
9112 In C++, if an initializer is present for a thread-local variable, it must
9113 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
9116 See @uref{http://people.redhat.com/drepper/tls.pdf,
9117 ELF Handling For Thread-Local Storage} for a detailed explanation of
9118 the four thread-local storage addressing models, and how the run-time
9119 is expected to function.
9122 * C99 Thread-Local Edits::
9123 * C++98 Thread-Local Edits::
9126 @node C99 Thread-Local Edits
9127 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
9129 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
9130 that document the exact semantics of the language extension.
9134 @cite{5.1.2 Execution environments}
9136 Add new text after paragraph 1
9139 Within either execution environment, a @dfn{thread} is a flow of
9140 control within a program. It is implementation defined whether
9141 or not there may be more than one thread associated with a program.
9142 It is implementation defined how threads beyond the first are
9143 created, the name and type of the function called at thread
9144 startup, and how threads may be terminated. However, objects
9145 with thread storage duration shall be initialized before thread
9150 @cite{6.2.4 Storage durations of objects}
9152 Add new text before paragraph 3
9155 An object whose identifier is declared with the storage-class
9156 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
9157 Its lifetime is the entire execution of the thread, and its
9158 stored value is initialized only once, prior to thread startup.
9162 @cite{6.4.1 Keywords}
9164 Add @code{__thread}.
9167 @cite{6.7.1 Storage-class specifiers}
9169 Add @code{__thread} to the list of storage class specifiers in
9172 Change paragraph 2 to
9175 With the exception of @code{__thread}, at most one storage-class
9176 specifier may be given [@dots{}]. The @code{__thread} specifier may
9177 be used alone, or immediately following @code{extern} or
9181 Add new text after paragraph 6
9184 The declaration of an identifier for a variable that has
9185 block scope that specifies @code{__thread} shall also
9186 specify either @code{extern} or @code{static}.
9188 The @code{__thread} specifier shall be used only with
9193 @node C++98 Thread-Local Edits
9194 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
9196 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
9197 that document the exact semantics of the language extension.
9201 @b{[intro.execution]}
9203 New text after paragraph 4
9206 A @dfn{thread} is a flow of control within the abstract machine.
9207 It is implementation defined whether or not there may be more than
9211 New text after paragraph 7
9214 It is unspecified whether additional action must be taken to
9215 ensure when and whether side effects are visible to other threads.
9221 Add @code{__thread}.
9224 @b{[basic.start.main]}
9226 Add after paragraph 5
9229 The thread that begins execution at the @code{main} function is called
9230 the @dfn{main thread}. It is implementation defined how functions
9231 beginning threads other than the main thread are designated or typed.
9232 A function so designated, as well as the @code{main} function, is called
9233 a @dfn{thread startup function}. It is implementation defined what
9234 happens if a thread startup function returns. It is implementation
9235 defined what happens to other threads when any thread calls @code{exit}.
9239 @b{[basic.start.init]}
9241 Add after paragraph 4
9244 The storage for an object of thread storage duration shall be
9245 statically initialized before the first statement of the thread startup
9246 function. An object of thread storage duration shall not require
9247 dynamic initialization.
9251 @b{[basic.start.term]}
9253 Add after paragraph 3
9256 The type of an object with thread storage duration shall not have a
9257 non-trivial destructor, nor shall it be an array type whose elements
9258 (directly or indirectly) have non-trivial destructors.
9264 Add ``thread storage duration'' to the list in paragraph 1.
9269 Thread, static, and automatic storage durations are associated with
9270 objects introduced by declarations [@dots{}].
9273 Add @code{__thread} to the list of specifiers in paragraph 3.
9276 @b{[basic.stc.thread]}
9278 New section before @b{[basic.stc.static]}
9281 The keyword @code{__thread} applied to a non-local object gives the
9282 object thread storage duration.
9284 A local variable or class data member declared both @code{static}
9285 and @code{__thread} gives the variable or member thread storage
9290 @b{[basic.stc.static]}
9295 All objects which have neither thread storage duration, dynamic
9296 storage duration nor are local [@dots{}].
9302 Add @code{__thread} to the list in paragraph 1.
9307 With the exception of @code{__thread}, at most one
9308 @var{storage-class-specifier} shall appear in a given
9309 @var{decl-specifier-seq}. The @code{__thread} specifier may
9310 be used alone, or immediately following the @code{extern} or
9311 @code{static} specifiers. [@dots{}]
9314 Add after paragraph 5
9317 The @code{__thread} specifier can be applied only to the names of objects
9318 and to anonymous unions.
9324 Add after paragraph 6
9327 Non-@code{static} members shall not be @code{__thread}.
9331 @node C++ Extensions
9332 @chapter Extensions to the C++ Language
9333 @cindex extensions, C++ language
9334 @cindex C++ language extensions
9336 The GNU compiler provides these extensions to the C++ language (and you
9337 can also use most of the C language extensions in your C++ programs). If you
9338 want to write code that checks whether these features are available, you can
9339 test for the GNU compiler the same way as for C programs: check for a
9340 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
9341 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
9342 Predefined Macros,cpp,The GNU C Preprocessor}).
9345 * Volatiles:: What constitutes an access to a volatile object.
9346 * Restricted Pointers:: C99 restricted pointers and references.
9347 * Vague Linkage:: Where G++ puts inlines, vtables and such.
9348 * C++ Interface:: You can use a single C++ header file for both
9349 declarations and definitions.
9350 * Template Instantiation:: Methods for ensuring that exactly one copy of
9351 each needed template instantiation is emitted.
9352 * Bound member functions:: You can extract a function pointer to the
9353 method denoted by a @samp{->*} or @samp{.*} expression.
9354 * C++ Attributes:: Variable, function, and type attributes for C++ only.
9355 * Strong Using:: Strong using-directives for namespace composition.
9356 * Java Exceptions:: Tweaking exception handling to work with Java.
9357 * Deprecated Features:: Things will disappear from g++.
9358 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
9362 @section When is a Volatile Object Accessed?
9363 @cindex accessing volatiles
9364 @cindex volatile read
9365 @cindex volatile write
9366 @cindex volatile access
9368 Both the C and C++ standard have the concept of volatile objects. These
9369 are normally accessed by pointers and used for accessing hardware. The
9370 standards encourage compilers to refrain from optimizations
9371 concerning accesses to volatile objects that it might perform on
9372 non-volatile objects. The C standard leaves it implementation defined
9373 as to what constitutes a volatile access. The C++ standard omits to
9374 specify this, except to say that C++ should behave in a similar manner
9375 to C with respect to volatiles, where possible. The minimum either
9376 standard specifies is that at a sequence point all previous accesses to
9377 volatile objects have stabilized and no subsequent accesses have
9378 occurred. Thus an implementation is free to reorder and combine
9379 volatile accesses which occur between sequence points, but cannot do so
9380 for accesses across a sequence point. The use of volatiles does not
9381 allow you to violate the restriction on updating objects multiple times
9382 within a sequence point.
9384 In most expressions, it is intuitively obvious what is a read and what is
9385 a write. For instance
9388 volatile int *dst = @var{somevalue};
9389 volatile int *src = @var{someothervalue};
9394 will cause a read of the volatile object pointed to by @var{src} and stores the
9395 value into the volatile object pointed to by @var{dst}. There is no
9396 guarantee that these reads and writes are atomic, especially for objects
9397 larger than @code{int}.
9399 Less obvious expressions are where something which looks like an access
9400 is used in a void context. An example would be,
9403 volatile int *src = @var{somevalue};
9407 With C, such expressions are rvalues, and as rvalues cause a read of
9408 the object, GCC interprets this as a read of the volatile being pointed
9409 to. The C++ standard specifies that such expressions do not undergo
9410 lvalue to rvalue conversion, and that the type of the dereferenced
9411 object may be incomplete. The C++ standard does not specify explicitly
9412 that it is this lvalue to rvalue conversion which is responsible for
9413 causing an access. However, there is reason to believe that it is,
9414 because otherwise certain simple expressions become undefined. However,
9415 because it would surprise most programmers, G++ treats dereferencing a
9416 pointer to volatile object of complete type in a void context as a read
9417 of the object. When the object has incomplete type, G++ issues a
9422 struct T @{int m;@};
9423 volatile S *ptr1 = @var{somevalue};
9424 volatile T *ptr2 = @var{somevalue};
9429 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
9430 causes a read of the object pointed to. If you wish to force an error on
9431 the first case, you must force a conversion to rvalue with, for instance
9432 a static cast, @code{static_cast<S>(*ptr1)}.
9434 When using a reference to volatile, G++ does not treat equivalent
9435 expressions as accesses to volatiles, but instead issues a warning that
9436 no volatile is accessed. The rationale for this is that otherwise it
9437 becomes difficult to determine where volatile access occur, and not
9438 possible to ignore the return value from functions returning volatile
9439 references. Again, if you wish to force a read, cast the reference to
9442 @node Restricted Pointers
9443 @section Restricting Pointer Aliasing
9444 @cindex restricted pointers
9445 @cindex restricted references
9446 @cindex restricted this pointer
9448 As with the C front end, G++ understands the C99 feature of restricted pointers,
9449 specified with the @code{__restrict__}, or @code{__restrict} type
9450 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
9451 language flag, @code{restrict} is not a keyword in C++.
9453 In addition to allowing restricted pointers, you can specify restricted
9454 references, which indicate that the reference is not aliased in the local
9458 void fn (int *__restrict__ rptr, int &__restrict__ rref)
9465 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
9466 @var{rref} refers to a (different) unaliased integer.
9468 You may also specify whether a member function's @var{this} pointer is
9469 unaliased by using @code{__restrict__} as a member function qualifier.
9472 void T::fn () __restrict__
9479 Within the body of @code{T::fn}, @var{this} will have the effective
9480 definition @code{T *__restrict__ const this}. Notice that the
9481 interpretation of a @code{__restrict__} member function qualifier is
9482 different to that of @code{const} or @code{volatile} qualifier, in that it
9483 is applied to the pointer rather than the object. This is consistent with
9484 other compilers which implement restricted pointers.
9486 As with all outermost parameter qualifiers, @code{__restrict__} is
9487 ignored in function definition matching. This means you only need to
9488 specify @code{__restrict__} in a function definition, rather than
9489 in a function prototype as well.
9492 @section Vague Linkage
9493 @cindex vague linkage
9495 There are several constructs in C++ which require space in the object
9496 file but are not clearly tied to a single translation unit. We say that
9497 these constructs have ``vague linkage''. Typically such constructs are
9498 emitted wherever they are needed, though sometimes we can be more
9502 @item Inline Functions
9503 Inline functions are typically defined in a header file which can be
9504 included in many different compilations. Hopefully they can usually be
9505 inlined, but sometimes an out-of-line copy is necessary, if the address
9506 of the function is taken or if inlining fails. In general, we emit an
9507 out-of-line copy in all translation units where one is needed. As an
9508 exception, we only emit inline virtual functions with the vtable, since
9509 it will always require a copy.
9511 Local static variables and string constants used in an inline function
9512 are also considered to have vague linkage, since they must be shared
9513 between all inlined and out-of-line instances of the function.
9517 C++ virtual functions are implemented in most compilers using a lookup
9518 table, known as a vtable. The vtable contains pointers to the virtual
9519 functions provided by a class, and each object of the class contains a
9520 pointer to its vtable (or vtables, in some multiple-inheritance
9521 situations). If the class declares any non-inline, non-pure virtual
9522 functions, the first one is chosen as the ``key method'' for the class,
9523 and the vtable is only emitted in the translation unit where the key
9526 @emph{Note:} If the chosen key method is later defined as inline, the
9527 vtable will still be emitted in every translation unit which defines it.
9528 Make sure that any inline virtuals are declared inline in the class
9529 body, even if they are not defined there.
9531 @item type_info objects
9534 C++ requires information about types to be written out in order to
9535 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
9536 For polymorphic classes (classes with virtual functions), the type_info
9537 object is written out along with the vtable so that @samp{dynamic_cast}
9538 can determine the dynamic type of a class object at runtime. For all
9539 other types, we write out the type_info object when it is used: when
9540 applying @samp{typeid} to an expression, throwing an object, or
9541 referring to a type in a catch clause or exception specification.
9543 @item Template Instantiations
9544 Most everything in this section also applies to template instantiations,
9545 but there are other options as well.
9546 @xref{Template Instantiation,,Where's the Template?}.
9550 When used with GNU ld version 2.8 or later on an ELF system such as
9551 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
9552 these constructs will be discarded at link time. This is known as
9555 On targets that don't support COMDAT, but do support weak symbols, GCC
9556 will use them. This way one copy will override all the others, but
9557 the unused copies will still take up space in the executable.
9559 For targets which do not support either COMDAT or weak symbols,
9560 most entities with vague linkage will be emitted as local symbols to
9561 avoid duplicate definition errors from the linker. This will not happen
9562 for local statics in inlines, however, as having multiple copies will
9563 almost certainly break things.
9565 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
9566 another way to control placement of these constructs.
9569 @section #pragma interface and implementation
9571 @cindex interface and implementation headers, C++
9572 @cindex C++ interface and implementation headers
9573 @cindex pragmas, interface and implementation
9575 @code{#pragma interface} and @code{#pragma implementation} provide the
9576 user with a way of explicitly directing the compiler to emit entities
9577 with vague linkage (and debugging information) in a particular
9580 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
9581 most cases, because of COMDAT support and the ``key method'' heuristic
9582 mentioned in @ref{Vague Linkage}. Using them can actually cause your
9583 program to grow due to unnecessary out-of-line copies of inline
9584 functions. Currently (3.4) the only benefit of these
9585 @code{#pragma}s is reduced duplication of debugging information, and
9586 that should be addressed soon on DWARF 2 targets with the use of
9590 @item #pragma interface
9591 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
9592 @kindex #pragma interface
9593 Use this directive in @emph{header files} that define object classes, to save
9594 space in most of the object files that use those classes. Normally,
9595 local copies of certain information (backup copies of inline member
9596 functions, debugging information, and the internal tables that implement
9597 virtual functions) must be kept in each object file that includes class
9598 definitions. You can use this pragma to avoid such duplication. When a
9599 header file containing @samp{#pragma interface} is included in a
9600 compilation, this auxiliary information will not be generated (unless
9601 the main input source file itself uses @samp{#pragma implementation}).
9602 Instead, the object files will contain references to be resolved at link
9605 The second form of this directive is useful for the case where you have
9606 multiple headers with the same name in different directories. If you
9607 use this form, you must specify the same string to @samp{#pragma
9610 @item #pragma implementation
9611 @itemx #pragma implementation "@var{objects}.h"
9612 @kindex #pragma implementation
9613 Use this pragma in a @emph{main input file}, when you want full output from
9614 included header files to be generated (and made globally visible). The
9615 included header file, in turn, should use @samp{#pragma interface}.
9616 Backup copies of inline member functions, debugging information, and the
9617 internal tables used to implement virtual functions are all generated in
9618 implementation files.
9620 @cindex implied @code{#pragma implementation}
9621 @cindex @code{#pragma implementation}, implied
9622 @cindex naming convention, implementation headers
9623 If you use @samp{#pragma implementation} with no argument, it applies to
9624 an include file with the same basename@footnote{A file's @dfn{basename}
9625 was the name stripped of all leading path information and of trailing
9626 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
9627 file. For example, in @file{allclass.cc}, giving just
9628 @samp{#pragma implementation}
9629 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
9631 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
9632 an implementation file whenever you would include it from
9633 @file{allclass.cc} even if you never specified @samp{#pragma
9634 implementation}. This was deemed to be more trouble than it was worth,
9635 however, and disabled.
9637 Use the string argument if you want a single implementation file to
9638 include code from multiple header files. (You must also use
9639 @samp{#include} to include the header file; @samp{#pragma
9640 implementation} only specifies how to use the file---it doesn't actually
9643 There is no way to split up the contents of a single header file into
9644 multiple implementation files.
9647 @cindex inlining and C++ pragmas
9648 @cindex C++ pragmas, effect on inlining
9649 @cindex pragmas in C++, effect on inlining
9650 @samp{#pragma implementation} and @samp{#pragma interface} also have an
9651 effect on function inlining.
9653 If you define a class in a header file marked with @samp{#pragma
9654 interface}, the effect on an inline function defined in that class is
9655 similar to an explicit @code{extern} declaration---the compiler emits
9656 no code at all to define an independent version of the function. Its
9657 definition is used only for inlining with its callers.
9659 @opindex fno-implement-inlines
9660 Conversely, when you include the same header file in a main source file
9661 that declares it as @samp{#pragma implementation}, the compiler emits
9662 code for the function itself; this defines a version of the function
9663 that can be found via pointers (or by callers compiled without
9664 inlining). If all calls to the function can be inlined, you can avoid
9665 emitting the function by compiling with @option{-fno-implement-inlines}.
9666 If any calls were not inlined, you will get linker errors.
9668 @node Template Instantiation
9669 @section Where's the Template?
9670 @cindex template instantiation
9672 C++ templates are the first language feature to require more
9673 intelligence from the environment than one usually finds on a UNIX
9674 system. Somehow the compiler and linker have to make sure that each
9675 template instance occurs exactly once in the executable if it is needed,
9676 and not at all otherwise. There are two basic approaches to this
9677 problem, which are referred to as the Borland model and the Cfront model.
9681 Borland C++ solved the template instantiation problem by adding the code
9682 equivalent of common blocks to their linker; the compiler emits template
9683 instances in each translation unit that uses them, and the linker
9684 collapses them together. The advantage of this model is that the linker
9685 only has to consider the object files themselves; there is no external
9686 complexity to worry about. This disadvantage is that compilation time
9687 is increased because the template code is being compiled repeatedly.
9688 Code written for this model tends to include definitions of all
9689 templates in the header file, since they must be seen to be
9693 The AT&T C++ translator, Cfront, solved the template instantiation
9694 problem by creating the notion of a template repository, an
9695 automatically maintained place where template instances are stored. A
9696 more modern version of the repository works as follows: As individual
9697 object files are built, the compiler places any template definitions and
9698 instantiations encountered in the repository. At link time, the link
9699 wrapper adds in the objects in the repository and compiles any needed
9700 instances that were not previously emitted. The advantages of this
9701 model are more optimal compilation speed and the ability to use the
9702 system linker; to implement the Borland model a compiler vendor also
9703 needs to replace the linker. The disadvantages are vastly increased
9704 complexity, and thus potential for error; for some code this can be
9705 just as transparent, but in practice it can been very difficult to build
9706 multiple programs in one directory and one program in multiple
9707 directories. Code written for this model tends to separate definitions
9708 of non-inline member templates into a separate file, which should be
9709 compiled separately.
9712 When used with GNU ld version 2.8 or later on an ELF system such as
9713 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
9714 Borland model. On other systems, G++ implements neither automatic
9717 A future version of G++ will support a hybrid model whereby the compiler
9718 will emit any instantiations for which the template definition is
9719 included in the compile, and store template definitions and
9720 instantiation context information into the object file for the rest.
9721 The link wrapper will extract that information as necessary and invoke
9722 the compiler to produce the remaining instantiations. The linker will
9723 then combine duplicate instantiations.
9725 In the mean time, you have the following options for dealing with
9726 template instantiations:
9731 Compile your template-using code with @option{-frepo}. The compiler will
9732 generate files with the extension @samp{.rpo} listing all of the
9733 template instantiations used in the corresponding object files which
9734 could be instantiated there; the link wrapper, @samp{collect2}, will
9735 then update the @samp{.rpo} files to tell the compiler where to place
9736 those instantiations and rebuild any affected object files. The
9737 link-time overhead is negligible after the first pass, as the compiler
9738 will continue to place the instantiations in the same files.
9740 This is your best option for application code written for the Borland
9741 model, as it will just work. Code written for the Cfront model will
9742 need to be modified so that the template definitions are available at
9743 one or more points of instantiation; usually this is as simple as adding
9744 @code{#include <tmethods.cc>} to the end of each template header.
9746 For library code, if you want the library to provide all of the template
9747 instantiations it needs, just try to link all of its object files
9748 together; the link will fail, but cause the instantiations to be
9749 generated as a side effect. Be warned, however, that this may cause
9750 conflicts if multiple libraries try to provide the same instantiations.
9751 For greater control, use explicit instantiation as described in the next
9755 @opindex fno-implicit-templates
9756 Compile your code with @option{-fno-implicit-templates} to disable the
9757 implicit generation of template instances, and explicitly instantiate
9758 all the ones you use. This approach requires more knowledge of exactly
9759 which instances you need than do the others, but it's less
9760 mysterious and allows greater control. You can scatter the explicit
9761 instantiations throughout your program, perhaps putting them in the
9762 translation units where the instances are used or the translation units
9763 that define the templates themselves; you can put all of the explicit
9764 instantiations you need into one big file; or you can create small files
9771 template class Foo<int>;
9772 template ostream& operator <<
9773 (ostream&, const Foo<int>&);
9776 for each of the instances you need, and create a template instantiation
9779 If you are using Cfront-model code, you can probably get away with not
9780 using @option{-fno-implicit-templates} when compiling files that don't
9781 @samp{#include} the member template definitions.
9783 If you use one big file to do the instantiations, you may want to
9784 compile it without @option{-fno-implicit-templates} so you get all of the
9785 instances required by your explicit instantiations (but not by any
9786 other files) without having to specify them as well.
9788 G++ has extended the template instantiation syntax given in the ISO
9789 standard to allow forward declaration of explicit instantiations
9790 (with @code{extern}), instantiation of the compiler support data for a
9791 template class (i.e.@: the vtable) without instantiating any of its
9792 members (with @code{inline}), and instantiation of only the static data
9793 members of a template class, without the support data or member
9794 functions (with (@code{static}):
9797 extern template int max (int, int);
9798 inline template class Foo<int>;
9799 static template class Foo<int>;
9803 Do nothing. Pretend G++ does implement automatic instantiation
9804 management. Code written for the Borland model will work fine, but
9805 each translation unit will contain instances of each of the templates it
9806 uses. In a large program, this can lead to an unacceptable amount of code
9810 @node Bound member functions
9811 @section Extracting the function pointer from a bound pointer to member function
9813 @cindex pointer to member function
9814 @cindex bound pointer to member function
9816 In C++, pointer to member functions (PMFs) are implemented using a wide
9817 pointer of sorts to handle all the possible call mechanisms; the PMF
9818 needs to store information about how to adjust the @samp{this} pointer,
9819 and if the function pointed to is virtual, where to find the vtable, and
9820 where in the vtable to look for the member function. If you are using
9821 PMFs in an inner loop, you should really reconsider that decision. If
9822 that is not an option, you can extract the pointer to the function that
9823 would be called for a given object/PMF pair and call it directly inside
9824 the inner loop, to save a bit of time.
9826 Note that you will still be paying the penalty for the call through a
9827 function pointer; on most modern architectures, such a call defeats the
9828 branch prediction features of the CPU@. This is also true of normal
9829 virtual function calls.
9831 The syntax for this extension is
9835 extern int (A::*fp)();
9836 typedef int (*fptr)(A *);
9838 fptr p = (fptr)(a.*fp);
9841 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
9842 no object is needed to obtain the address of the function. They can be
9843 converted to function pointers directly:
9846 fptr p1 = (fptr)(&A::foo);
9849 @opindex Wno-pmf-conversions
9850 You must specify @option{-Wno-pmf-conversions} to use this extension.
9852 @node C++ Attributes
9853 @section C++-Specific Variable, Function, and Type Attributes
9855 Some attributes only make sense for C++ programs.
9858 @item init_priority (@var{priority})
9859 @cindex init_priority attribute
9862 In Standard C++, objects defined at namespace scope are guaranteed to be
9863 initialized in an order in strict accordance with that of their definitions
9864 @emph{in a given translation unit}. No guarantee is made for initializations
9865 across translation units. However, GNU C++ allows users to control the
9866 order of initialization of objects defined at namespace scope with the
9867 @code{init_priority} attribute by specifying a relative @var{priority},
9868 a constant integral expression currently bounded between 101 and 65535
9869 inclusive. Lower numbers indicate a higher priority.
9871 In the following example, @code{A} would normally be created before
9872 @code{B}, but the @code{init_priority} attribute has reversed that order:
9875 Some_Class A __attribute__ ((init_priority (2000)));
9876 Some_Class B __attribute__ ((init_priority (543)));
9880 Note that the particular values of @var{priority} do not matter; only their
9883 @item java_interface
9884 @cindex java_interface attribute
9886 This type attribute informs C++ that the class is a Java interface. It may
9887 only be applied to classes declared within an @code{extern "Java"} block.
9888 Calls to methods declared in this interface will be dispatched using GCJ's
9889 interface table mechanism, instead of regular virtual table dispatch.
9893 See also @xref{Strong Using}.
9896 @section Strong Using
9898 @strong{Caution:} The semantics of this extension are not fully
9899 defined. Users should refrain from using this extension as its
9900 semantics may change subtly over time. It is possible that this
9901 extension wil be removed in future versions of G++.
9903 A using-directive with @code{__attribute ((strong))} is stronger
9904 than a normal using-directive in two ways:
9908 Templates from the used namespace can be specialized as though they were members of the using namespace.
9911 The using namespace is considered an associated namespace of all
9912 templates in the used namespace for purposes of argument-dependent
9916 This is useful for composing a namespace transparently from
9917 implementation namespaces. For example:
9922 template <class T> struct A @{ @};
9924 using namespace debug __attribute ((__strong__));
9925 template <> struct A<int> @{ @}; // @r{ok to specialize}
9927 template <class T> void f (A<T>);
9932 f (std::A<float>()); // @r{lookup finds} std::f
9937 @node Java Exceptions
9938 @section Java Exceptions
9940 The Java language uses a slightly different exception handling model
9941 from C++. Normally, GNU C++ will automatically detect when you are
9942 writing C++ code that uses Java exceptions, and handle them
9943 appropriately. However, if C++ code only needs to execute destructors
9944 when Java exceptions are thrown through it, GCC will guess incorrectly.
9945 Sample problematic code is:
9948 struct S @{ ~S(); @};
9949 extern void bar(); // @r{is written in Java, and may throw exceptions}
9958 The usual effect of an incorrect guess is a link failure, complaining of
9959 a missing routine called @samp{__gxx_personality_v0}.
9961 You can inform the compiler that Java exceptions are to be used in a
9962 translation unit, irrespective of what it might think, by writing
9963 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
9964 @samp{#pragma} must appear before any functions that throw or catch
9965 exceptions, or run destructors when exceptions are thrown through them.
9967 You cannot mix Java and C++ exceptions in the same translation unit. It
9968 is believed to be safe to throw a C++ exception from one file through
9969 another file compiled for the Java exception model, or vice versa, but
9970 there may be bugs in this area.
9972 @node Deprecated Features
9973 @section Deprecated Features
9975 In the past, the GNU C++ compiler was extended to experiment with new
9976 features, at a time when the C++ language was still evolving. Now that
9977 the C++ standard is complete, some of those features are superseded by
9978 superior alternatives. Using the old features might cause a warning in
9979 some cases that the feature will be dropped in the future. In other
9980 cases, the feature might be gone already.
9982 While the list below is not exhaustive, it documents some of the options
9983 that are now deprecated:
9986 @item -fexternal-templates
9987 @itemx -falt-external-templates
9988 These are two of the many ways for G++ to implement template
9989 instantiation. @xref{Template Instantiation}. The C++ standard clearly
9990 defines how template definitions have to be organized across
9991 implementation units. G++ has an implicit instantiation mechanism that
9992 should work just fine for standard-conforming code.
9994 @item -fstrict-prototype
9995 @itemx -fno-strict-prototype
9996 Previously it was possible to use an empty prototype parameter list to
9997 indicate an unspecified number of parameters (like C), rather than no
9998 parameters, as C++ demands. This feature has been removed, except where
9999 it is required for backwards compatibility @xref{Backwards Compatibility}.
10002 G++ allows a virtual function returning @samp{void *} to be overridden
10003 by one returning a different pointer type. This extension to the
10004 covariant return type rules is now deprecated and will be removed from a
10007 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
10008 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
10009 and will be removed in a future version. Code using these operators
10010 should be modified to use @code{std::min} and @code{std::max} instead.
10012 The named return value extension has been deprecated, and is now
10015 The use of initializer lists with new expressions has been deprecated,
10016 and is now removed from G++.
10018 Floating and complex non-type template parameters have been deprecated,
10019 and are now removed from G++.
10021 The implicit typename extension has been deprecated and is now
10024 The use of default arguments in function pointers, function typedefs and
10025 and other places where they are not permitted by the standard is
10026 deprecated and will be removed from a future version of G++.
10028 G++ allows floating-point literals to appear in integral constant expressions,
10029 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
10030 This extension is deprecated and will be removed from a future version.
10032 G++ allows static data members of const floating-point type to be declared
10033 with an initializer in a class definition. The standard only allows
10034 initializers for static members of const integral types and const
10035 enumeration types so this extension has been deprecated and will be removed
10036 from a future version.
10038 @node Backwards Compatibility
10039 @section Backwards Compatibility
10040 @cindex Backwards Compatibility
10041 @cindex ARM [Annotated C++ Reference Manual]
10043 Now that there is a definitive ISO standard C++, G++ has a specification
10044 to adhere to. The C++ language evolved over time, and features that
10045 used to be acceptable in previous drafts of the standard, such as the ARM
10046 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
10047 compilation of C++ written to such drafts, G++ contains some backwards
10048 compatibilities. @emph{All such backwards compatibility features are
10049 liable to disappear in future versions of G++.} They should be considered
10050 deprecated @xref{Deprecated Features}.
10054 If a variable is declared at for scope, it used to remain in scope until
10055 the end of the scope which contained the for statement (rather than just
10056 within the for scope). G++ retains this, but issues a warning, if such a
10057 variable is accessed outside the for scope.
10059 @item Implicit C language
10060 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
10061 scope to set the language. On such systems, all header files are
10062 implicitly scoped inside a C language scope. Also, an empty prototype
10063 @code{()} will be treated as an unspecified number of arguments, rather
10064 than no arguments, as C++ demands.