1 @c Copyright (C) 1988,1989,1992,1993,1994,1996,1998,1999,2000,2001,2002,2003,2004,2005
2 @c Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
7 @chapter Extensions to the C Language Family
8 @cindex extensions, C language
9 @cindex C language extensions
12 GNU C provides several language features not found in ISO standard C@.
13 (The @option{-pedantic} option directs GCC to print a warning message if
14 any of these features is used.) To test for the availability of these
15 features in conditional compilation, check for a predefined macro
16 @code{__GNUC__}, which is always defined under GCC@.
18 These extensions are available in C and Objective-C@. Most of them are
19 also available in C++. @xref{C++ Extensions,,Extensions to the
20 C++ Language}, for extensions that apply @emph{only} to C++.
22 Some features that are in ISO C99 but not C89 or C++ are also, as
23 extensions, accepted by GCC in C89 mode and in C++.
26 * Statement Exprs:: Putting statements and declarations inside expressions.
27 * Local Labels:: Labels local to a block.
28 * Labels as Values:: Getting pointers to labels, and computed gotos.
29 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
30 * Constructing Calls:: Dispatching a call to another function.
31 * Typeof:: @code{typeof}: referring to the type of an expression.
32 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
33 * Long Long:: Double-word integers---@code{long long int}.
34 * Complex:: Data types for complex numbers.
35 * Hex Floats:: Hexadecimal floating-point constants.
36 * Zero Length:: Zero-length arrays.
37 * Variable Length:: Arrays whose length is computed at run time.
38 * Empty Structures:: Structures with no members.
39 * Variadic Macros:: Macros with a variable number of arguments.
40 * Escaped Newlines:: Slightly looser rules for escaped newlines.
41 * Subscripting:: Any array can be subscripted, even if not an lvalue.
42 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
43 * Initializers:: Non-constant initializers.
44 * Compound Literals:: Compound literals give structures, unions
46 * Designated Inits:: Labeling elements of initializers.
47 * Cast to Union:: Casting to union type from any member of the union.
48 * Case Ranges:: `case 1 ... 9' and such.
49 * Mixed Declarations:: Mixing declarations and code.
50 * Function Attributes:: Declaring that functions have no side effects,
51 or that they can never return.
52 * Attribute Syntax:: Formal syntax for attributes.
53 * Function Prototypes:: Prototype declarations and old-style definitions.
54 * C++ Comments:: C++ comments are recognized.
55 * Dollar Signs:: Dollar sign is allowed in identifiers.
56 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
57 * Variable Attributes:: Specifying attributes of variables.
58 * Type Attributes:: Specifying attributes of types.
59 * Alignment:: Inquiring about the alignment of a type or variable.
60 * Inline:: Defining inline functions (as fast as macros).
61 * Extended Asm:: Assembler instructions with C expressions as operands.
62 (With them you can define ``built-in'' functions.)
63 * Constraints:: Constraints for asm operands
64 * Asm Labels:: Specifying the assembler name to use for a C symbol.
65 * Explicit Reg Vars:: Defining variables residing in specified registers.
66 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
67 * Incomplete Enums:: @code{enum foo;}, with details to follow.
68 * Function Names:: Printable strings which are the name of the current
70 * Return Address:: Getting the return or frame address of a function.
71 * Vector Extensions:: Using vector instructions through built-in functions.
72 * Offsetof:: Special syntax for implementing @code{offsetof}.
73 * Atomic Builtins:: Built-in functions for atomic memory access.
74 * Object Size Checking:: Built-in functions for limited buffer overflow
76 * Other Builtins:: Other built-in functions.
77 * Target Builtins:: Built-in functions specific to particular targets.
78 * Target Format Checks:: Format checks specific to particular targets.
79 * Pragmas:: Pragmas accepted by GCC.
80 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
81 * Thread-Local:: Per-thread variables.
85 @section Statements and Declarations in Expressions
86 @cindex statements inside expressions
87 @cindex declarations inside expressions
88 @cindex expressions containing statements
89 @cindex macros, statements in expressions
91 @c the above section title wrapped and causes an underfull hbox.. i
92 @c changed it from "within" to "in". --mew 4feb93
93 A compound statement enclosed in parentheses may appear as an expression
94 in GNU C@. This allows you to use loops, switches, and local variables
97 Recall that a compound statement is a sequence of statements surrounded
98 by braces; in this construct, parentheses go around the braces. For
102 (@{ int y = foo (); int z;
109 is a valid (though slightly more complex than necessary) expression
110 for the absolute value of @code{foo ()}.
112 The last thing in the compound statement should be an expression
113 followed by a semicolon; the value of this subexpression serves as the
114 value of the entire construct. (If you use some other kind of statement
115 last within the braces, the construct has type @code{void}, and thus
116 effectively no value.)
118 This feature is especially useful in making macro definitions ``safe'' (so
119 that they evaluate each operand exactly once). For example, the
120 ``maximum'' function is commonly defined as a macro in standard C as
124 #define max(a,b) ((a) > (b) ? (a) : (b))
128 @cindex side effects, macro argument
129 But this definition computes either @var{a} or @var{b} twice, with bad
130 results if the operand has side effects. In GNU C, if you know the
131 type of the operands (here taken as @code{int}), you can define
132 the macro safely as follows:
135 #define maxint(a,b) \
136 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
139 Embedded statements are not allowed in constant expressions, such as
140 the value of an enumeration constant, the width of a bit-field, or
141 the initial value of a static variable.
143 If you don't know the type of the operand, you can still do this, but you
144 must use @code{typeof} (@pxref{Typeof}).
146 In G++, the result value of a statement expression undergoes array and
147 function pointer decay, and is returned by value to the enclosing
148 expression. For instance, if @code{A} is a class, then
157 will construct a temporary @code{A} object to hold the result of the
158 statement expression, and that will be used to invoke @code{Foo}.
159 Therefore the @code{this} pointer observed by @code{Foo} will not be the
162 Any temporaries created within a statement within a statement expression
163 will be destroyed at the statement's end. This makes statement
164 expressions inside macros slightly different from function calls. In
165 the latter case temporaries introduced during argument evaluation will
166 be destroyed at the end of the statement that includes the function
167 call. In the statement expression case they will be destroyed during
168 the statement expression. For instance,
171 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
172 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
182 will have different places where temporaries are destroyed. For the
183 @code{macro} case, the temporary @code{X} will be destroyed just after
184 the initialization of @code{b}. In the @code{function} case that
185 temporary will be destroyed when the function returns.
187 These considerations mean that it is probably a bad idea to use
188 statement-expressions of this form in header files that are designed to
189 work with C++. (Note that some versions of the GNU C Library contained
190 header files using statement-expression that lead to precisely this
193 Jumping into a statement expression with @code{goto} or using a
194 @code{switch} statement outside the statement expression with a
195 @code{case} or @code{default} label inside the statement expression is
196 not permitted. Jumping into a statement expression with a computed
197 @code{goto} (@pxref{Labels as Values}) yields undefined behavior.
198 Jumping out of a statement expression is permitted, but if the
199 statement expression is part of a larger expression then it is
200 unspecified which other subexpressions of that expression have been
201 evaluated except where the language definition requires certain
202 subexpressions to be evaluated before or after the statement
203 expression. In any case, as with a function call the evaluation of a
204 statement expression is not interleaved with the evaluation of other
205 parts of the containing expression. For example,
208 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
212 will call @code{foo} and @code{bar1} and will not call @code{baz} but
213 may or may not call @code{bar2}. If @code{bar2} is called, it will be
214 called after @code{foo} and before @code{bar1}
217 @section Locally Declared Labels
219 @cindex macros, local labels
221 GCC allows you to declare @dfn{local labels} in any nested block
222 scope. A local label is just like an ordinary label, but you can
223 only reference it (with a @code{goto} statement, or by taking its
224 address) within the block in which it was declared.
226 A local label declaration looks like this:
229 __label__ @var{label};
236 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
239 Local label declarations must come at the beginning of the block,
240 before any ordinary declarations or statements.
242 The label declaration defines the label @emph{name}, but does not define
243 the label itself. You must do this in the usual way, with
244 @code{@var{label}:}, within the statements of the statement expression.
246 The local label feature is useful for complex macros. If a macro
247 contains nested loops, a @code{goto} can be useful for breaking out of
248 them. However, an ordinary label whose scope is the whole function
249 cannot be used: if the macro can be expanded several times in one
250 function, the label will be multiply defined in that function. A
251 local label avoids this problem. For example:
254 #define SEARCH(value, array, target) \
257 typeof (target) _SEARCH_target = (target); \
258 typeof (*(array)) *_SEARCH_array = (array); \
261 for (i = 0; i < max; i++) \
262 for (j = 0; j < max; j++) \
263 if (_SEARCH_array[i][j] == _SEARCH_target) \
264 @{ (value) = i; goto found; @} \
270 This could also be written using a statement-expression:
273 #define SEARCH(array, target) \
276 typeof (target) _SEARCH_target = (target); \
277 typeof (*(array)) *_SEARCH_array = (array); \
280 for (i = 0; i < max; i++) \
281 for (j = 0; j < max; j++) \
282 if (_SEARCH_array[i][j] == _SEARCH_target) \
283 @{ value = i; goto found; @} \
290 Local label declarations also make the labels they declare visible to
291 nested functions, if there are any. @xref{Nested Functions}, for details.
293 @node Labels as Values
294 @section Labels as Values
295 @cindex labels as values
296 @cindex computed gotos
297 @cindex goto with computed label
298 @cindex address of a label
300 You can get the address of a label defined in the current function
301 (or a containing function) with the unary operator @samp{&&}. The
302 value has type @code{void *}. This value is a constant and can be used
303 wherever a constant of that type is valid. For example:
311 To use these values, you need to be able to jump to one. This is done
312 with the computed goto statement@footnote{The analogous feature in
313 Fortran is called an assigned goto, but that name seems inappropriate in
314 C, where one can do more than simply store label addresses in label
315 variables.}, @code{goto *@var{exp};}. For example,
322 Any expression of type @code{void *} is allowed.
324 One way of using these constants is in initializing a static array that
325 will serve as a jump table:
328 static void *array[] = @{ &&foo, &&bar, &&hack @};
331 Then you can select a label with indexing, like this:
338 Note that this does not check whether the subscript is in bounds---array
339 indexing in C never does that.
341 Such an array of label values serves a purpose much like that of the
342 @code{switch} statement. The @code{switch} statement is cleaner, so
343 use that rather than an array unless the problem does not fit a
344 @code{switch} statement very well.
346 Another use of label values is in an interpreter for threaded code.
347 The labels within the interpreter function can be stored in the
348 threaded code for super-fast dispatching.
350 You may not use this mechanism to jump to code in a different function.
351 If you do that, totally unpredictable things will happen. The best way to
352 avoid this is to store the label address only in automatic variables and
353 never pass it as an argument.
355 An alternate way to write the above example is
358 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
360 goto *(&&foo + array[i]);
364 This is more friendly to code living in shared libraries, as it reduces
365 the number of dynamic relocations that are needed, and by consequence,
366 allows the data to be read-only.
368 @node Nested Functions
369 @section Nested Functions
370 @cindex nested functions
371 @cindex downward funargs
374 A @dfn{nested function} is a function defined inside another function.
375 (Nested functions are not supported for GNU C++.) The nested function's
376 name is local to the block where it is defined. For example, here we
377 define a nested function named @code{square}, and call it twice:
381 foo (double a, double b)
383 double square (double z) @{ return z * z; @}
385 return square (a) + square (b);
390 The nested function can access all the variables of the containing
391 function that are visible at the point of its definition. This is
392 called @dfn{lexical scoping}. For example, here we show a nested
393 function which uses an inherited variable named @code{offset}:
397 bar (int *array, int offset, int size)
399 int access (int *array, int index)
400 @{ return array[index + offset]; @}
403 for (i = 0; i < size; i++)
404 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
409 Nested function definitions are permitted within functions in the places
410 where variable definitions are allowed; that is, in any block, mixed
411 with the other declarations and statements in the block.
413 It is possible to call the nested function from outside the scope of its
414 name by storing its address or passing the address to another function:
417 hack (int *array, int size)
419 void store (int index, int value)
420 @{ array[index] = value; @}
422 intermediate (store, size);
426 Here, the function @code{intermediate} receives the address of
427 @code{store} as an argument. If @code{intermediate} calls @code{store},
428 the arguments given to @code{store} are used to store into @code{array}.
429 But this technique works only so long as the containing function
430 (@code{hack}, in this example) does not exit.
432 If you try to call the nested function through its address after the
433 containing function has exited, all hell will break loose. If you try
434 to call it after a containing scope level has exited, and if it refers
435 to some of the variables that are no longer in scope, you may be lucky,
436 but it's not wise to take the risk. If, however, the nested function
437 does not refer to anything that has gone out of scope, you should be
440 GCC implements taking the address of a nested function using a technique
441 called @dfn{trampolines}. A paper describing them is available as
444 @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}.
446 A nested function can jump to a label inherited from a containing
447 function, provided the label was explicitly declared in the containing
448 function (@pxref{Local Labels}). Such a jump returns instantly to the
449 containing function, exiting the nested function which did the
450 @code{goto} and any intermediate functions as well. Here is an example:
454 bar (int *array, int offset, int size)
457 int access (int *array, int index)
461 return array[index + offset];
465 for (i = 0; i < size; i++)
466 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
470 /* @r{Control comes here from @code{access}
471 if it detects an error.} */
478 A nested function always has no linkage. Declaring one with
479 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
480 before its definition, use @code{auto} (which is otherwise meaningless
481 for function declarations).
484 bar (int *array, int offset, int size)
487 auto int access (int *, int);
489 int access (int *array, int index)
493 return array[index + offset];
499 @node Constructing Calls
500 @section Constructing Function Calls
501 @cindex constructing calls
502 @cindex forwarding calls
504 Using the built-in functions described below, you can record
505 the arguments a function received, and call another function
506 with the same arguments, without knowing the number or types
509 You can also record the return value of that function call,
510 and later return that value, without knowing what data type
511 the function tried to return (as long as your caller expects
514 However, these built-in functions may interact badly with some
515 sophisticated features or other extensions of the language. It
516 is, therefore, not recommended to use them outside very simple
517 functions acting as mere forwarders for their arguments.
519 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
520 This built-in function returns a pointer to data
521 describing how to perform a call with the same arguments as were passed
522 to the current function.
524 The function saves the arg pointer register, structure value address,
525 and all registers that might be used to pass arguments to a function
526 into a block of memory allocated on the stack. Then it returns the
527 address of that block.
530 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
531 This built-in function invokes @var{function}
532 with a copy of the parameters described by @var{arguments}
535 The value of @var{arguments} should be the value returned by
536 @code{__builtin_apply_args}. The argument @var{size} specifies the size
537 of the stack argument data, in bytes.
539 This function returns a pointer to data describing
540 how to return whatever value was returned by @var{function}. The data
541 is saved in a block of memory allocated on the stack.
543 It is not always simple to compute the proper value for @var{size}. The
544 value is used by @code{__builtin_apply} to compute the amount of data
545 that should be pushed on the stack and copied from the incoming argument
549 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
550 This built-in function returns the value described by @var{result} from
551 the containing function. You should specify, for @var{result}, a value
552 returned by @code{__builtin_apply}.
556 @section Referring to a Type with @code{typeof}
559 @cindex macros, types of arguments
561 Another way to refer to the type of an expression is with @code{typeof}.
562 The syntax of using of this keyword looks like @code{sizeof}, but the
563 construct acts semantically like a type name defined with @code{typedef}.
565 There are two ways of writing the argument to @code{typeof}: with an
566 expression or with a type. Here is an example with an expression:
573 This assumes that @code{x} is an array of pointers to functions;
574 the type described is that of the values of the functions.
576 Here is an example with a typename as the argument:
583 Here the type described is that of pointers to @code{int}.
585 If you are writing a header file that must work when included in ISO C
586 programs, write @code{__typeof__} instead of @code{typeof}.
587 @xref{Alternate Keywords}.
589 A @code{typeof}-construct can be used anywhere a typedef name could be
590 used. For example, you can use it in a declaration, in a cast, or inside
591 of @code{sizeof} or @code{typeof}.
593 @code{typeof} is often useful in conjunction with the
594 statements-within-expressions feature. Here is how the two together can
595 be used to define a safe ``maximum'' macro that operates on any
596 arithmetic type and evaluates each of its arguments exactly once:
600 (@{ typeof (a) _a = (a); \
601 typeof (b) _b = (b); \
602 _a > _b ? _a : _b; @})
605 @cindex underscores in variables in macros
606 @cindex @samp{_} in variables in macros
607 @cindex local variables in macros
608 @cindex variables, local, in macros
609 @cindex macros, local variables in
611 The reason for using names that start with underscores for the local
612 variables is to avoid conflicts with variable names that occur within the
613 expressions that are substituted for @code{a} and @code{b}. Eventually we
614 hope to design a new form of declaration syntax that allows you to declare
615 variables whose scopes start only after their initializers; this will be a
616 more reliable way to prevent such conflicts.
619 Some more examples of the use of @code{typeof}:
623 This declares @code{y} with the type of what @code{x} points to.
630 This declares @code{y} as an array of such values.
637 This declares @code{y} as an array of pointers to characters:
640 typeof (typeof (char *)[4]) y;
644 It is equivalent to the following traditional C declaration:
650 To see the meaning of the declaration using @code{typeof}, and why it
651 might be a useful way to write, rewrite it with these macros:
654 #define pointer(T) typeof(T *)
655 #define array(T, N) typeof(T [N])
659 Now the declaration can be rewritten this way:
662 array (pointer (char), 4) y;
666 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
667 pointers to @code{char}.
670 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
671 a more limited extension which permitted one to write
674 typedef @var{T} = @var{expr};
678 with the effect of declaring @var{T} to have the type of the expression
679 @var{expr}. This extension does not work with GCC 3 (versions between
680 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
681 relies on it should be rewritten to use @code{typeof}:
684 typedef typeof(@var{expr}) @var{T};
688 This will work with all versions of GCC@.
691 @section Conditionals with Omitted Operands
692 @cindex conditional expressions, extensions
693 @cindex omitted middle-operands
694 @cindex middle-operands, omitted
695 @cindex extensions, @code{?:}
696 @cindex @code{?:} extensions
698 The middle operand in a conditional expression may be omitted. Then
699 if the first operand is nonzero, its value is the value of the conditional
702 Therefore, the expression
709 has the value of @code{x} if that is nonzero; otherwise, the value of
712 This example is perfectly equivalent to
718 @cindex side effect in ?:
719 @cindex ?: side effect
721 In this simple case, the ability to omit the middle operand is not
722 especially useful. When it becomes useful is when the first operand does,
723 or may (if it is a macro argument), contain a side effect. Then repeating
724 the operand in the middle would perform the side effect twice. Omitting
725 the middle operand uses the value already computed without the undesirable
726 effects of recomputing it.
729 @section Double-Word Integers
730 @cindex @code{long long} data types
731 @cindex double-word arithmetic
732 @cindex multiprecision arithmetic
733 @cindex @code{LL} integer suffix
734 @cindex @code{ULL} integer suffix
736 ISO C99 supports data types for integers that are at least 64 bits wide,
737 and as an extension GCC supports them in C89 mode and in C++.
738 Simply write @code{long long int} for a signed integer, or
739 @code{unsigned long long int} for an unsigned integer. To make an
740 integer constant of type @code{long long int}, add the suffix @samp{LL}
741 to the integer. To make an integer constant of type @code{unsigned long
742 long int}, add the suffix @samp{ULL} to the integer.
744 You can use these types in arithmetic like any other integer types.
745 Addition, subtraction, and bitwise boolean operations on these types
746 are open-coded on all types of machines. Multiplication is open-coded
747 if the machine supports fullword-to-doubleword a widening multiply
748 instruction. Division and shifts are open-coded only on machines that
749 provide special support. The operations that are not open-coded use
750 special library routines that come with GCC@.
752 There may be pitfalls when you use @code{long long} types for function
753 arguments, unless you declare function prototypes. If a function
754 expects type @code{int} for its argument, and you pass a value of type
755 @code{long long int}, confusion will result because the caller and the
756 subroutine will disagree about the number of bytes for the argument.
757 Likewise, if the function expects @code{long long int} and you pass
758 @code{int}. The best way to avoid such problems is to use prototypes.
761 @section Complex Numbers
762 @cindex complex numbers
763 @cindex @code{_Complex} keyword
764 @cindex @code{__complex__} keyword
766 ISO C99 supports complex floating data types, and as an extension GCC
767 supports them in C89 mode and in C++, and supports complex integer data
768 types which are not part of ISO C99. You can declare complex types
769 using the keyword @code{_Complex}. As an extension, the older GNU
770 keyword @code{__complex__} is also supported.
772 For example, @samp{_Complex double x;} declares @code{x} as a
773 variable whose real part and imaginary part are both of type
774 @code{double}. @samp{_Complex short int y;} declares @code{y} to
775 have real and imaginary parts of type @code{short int}; this is not
776 likely to be useful, but it shows that the set of complex types is
779 To write a constant with a complex data type, use the suffix @samp{i} or
780 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
781 has type @code{_Complex float} and @code{3i} has type
782 @code{_Complex int}. Such a constant always has a pure imaginary
783 value, but you can form any complex value you like by adding one to a
784 real constant. This is a GNU extension; if you have an ISO C99
785 conforming C library (such as GNU libc), and want to construct complex
786 constants of floating type, you should include @code{<complex.h>} and
787 use the macros @code{I} or @code{_Complex_I} instead.
789 @cindex @code{__real__} keyword
790 @cindex @code{__imag__} keyword
791 To extract the real part of a complex-valued expression @var{exp}, write
792 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
793 extract the imaginary part. This is a GNU extension; for values of
794 floating type, you should use the ISO C99 functions @code{crealf},
795 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
796 @code{cimagl}, declared in @code{<complex.h>} and also provided as
797 built-in functions by GCC@.
799 @cindex complex conjugation
800 The operator @samp{~} performs complex conjugation when used on a value
801 with a complex type. This is a GNU extension; for values of
802 floating type, you should use the ISO C99 functions @code{conjf},
803 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
804 provided as built-in functions by GCC@.
806 GCC can allocate complex automatic variables in a noncontiguous
807 fashion; it's even possible for the real part to be in a register while
808 the imaginary part is on the stack (or vice-versa). Only the DWARF2
809 debug info format can represent this, so use of DWARF2 is recommended.
810 If you are using the stabs debug info format, GCC describes a noncontiguous
811 complex variable as if it were two separate variables of noncomplex type.
812 If the variable's actual name is @code{foo}, the two fictitious
813 variables are named @code{foo$real} and @code{foo$imag}. You can
814 examine and set these two fictitious variables with your debugger.
820 ISO C99 supports floating-point numbers written not only in the usual
821 decimal notation, such as @code{1.55e1}, but also numbers such as
822 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
823 supports this in C89 mode (except in some cases when strictly
824 conforming) and in C++. In that format the
825 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
826 mandatory. The exponent is a decimal number that indicates the power of
827 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
834 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
835 is the same as @code{1.55e1}.
837 Unlike for floating-point numbers in the decimal notation the exponent
838 is always required in the hexadecimal notation. Otherwise the compiler
839 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
840 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
841 extension for floating-point constants of type @code{float}.
844 @section Arrays of Length Zero
845 @cindex arrays of length zero
846 @cindex zero-length arrays
847 @cindex length-zero arrays
848 @cindex flexible array members
850 Zero-length arrays are allowed in GNU C@. They are very useful as the
851 last element of a structure which is really a header for a variable-length
860 struct line *thisline = (struct line *)
861 malloc (sizeof (struct line) + this_length);
862 thisline->length = this_length;
865 In ISO C90, you would have to give @code{contents} a length of 1, which
866 means either you waste space or complicate the argument to @code{malloc}.
868 In ISO C99, you would use a @dfn{flexible array member}, which is
869 slightly different in syntax and semantics:
873 Flexible array members are written as @code{contents[]} without
877 Flexible array members have incomplete type, and so the @code{sizeof}
878 operator may not be applied. As a quirk of the original implementation
879 of zero-length arrays, @code{sizeof} evaluates to zero.
882 Flexible array members may only appear as the last member of a
883 @code{struct} that is otherwise non-empty.
886 A structure containing a flexible array member, or a union containing
887 such a structure (possibly recursively), may not be a member of a
888 structure or an element of an array. (However, these uses are
889 permitted by GCC as extensions.)
892 GCC versions before 3.0 allowed zero-length arrays to be statically
893 initialized, as if they were flexible arrays. In addition to those
894 cases that were useful, it also allowed initializations in situations
895 that would corrupt later data. Non-empty initialization of zero-length
896 arrays is now treated like any case where there are more initializer
897 elements than the array holds, in that a suitable warning about "excess
898 elements in array" is given, and the excess elements (all of them, in
899 this case) are ignored.
901 Instead GCC allows static initialization of flexible array members.
902 This is equivalent to defining a new structure containing the original
903 structure followed by an array of sufficient size to contain the data.
904 I.e.@: in the following, @code{f1} is constructed as if it were declared
910 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
913 struct f1 f1; int data[3];
914 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
918 The convenience of this extension is that @code{f1} has the desired
919 type, eliminating the need to consistently refer to @code{f2.f1}.
921 This has symmetry with normal static arrays, in that an array of
922 unknown size is also written with @code{[]}.
924 Of course, this extension only makes sense if the extra data comes at
925 the end of a top-level object, as otherwise we would be overwriting
926 data at subsequent offsets. To avoid undue complication and confusion
927 with initialization of deeply nested arrays, we simply disallow any
928 non-empty initialization except when the structure is the top-level
932 struct foo @{ int x; int y[]; @};
933 struct bar @{ struct foo z; @};
935 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
936 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
937 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
938 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
941 @node Empty Structures
942 @section Structures With No Members
943 @cindex empty structures
944 @cindex zero-size structures
946 GCC permits a C structure to have no members:
953 The structure will have size zero. In C++, empty structures are part
954 of the language. G++ treats empty structures as if they had a single
955 member of type @code{char}.
957 @node Variable Length
958 @section Arrays of Variable Length
959 @cindex variable-length arrays
960 @cindex arrays of variable length
963 Variable-length automatic arrays are allowed in ISO C99, and as an
964 extension GCC accepts them in C89 mode and in C++. (However, GCC's
965 implementation of variable-length arrays does not yet conform in detail
966 to the ISO C99 standard.) These arrays are
967 declared like any other automatic arrays, but with a length that is not
968 a constant expression. The storage is allocated at the point of
969 declaration and deallocated when the brace-level is exited. For
974 concat_fopen (char *s1, char *s2, char *mode)
976 char str[strlen (s1) + strlen (s2) + 1];
979 return fopen (str, mode);
983 @cindex scope of a variable length array
984 @cindex variable-length array scope
985 @cindex deallocating variable length arrays
986 Jumping or breaking out of the scope of the array name deallocates the
987 storage. Jumping into the scope is not allowed; you get an error
990 @cindex @code{alloca} vs variable-length arrays
991 You can use the function @code{alloca} to get an effect much like
992 variable-length arrays. The function @code{alloca} is available in
993 many other C implementations (but not in all). On the other hand,
994 variable-length arrays are more elegant.
996 There are other differences between these two methods. Space allocated
997 with @code{alloca} exists until the containing @emph{function} returns.
998 The space for a variable-length array is deallocated as soon as the array
999 name's scope ends. (If you use both variable-length arrays and
1000 @code{alloca} in the same function, deallocation of a variable-length array
1001 will also deallocate anything more recently allocated with @code{alloca}.)
1003 You can also use variable-length arrays as arguments to functions:
1007 tester (int len, char data[len][len])
1013 The length of an array is computed once when the storage is allocated
1014 and is remembered for the scope of the array in case you access it with
1017 If you want to pass the array first and the length afterward, you can
1018 use a forward declaration in the parameter list---another GNU extension.
1022 tester (int len; char data[len][len], int len)
1028 @cindex parameter forward declaration
1029 The @samp{int len} before the semicolon is a @dfn{parameter forward
1030 declaration}, and it serves the purpose of making the name @code{len}
1031 known when the declaration of @code{data} is parsed.
1033 You can write any number of such parameter forward declarations in the
1034 parameter list. They can be separated by commas or semicolons, but the
1035 last one must end with a semicolon, which is followed by the ``real''
1036 parameter declarations. Each forward declaration must match a ``real''
1037 declaration in parameter name and data type. ISO C99 does not support
1038 parameter forward declarations.
1040 @node Variadic Macros
1041 @section Macros with a Variable Number of Arguments.
1042 @cindex variable number of arguments
1043 @cindex macro with variable arguments
1044 @cindex rest argument (in macro)
1045 @cindex variadic macros
1047 In the ISO C standard of 1999, a macro can be declared to accept a
1048 variable number of arguments much as a function can. The syntax for
1049 defining the macro is similar to that of a function. Here is an
1053 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1056 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1057 such a macro, it represents the zero or more tokens until the closing
1058 parenthesis that ends the invocation, including any commas. This set of
1059 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1060 wherever it appears. See the CPP manual for more information.
1062 GCC has long supported variadic macros, and used a different syntax that
1063 allowed you to give a name to the variable arguments just like any other
1064 argument. Here is an example:
1067 #define debug(format, args...) fprintf (stderr, format, args)
1070 This is in all ways equivalent to the ISO C example above, but arguably
1071 more readable and descriptive.
1073 GNU CPP has two further variadic macro extensions, and permits them to
1074 be used with either of the above forms of macro definition.
1076 In standard C, you are not allowed to leave the variable argument out
1077 entirely; but you are allowed to pass an empty argument. For example,
1078 this invocation is invalid in ISO C, because there is no comma after
1085 GNU CPP permits you to completely omit the variable arguments in this
1086 way. In the above examples, the compiler would complain, though since
1087 the expansion of the macro still has the extra comma after the format
1090 To help solve this problem, CPP behaves specially for variable arguments
1091 used with the token paste operator, @samp{##}. If instead you write
1094 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1097 and if the variable arguments are omitted or empty, the @samp{##}
1098 operator causes the preprocessor to remove the comma before it. If you
1099 do provide some variable arguments in your macro invocation, GNU CPP
1100 does not complain about the paste operation and instead places the
1101 variable arguments after the comma. Just like any other pasted macro
1102 argument, these arguments are not macro expanded.
1104 @node Escaped Newlines
1105 @section Slightly Looser Rules for Escaped Newlines
1106 @cindex escaped newlines
1107 @cindex newlines (escaped)
1109 Recently, the preprocessor has relaxed its treatment of escaped
1110 newlines. Previously, the newline had to immediately follow a
1111 backslash. The current implementation allows whitespace in the form
1112 of spaces, horizontal and vertical tabs, and form feeds between the
1113 backslash and the subsequent newline. The preprocessor issues a
1114 warning, but treats it as a valid escaped newline and combines the two
1115 lines to form a single logical line. This works within comments and
1116 tokens, as well as between tokens. Comments are @emph{not} treated as
1117 whitespace for the purposes of this relaxation, since they have not
1118 yet been replaced with spaces.
1121 @section Non-Lvalue Arrays May Have Subscripts
1122 @cindex subscripting
1123 @cindex arrays, non-lvalue
1125 @cindex subscripting and function values
1126 In ISO C99, arrays that are not lvalues still decay to pointers, and
1127 may be subscripted, although they may not be modified or used after
1128 the next sequence point and the unary @samp{&} operator may not be
1129 applied to them. As an extension, GCC allows such arrays to be
1130 subscripted in C89 mode, though otherwise they do not decay to
1131 pointers outside C99 mode. For example,
1132 this is valid in GNU C though not valid in C89:
1136 struct foo @{int a[4];@};
1142 return f().a[index];
1148 @section Arithmetic on @code{void}- and Function-Pointers
1149 @cindex void pointers, arithmetic
1150 @cindex void, size of pointer to
1151 @cindex function pointers, arithmetic
1152 @cindex function, size of pointer to
1154 In GNU C, addition and subtraction operations are supported on pointers to
1155 @code{void} and on pointers to functions. This is done by treating the
1156 size of a @code{void} or of a function as 1.
1158 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1159 and on function types, and returns 1.
1161 @opindex Wpointer-arith
1162 The option @option{-Wpointer-arith} requests a warning if these extensions
1166 @section Non-Constant Initializers
1167 @cindex initializers, non-constant
1168 @cindex non-constant initializers
1170 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1171 automatic variable are not required to be constant expressions in GNU C@.
1172 Here is an example of an initializer with run-time varying elements:
1175 foo (float f, float g)
1177 float beat_freqs[2] = @{ f-g, f+g @};
1182 @node Compound Literals
1183 @section Compound Literals
1184 @cindex constructor expressions
1185 @cindex initializations in expressions
1186 @cindex structures, constructor expression
1187 @cindex expressions, constructor
1188 @cindex compound literals
1189 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1191 ISO C99 supports compound literals. A compound literal looks like
1192 a cast containing an initializer. Its value is an object of the
1193 type specified in the cast, containing the elements specified in
1194 the initializer; it is an lvalue. As an extension, GCC supports
1195 compound literals in C89 mode and in C++.
1197 Usually, the specified type is a structure. Assume that
1198 @code{struct foo} and @code{structure} are declared as shown:
1201 struct foo @{int a; char b[2];@} structure;
1205 Here is an example of constructing a @code{struct foo} with a compound literal:
1208 structure = ((struct foo) @{x + y, 'a', 0@});
1212 This is equivalent to writing the following:
1216 struct foo temp = @{x + y, 'a', 0@};
1221 You can also construct an array. If all the elements of the compound literal
1222 are (made up of) simple constant expressions, suitable for use in
1223 initializers of objects of static storage duration, then the compound
1224 literal can be coerced to a pointer to its first element and used in
1225 such an initializer, as shown here:
1228 char **foo = (char *[]) @{ "x", "y", "z" @};
1231 Compound literals for scalar types and union types are is
1232 also allowed, but then the compound literal is equivalent
1235 As a GNU extension, GCC allows initialization of objects with static storage
1236 duration by compound literals (which is not possible in ISO C99, because
1237 the initializer is not a constant).
1238 It is handled as if the object was initialized only with the bracket
1239 enclosed list if compound literal's and object types match.
1240 The initializer list of the compound literal must be constant.
1241 If the object being initialized has array type of unknown size, the size is
1242 determined by compound literal size.
1245 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1246 static int y[] = (int []) @{1, 2, 3@};
1247 static int z[] = (int [3]) @{1@};
1251 The above lines are equivalent to the following:
1253 static struct foo x = @{1, 'a', 'b'@};
1254 static int y[] = @{1, 2, 3@};
1255 static int z[] = @{1, 0, 0@};
1258 @node Designated Inits
1259 @section Designated Initializers
1260 @cindex initializers with labeled elements
1261 @cindex labeled elements in initializers
1262 @cindex case labels in initializers
1263 @cindex designated initializers
1265 Standard C89 requires the elements of an initializer to appear in a fixed
1266 order, the same as the order of the elements in the array or structure
1269 In ISO C99 you can give the elements in any order, specifying the array
1270 indices or structure field names they apply to, and GNU C allows this as
1271 an extension in C89 mode as well. This extension is not
1272 implemented in GNU C++.
1274 To specify an array index, write
1275 @samp{[@var{index}] =} before the element value. For example,
1278 int a[6] = @{ [4] = 29, [2] = 15 @};
1285 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1289 The index values must be constant expressions, even if the array being
1290 initialized is automatic.
1292 An alternative syntax for this which has been obsolete since GCC 2.5 but
1293 GCC still accepts is to write @samp{[@var{index}]} before the element
1294 value, with no @samp{=}.
1296 To initialize a range of elements to the same value, write
1297 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1298 extension. For example,
1301 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1305 If the value in it has side-effects, the side-effects will happen only once,
1306 not for each initialized field by the range initializer.
1309 Note that the length of the array is the highest value specified
1312 In a structure initializer, specify the name of a field to initialize
1313 with @samp{.@var{fieldname} =} before the element value. For example,
1314 given the following structure,
1317 struct point @{ int x, y; @};
1321 the following initialization
1324 struct point p = @{ .y = yvalue, .x = xvalue @};
1331 struct point p = @{ xvalue, yvalue @};
1334 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1335 @samp{@var{fieldname}:}, as shown here:
1338 struct point p = @{ y: yvalue, x: xvalue @};
1342 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1343 @dfn{designator}. You can also use a designator (or the obsolete colon
1344 syntax) when initializing a union, to specify which element of the union
1345 should be used. For example,
1348 union foo @{ int i; double d; @};
1350 union foo f = @{ .d = 4 @};
1354 will convert 4 to a @code{double} to store it in the union using
1355 the second element. By contrast, casting 4 to type @code{union foo}
1356 would store it into the union as the integer @code{i}, since it is
1357 an integer. (@xref{Cast to Union}.)
1359 You can combine this technique of naming elements with ordinary C
1360 initialization of successive elements. Each initializer element that
1361 does not have a designator applies to the next consecutive element of the
1362 array or structure. For example,
1365 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1372 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1375 Labeling the elements of an array initializer is especially useful
1376 when the indices are characters or belong to an @code{enum} type.
1381 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1382 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1385 @cindex designator lists
1386 You can also write a series of @samp{.@var{fieldname}} and
1387 @samp{[@var{index}]} designators before an @samp{=} to specify a
1388 nested subobject to initialize; the list is taken relative to the
1389 subobject corresponding to the closest surrounding brace pair. For
1390 example, with the @samp{struct point} declaration above:
1393 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1397 If the same field is initialized multiple times, it will have value from
1398 the last initialization. If any such overridden initialization has
1399 side-effect, it is unspecified whether the side-effect happens or not.
1400 Currently, GCC will discard them and issue a warning.
1403 @section Case Ranges
1405 @cindex ranges in case statements
1407 You can specify a range of consecutive values in a single @code{case} label,
1411 case @var{low} ... @var{high}:
1415 This has the same effect as the proper number of individual @code{case}
1416 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1418 This feature is especially useful for ranges of ASCII character codes:
1424 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1425 it may be parsed wrong when you use it with integer values. For example,
1440 @section Cast to a Union Type
1441 @cindex cast to a union
1442 @cindex union, casting to a
1444 A cast to union type is similar to other casts, except that the type
1445 specified is a union type. You can specify the type either with
1446 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1447 a constructor though, not a cast, and hence does not yield an lvalue like
1448 normal casts. (@xref{Compound Literals}.)
1450 The types that may be cast to the union type are those of the members
1451 of the union. Thus, given the following union and variables:
1454 union foo @{ int i; double d; @};
1460 both @code{x} and @code{y} can be cast to type @code{union foo}.
1462 Using the cast as the right-hand side of an assignment to a variable of
1463 union type is equivalent to storing in a member of the union:
1468 u = (union foo) x @equiv{} u.i = x
1469 u = (union foo) y @equiv{} u.d = y
1472 You can also use the union cast as a function argument:
1475 void hack (union foo);
1477 hack ((union foo) x);
1480 @node Mixed Declarations
1481 @section Mixed Declarations and Code
1482 @cindex mixed declarations and code
1483 @cindex declarations, mixed with code
1484 @cindex code, mixed with declarations
1486 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1487 within compound statements. As an extension, GCC also allows this in
1488 C89 mode. For example, you could do:
1497 Each identifier is visible from where it is declared until the end of
1498 the enclosing block.
1500 @node Function Attributes
1501 @section Declaring Attributes of Functions
1502 @cindex function attributes
1503 @cindex declaring attributes of functions
1504 @cindex functions that never return
1505 @cindex functions that return more than once
1506 @cindex functions that have no side effects
1507 @cindex functions in arbitrary sections
1508 @cindex functions that behave like malloc
1509 @cindex @code{volatile} applied to function
1510 @cindex @code{const} applied to function
1511 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1512 @cindex functions with non-null pointer arguments
1513 @cindex functions that are passed arguments in registers on the 386
1514 @cindex functions that pop the argument stack on the 386
1515 @cindex functions that do not pop the argument stack on the 386
1517 In GNU C, you declare certain things about functions called in your program
1518 which help the compiler optimize function calls and check your code more
1521 The keyword @code{__attribute__} allows you to specify special
1522 attributes when making a declaration. This keyword is followed by an
1523 attribute specification inside double parentheses. The following
1524 attributes are currently defined for functions on all targets:
1525 @code{noreturn}, @code{returns_twice}, @code{noinline}, @code{always_inline},
1526 @code{pure}, @code{const}, @code{nothrow}, @code{sentinel},
1527 @code{format}, @code{format_arg}, @code{no_instrument_function},
1528 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1529 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
1530 @code{alias}, @code{warn_unused_result}, @code{nonnull}
1531 and @code{externally_visible}. Several other
1532 attributes are defined for functions on particular target systems. Other
1533 attributes, including @code{section} are supported for variables declarations
1534 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1536 You may also specify attributes with @samp{__} preceding and following
1537 each keyword. This allows you to use them in header files without
1538 being concerned about a possible macro of the same name. For example,
1539 you may use @code{__noreturn__} instead of @code{noreturn}.
1541 @xref{Attribute Syntax}, for details of the exact syntax for using
1545 @c Keep this table alphabetized by attribute name. Treat _ as space.
1547 @item alias ("@var{target}")
1548 @cindex @code{alias} attribute
1549 The @code{alias} attribute causes the declaration to be emitted as an
1550 alias for another symbol, which must be specified. For instance,
1553 void __f () @{ /* @r{Do something.} */; @}
1554 void f () __attribute__ ((weak, alias ("__f")));
1557 declares @samp{f} to be a weak alias for @samp{__f}. In C++, the
1558 mangled name for the target must be used. It is an error if @samp{__f}
1559 is not defined in the same translation unit.
1561 Not all target machines support this attribute.
1564 @cindex @code{always_inline} function attribute
1565 Generally, functions are not inlined unless optimization is specified.
1566 For functions declared inline, this attribute inlines the function even
1567 if no optimization level was specified.
1570 @cindex functions that do pop the argument stack on the 386
1572 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1573 assume that the calling function will pop off the stack space used to
1574 pass arguments. This is
1575 useful to override the effects of the @option{-mrtd} switch.
1578 @cindex @code{const} function attribute
1579 Many functions do not examine any values except their arguments, and
1580 have no effects except the return value. Basically this is just slightly
1581 more strict class than the @code{pure} attribute below, since function is not
1582 allowed to read global memory.
1584 @cindex pointer arguments
1585 Note that a function that has pointer arguments and examines the data
1586 pointed to must @emph{not} be declared @code{const}. Likewise, a
1587 function that calls a non-@code{const} function usually must not be
1588 @code{const}. It does not make sense for a @code{const} function to
1591 The attribute @code{const} is not implemented in GCC versions earlier
1592 than 2.5. An alternative way to declare that a function has no side
1593 effects, which works in the current version and in some older versions,
1597 typedef int intfn ();
1599 extern const intfn square;
1602 This approach does not work in GNU C++ from 2.6.0 on, since the language
1603 specifies that the @samp{const} must be attached to the return value.
1607 @cindex @code{constructor} function attribute
1608 @cindex @code{destructor} function attribute
1609 The @code{constructor} attribute causes the function to be called
1610 automatically before execution enters @code{main ()}. Similarly, the
1611 @code{destructor} attribute causes the function to be called
1612 automatically after @code{main ()} has completed or @code{exit ()} has
1613 been called. Functions with these attributes are useful for
1614 initializing data that will be used implicitly during the execution of
1617 These attributes are not currently implemented for Objective-C@.
1620 @cindex @code{deprecated} attribute.
1621 The @code{deprecated} attribute results in a warning if the function
1622 is used anywhere in the source file. This is useful when identifying
1623 functions that are expected to be removed in a future version of a
1624 program. The warning also includes the location of the declaration
1625 of the deprecated function, to enable users to easily find further
1626 information about why the function is deprecated, or what they should
1627 do instead. Note that the warnings only occurs for uses:
1630 int old_fn () __attribute__ ((deprecated));
1632 int (*fn_ptr)() = old_fn;
1635 results in a warning on line 3 but not line 2.
1637 The @code{deprecated} attribute can also be used for variables and
1638 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
1641 @cindex @code{__declspec(dllexport)}
1642 On Microsoft Windows targets and Symbian OS targets the
1643 @code{dllexport} attribute causes the compiler to provide a global
1644 pointer to a pointer in a DLL, so that it can be referenced with the
1645 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
1646 name is formed by combining @code{_imp__} and the function or variable
1649 You can use @code{__declspec(dllexport)} as a synonym for
1650 @code{__attribute__ ((dllexport))} for compatibility with other
1653 On systems that support the @code{visibility} attribute, this
1654 attribute also implies ``default'' visibility, unless a
1655 @code{visibility} attribute is explicitly specified. You should avoid
1656 the use of @code{dllexport} with ``hidden'' or ``internal''
1657 visibility; in the future GCC may issue an error for those cases.
1659 Currently, the @code{dllexport} attribute is ignored for inlined
1660 functions, unless the @option{-fkeep-inline-functions} flag has been
1661 used. The attribute is also ignored for undefined symbols.
1663 When applied to C++ classes, the attribute marks defined non-inlined
1664 member functions and static data members as exports. Static consts
1665 initialized in-class are not marked unless they are also defined
1668 For Microsoft Windows targets there are alternative methods for
1669 including the symbol in the DLL's export table such as using a
1670 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
1671 the @option{--export-all} linker flag.
1674 @cindex @code{__declspec(dllimport)}
1675 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
1676 attribute causes the compiler to reference a function or variable via
1677 a global pointer to a pointer that is set up by the DLL exporting the
1678 symbol. The attribute implies @code{extern} storage. On Microsoft
1679 Windows targets, the pointer name is formed by combining @code{_imp__}
1680 and the function or variable name.
1682 You can use @code{__declspec(dllimport)} as a synonym for
1683 @code{__attribute__ ((dllimport))} for compatibility with other
1686 Currently, the attribute is ignored for inlined functions. If the
1687 attribute is applied to a symbol @emph{definition}, an error is reported.
1688 If a symbol previously declared @code{dllimport} is later defined, the
1689 attribute is ignored in subsequent references, and a warning is emitted.
1690 The attribute is also overridden by a subsequent declaration as
1693 When applied to C++ classes, the attribute marks non-inlined
1694 member functions and static data members as imports. However, the
1695 attribute is ignored for virtual methods to allow creation of vtables
1698 On the SH Symbian OS target the @code{dllimport} attribute also has
1699 another affect---it can cause the vtable and run-time type information
1700 for a class to be exported. This happens when the class has a
1701 dllimport'ed constructor or a non-inline, non-pure virtual function
1702 and, for either of those two conditions, the class also has a inline
1703 constructor or destructor and has a key function that is defined in
1704 the current translation unit.
1706 For Microsoft Windows based targets the use of the @code{dllimport}
1707 attribute on functions is not necessary, but provides a small
1708 performance benefit by eliminating a thunk in the DLL@. The use of the
1709 @code{dllimport} attribute on imported variables was required on older
1710 versions of the GNU linker, but can now be avoided by passing the
1711 @option{--enable-auto-import} switch to the GNU linker. As with
1712 functions, using the attribute for a variable eliminates a thunk in
1715 One drawback to using this attribute is that a pointer to a function
1716 or variable marked as @code{dllimport} cannot be used as a constant
1717 address. On Microsoft Windows targets, the attribute can be disabled
1718 for functions by setting the @option{-mnop-fun-dllimport} flag.
1721 @cindex eight bit data on the H8/300, H8/300H, and H8S
1722 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1723 variable should be placed into the eight bit data section.
1724 The compiler will generate more efficient code for certain operations
1725 on data in the eight bit data area. Note the eight bit data area is limited to
1728 You must use GAS and GLD from GNU binutils version 2.7 or later for
1729 this attribute to work correctly.
1731 @item exception_handler
1732 @cindex exception handler functions on the Blackfin processor
1733 Use this attribute on the Blackfin to indicate that the specified function
1734 is an exception handler. The compiler will generate function entry and
1735 exit sequences suitable for use in an exception handler when this
1736 attribute is present.
1739 @cindex functions which handle memory bank switching
1740 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
1741 use a calling convention that takes care of switching memory banks when
1742 entering and leaving a function. This calling convention is also the
1743 default when using the @option{-mlong-calls} option.
1745 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
1746 to call and return from a function.
1748 On 68HC11 the compiler will generate a sequence of instructions
1749 to invoke a board-specific routine to switch the memory bank and call the
1750 real function. The board-specific routine simulates a @code{call}.
1751 At the end of a function, it will jump to a board-specific routine
1752 instead of using @code{rts}. The board-specific return routine simulates
1756 @cindex functions that pop the argument stack on the 386
1757 On the Intel 386, the @code{fastcall} attribute causes the compiler to
1758 pass the first argument (if of integral type) in the register ECX and
1759 the second argument (if of integral type) in the register EDX@. Subsequent
1760 and other typed arguments are passed on the stack. The called function will
1761 pop the arguments off the stack. If the number of arguments is variable all
1762 arguments are pushed on the stack.
1764 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
1765 @cindex @code{format} function attribute
1767 The @code{format} attribute specifies that a function takes @code{printf},
1768 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
1769 should be type-checked against a format string. For example, the
1774 my_printf (void *my_object, const char *my_format, ...)
1775 __attribute__ ((format (printf, 2, 3)));
1779 causes the compiler to check the arguments in calls to @code{my_printf}
1780 for consistency with the @code{printf} style format string argument
1783 The parameter @var{archetype} determines how the format string is
1784 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
1785 or @code{strfmon}. (You can also use @code{__printf__},
1786 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
1787 parameter @var{string-index} specifies which argument is the format
1788 string argument (starting from 1), while @var{first-to-check} is the
1789 number of the first argument to check against the format string. For
1790 functions where the arguments are not available to be checked (such as
1791 @code{vprintf}), specify the third parameter as zero. In this case the
1792 compiler only checks the format string for consistency. For
1793 @code{strftime} formats, the third parameter is required to be zero.
1794 Since non-static C++ methods have an implicit @code{this} argument, the
1795 arguments of such methods should be counted from two, not one, when
1796 giving values for @var{string-index} and @var{first-to-check}.
1798 In the example above, the format string (@code{my_format}) is the second
1799 argument of the function @code{my_print}, and the arguments to check
1800 start with the third argument, so the correct parameters for the format
1801 attribute are 2 and 3.
1803 @opindex ffreestanding
1804 @opindex fno-builtin
1805 The @code{format} attribute allows you to identify your own functions
1806 which take format strings as arguments, so that GCC can check the
1807 calls to these functions for errors. The compiler always (unless
1808 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
1809 for the standard library functions @code{printf}, @code{fprintf},
1810 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
1811 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
1812 warnings are requested (using @option{-Wformat}), so there is no need to
1813 modify the header file @file{stdio.h}. In C99 mode, the functions
1814 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
1815 @code{vsscanf} are also checked. Except in strictly conforming C
1816 standard modes, the X/Open function @code{strfmon} is also checked as
1817 are @code{printf_unlocked} and @code{fprintf_unlocked}.
1818 @xref{C Dialect Options,,Options Controlling C Dialect}.
1820 The target may provide additional types of format checks.
1821 @xref{Target Format Checks,,Format Checks Specific to Particular
1824 @item format_arg (@var{string-index})
1825 @cindex @code{format_arg} function attribute
1826 @opindex Wformat-nonliteral
1827 The @code{format_arg} attribute specifies that a function takes a format
1828 string for a @code{printf}, @code{scanf}, @code{strftime} or
1829 @code{strfmon} style function and modifies it (for example, to translate
1830 it into another language), so the result can be passed to a
1831 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
1832 function (with the remaining arguments to the format function the same
1833 as they would have been for the unmodified string). For example, the
1838 my_dgettext (char *my_domain, const char *my_format)
1839 __attribute__ ((format_arg (2)));
1843 causes the compiler to check the arguments in calls to a @code{printf},
1844 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
1845 format string argument is a call to the @code{my_dgettext} function, for
1846 consistency with the format string argument @code{my_format}. If the
1847 @code{format_arg} attribute had not been specified, all the compiler
1848 could tell in such calls to format functions would be that the format
1849 string argument is not constant; this would generate a warning when
1850 @option{-Wformat-nonliteral} is used, but the calls could not be checked
1851 without the attribute.
1853 The parameter @var{string-index} specifies which argument is the format
1854 string argument (starting from one). Since non-static C++ methods have
1855 an implicit @code{this} argument, the arguments of such methods should
1856 be counted from two.
1858 The @code{format-arg} attribute allows you to identify your own
1859 functions which modify format strings, so that GCC can check the
1860 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
1861 type function whose operands are a call to one of your own function.
1862 The compiler always treats @code{gettext}, @code{dgettext}, and
1863 @code{dcgettext} in this manner except when strict ISO C support is
1864 requested by @option{-ansi} or an appropriate @option{-std} option, or
1865 @option{-ffreestanding} or @option{-fno-builtin}
1866 is used. @xref{C Dialect Options,,Options
1867 Controlling C Dialect}.
1869 @item function_vector
1870 @cindex calling functions through the function vector on the H8/300 processors
1871 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1872 function should be called through the function vector. Calling a
1873 function through the function vector will reduce code size, however;
1874 the function vector has a limited size (maximum 128 entries on the H8/300
1875 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
1877 You must use GAS and GLD from GNU binutils version 2.7 or later for
1878 this attribute to work correctly.
1881 @cindex interrupt handler functions
1882 Use this attribute on the ARM, AVR, C4x, M32R/D and Xstormy16 ports to indicate
1883 that the specified function is an interrupt handler. The compiler will
1884 generate function entry and exit sequences suitable for use in an
1885 interrupt handler when this attribute is present.
1887 Note, interrupt handlers for the Blackfin, m68k, H8/300, H8/300H, H8S, and
1888 SH processors can be specified via the @code{interrupt_handler} attribute.
1890 Note, on the AVR, interrupts will be enabled inside the function.
1892 Note, for the ARM, you can specify the kind of interrupt to be handled by
1893 adding an optional parameter to the interrupt attribute like this:
1896 void f () __attribute__ ((interrupt ("IRQ")));
1899 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
1901 @item interrupt_handler
1902 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
1903 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
1904 indicate that the specified function is an interrupt handler. The compiler
1905 will generate function entry and exit sequences suitable for use in an
1906 interrupt handler when this attribute is present.
1909 @cindex User stack pointer in interrupts on the Blackfin
1910 When used together with @code{interrupt_handler}, @code{exception_handler}
1911 or @code{nmi_handler}, code will be generated to load the stack pointer
1912 from the USP register in the function prologue.
1914 @item long_call/short_call
1915 @cindex indirect calls on ARM
1916 This attribute specifies how a particular function is called on
1917 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
1918 command line switch and @code{#pragma long_calls} settings. The
1919 @code{long_call} attribute causes the compiler to always call the
1920 function by first loading its address into a register and then using the
1921 contents of that register. The @code{short_call} attribute always places
1922 the offset to the function from the call site into the @samp{BL}
1923 instruction directly.
1925 @item longcall/shortcall
1926 @cindex functions called via pointer on the RS/6000 and PowerPC
1927 On the RS/6000 and PowerPC, the @code{longcall} attribute causes the
1928 compiler to always call this function via a pointer, just as it would if
1929 the @option{-mlongcall} option had been specified. The @code{shortcall}
1930 attribute causes the compiler not to do this. These attributes override
1931 both the @option{-mlongcall} switch and the @code{#pragma longcall}
1934 @xref{RS/6000 and PowerPC Options}, for more information on whether long
1935 calls are necessary.
1938 @cindex @code{malloc} attribute
1939 The @code{malloc} attribute is used to tell the compiler that a function
1940 may be treated as if any non-@code{NULL} pointer it returns cannot
1941 alias any other pointer valid when the function returns.
1942 This will often improve optimization.
1943 Standard functions with this property include @code{malloc} and
1944 @code{calloc}. @code{realloc}-like functions have this property as
1945 long as the old pointer is never referred to (including comparing it
1946 to the new pointer) after the function returns a non-@code{NULL}
1949 @item model (@var{model-name})
1950 @cindex function addressability on the M32R/D
1951 @cindex variable addressability on the IA-64
1953 On the M32R/D, use this attribute to set the addressability of an
1954 object, and of the code generated for a function. The identifier
1955 @var{model-name} is one of @code{small}, @code{medium}, or
1956 @code{large}, representing each of the code models.
1958 Small model objects live in the lower 16MB of memory (so that their
1959 addresses can be loaded with the @code{ld24} instruction), and are
1960 callable with the @code{bl} instruction.
1962 Medium model objects may live anywhere in the 32-bit address space (the
1963 compiler will generate @code{seth/add3} instructions to load their addresses),
1964 and are callable with the @code{bl} instruction.
1966 Large model objects may live anywhere in the 32-bit address space (the
1967 compiler will generate @code{seth/add3} instructions to load their addresses),
1968 and may not be reachable with the @code{bl} instruction (the compiler will
1969 generate the much slower @code{seth/add3/jl} instruction sequence).
1971 On IA-64, use this attribute to set the addressability of an object.
1972 At present, the only supported identifier for @var{model-name} is
1973 @code{small}, indicating addressability via ``small'' (22-bit)
1974 addresses (so that their addresses can be loaded with the @code{addl}
1975 instruction). Caveat: such addressing is by definition not position
1976 independent and hence this attribute must not be used for objects
1977 defined by shared libraries.
1980 @cindex function without a prologue/epilogue code
1981 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
1982 specified function does not need prologue/epilogue sequences generated by
1983 the compiler. It is up to the programmer to provide these sequences.
1986 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
1987 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
1988 use the normal calling convention based on @code{jsr} and @code{rts}.
1989 This attribute can be used to cancel the effect of the @option{-mlong-calls}
1993 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
1994 Use this attribute together with @code{interrupt_handler},
1995 @code{exception_handler} or @code{nmi_handler} to indicate that the function
1996 entry code should enable nested interrupts or exceptions.
1999 @cindex NMI handler functions on the Blackfin processor
2000 Use this attribute on the Blackfin to indicate that the specified function
2001 is an NMI handler. The compiler will generate function entry and
2002 exit sequences suitable for use in an NMI handler when this
2003 attribute is present.
2005 @item no_instrument_function
2006 @cindex @code{no_instrument_function} function attribute
2007 @opindex finstrument-functions
2008 If @option{-finstrument-functions} is given, profiling function calls will
2009 be generated at entry and exit of most user-compiled functions.
2010 Functions with this attribute will not be so instrumented.
2013 @cindex @code{noinline} function attribute
2014 This function attribute prevents a function from being considered for
2017 @item nonnull (@var{arg-index}, @dots{})
2018 @cindex @code{nonnull} function attribute
2019 The @code{nonnull} attribute specifies that some function parameters should
2020 be non-null pointers. For instance, the declaration:
2024 my_memcpy (void *dest, const void *src, size_t len)
2025 __attribute__((nonnull (1, 2)));
2029 causes the compiler to check that, in calls to @code{my_memcpy},
2030 arguments @var{dest} and @var{src} are non-null. If the compiler
2031 determines that a null pointer is passed in an argument slot marked
2032 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2033 is issued. The compiler may also choose to make optimizations based
2034 on the knowledge that certain function arguments will not be null.
2036 If no argument index list is given to the @code{nonnull} attribute,
2037 all pointer arguments are marked as non-null. To illustrate, the
2038 following declaration is equivalent to the previous example:
2042 my_memcpy (void *dest, const void *src, size_t len)
2043 __attribute__((nonnull));
2047 @cindex @code{noreturn} function attribute
2048 A few standard library functions, such as @code{abort} and @code{exit},
2049 cannot return. GCC knows this automatically. Some programs define
2050 their own functions that never return. You can declare them
2051 @code{noreturn} to tell the compiler this fact. For example,
2055 void fatal () __attribute__ ((noreturn));
2058 fatal (/* @r{@dots{}} */)
2060 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2066 The @code{noreturn} keyword tells the compiler to assume that
2067 @code{fatal} cannot return. It can then optimize without regard to what
2068 would happen if @code{fatal} ever did return. This makes slightly
2069 better code. More importantly, it helps avoid spurious warnings of
2070 uninitialized variables.
2072 The @code{noreturn} keyword does not affect the exceptional path when that
2073 applies: a @code{noreturn}-marked function may still return to the caller
2074 by throwing an exception or calling @code{longjmp}.
2076 Do not assume that registers saved by the calling function are
2077 restored before calling the @code{noreturn} function.
2079 It does not make sense for a @code{noreturn} function to have a return
2080 type other than @code{void}.
2082 The attribute @code{noreturn} is not implemented in GCC versions
2083 earlier than 2.5. An alternative way to declare that a function does
2084 not return, which works in the current version and in some older
2085 versions, is as follows:
2088 typedef void voidfn ();
2090 volatile voidfn fatal;
2093 This approach does not work in GNU C++.
2096 @cindex @code{nothrow} function attribute
2097 The @code{nothrow} attribute is used to inform the compiler that a
2098 function cannot throw an exception. For example, most functions in
2099 the standard C library can be guaranteed not to throw an exception
2100 with the notable exceptions of @code{qsort} and @code{bsearch} that
2101 take function pointer arguments. The @code{nothrow} attribute is not
2102 implemented in GCC versions earlier than 3.3.
2105 @cindex @code{pure} function attribute
2106 Many functions have no effects except the return value and their
2107 return value depends only on the parameters and/or global variables.
2108 Such a function can be subject
2109 to common subexpression elimination and loop optimization just as an
2110 arithmetic operator would be. These functions should be declared
2111 with the attribute @code{pure}. For example,
2114 int square (int) __attribute__ ((pure));
2118 says that the hypothetical function @code{square} is safe to call
2119 fewer times than the program says.
2121 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2122 Interesting non-pure functions are functions with infinite loops or those
2123 depending on volatile memory or other system resource, that may change between
2124 two consecutive calls (such as @code{feof} in a multithreading environment).
2126 The attribute @code{pure} is not implemented in GCC versions earlier
2129 @item regparm (@var{number})
2130 @cindex @code{regparm} attribute
2131 @cindex functions that are passed arguments in registers on the 386
2132 On the Intel 386, the @code{regparm} attribute causes the compiler to
2133 pass arguments number one to @var{number} if they are of integral type
2134 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2135 take a variable number of arguments will continue to be passed all of their
2136 arguments on the stack.
2138 Beware that on some ELF systems this attribute is unsuitable for
2139 global functions in shared libraries with lazy binding (which is the
2140 default). Lazy binding will send the first call via resolving code in
2141 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2142 per the standard calling conventions. Solaris 8 is affected by this.
2143 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2144 safe since the loaders there save all registers. (Lazy binding can be
2145 disabled with the linker or the loader if desired, to avoid the
2149 @cindex @code{sseregparm} attribute
2150 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2151 causes the compiler to pass up to 8 floating point arguments in
2152 SSE registers instead of on the stack. Functions that take a
2153 variable number of arguments will continue to pass all of their
2154 floating point arguments on the stack.
2157 @cindex @code{returns_twice} attribute
2158 The @code{returns_twice} attribute tells the compiler that a function may
2159 return more than one time. The compiler will ensure that all registers
2160 are dead before calling such a function and will emit a warning about
2161 the variables that may be clobbered after the second return from the
2162 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2163 The @code{longjmp}-like counterpart of such function, if any, might need
2164 to be marked with the @code{noreturn} attribute.
2167 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2168 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2169 all registers except the stack pointer should be saved in the prologue
2170 regardless of whether they are used or not.
2172 @item section ("@var{section-name}")
2173 @cindex @code{section} function attribute
2174 Normally, the compiler places the code it generates in the @code{text} section.
2175 Sometimes, however, you need additional sections, or you need certain
2176 particular functions to appear in special sections. The @code{section}
2177 attribute specifies that a function lives in a particular section.
2178 For example, the declaration:
2181 extern void foobar (void) __attribute__ ((section ("bar")));
2185 puts the function @code{foobar} in the @code{bar} section.
2187 Some file formats do not support arbitrary sections so the @code{section}
2188 attribute is not available on all platforms.
2189 If you need to map the entire contents of a module to a particular
2190 section, consider using the facilities of the linker instead.
2193 @cindex @code{sentinel} function attribute
2194 This function attribute ensures that a parameter in a function call is
2195 an explicit @code{NULL}. The attribute is only valid on variadic
2196 functions. By default, the sentinel is located at position zero, the
2197 last parameter of the function call. If an optional integer position
2198 argument P is supplied to the attribute, the sentinel must be located at
2199 position P counting backwards from the end of the argument list.
2202 __attribute__ ((sentinel))
2204 __attribute__ ((sentinel(0)))
2207 The attribute is automatically set with a position of 0 for the built-in
2208 functions @code{execl} and @code{execlp}. The built-in function
2209 @code{execle} has the attribute set with a position of 1.
2211 A valid @code{NULL} in this context is defined as zero with any pointer
2212 type. If your system defines the @code{NULL} macro with an integer type
2213 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2214 with a copy that redefines NULL appropriately.
2216 The warnings for missing or incorrect sentinels are enabled with
2220 See long_call/short_call.
2223 See longcall/shortcall.
2226 @cindex signal handler functions on the AVR processors
2227 Use this attribute on the AVR to indicate that the specified
2228 function is a signal handler. The compiler will generate function
2229 entry and exit sequences suitable for use in a signal handler when this
2230 attribute is present. Interrupts will be disabled inside the function.
2233 Use this attribute on the SH to indicate an @code{interrupt_handler}
2234 function should switch to an alternate stack. It expects a string
2235 argument that names a global variable holding the address of the
2240 void f () __attribute__ ((interrupt_handler,
2241 sp_switch ("alt_stack")));
2245 @cindex functions that pop the argument stack on the 386
2246 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2247 assume that the called function will pop off the stack space used to
2248 pass arguments, unless it takes a variable number of arguments.
2251 @cindex tiny data section on the H8/300H and H8S
2252 Use this attribute on the H8/300H and H8S to indicate that the specified
2253 variable should be placed into the tiny data section.
2254 The compiler will generate more efficient code for loads and stores
2255 on data in the tiny data section. Note the tiny data area is limited to
2256 slightly under 32kbytes of data.
2259 Use this attribute on the SH for an @code{interrupt_handler} to return using
2260 @code{trapa} instead of @code{rte}. This attribute expects an integer
2261 argument specifying the trap number to be used.
2264 @cindex @code{unused} attribute.
2265 This attribute, attached to a function, means that the function is meant
2266 to be possibly unused. GCC will not produce a warning for this
2270 @cindex @code{used} attribute.
2271 This attribute, attached to a function, means that code must be emitted
2272 for the function even if it appears that the function is not referenced.
2273 This is useful, for example, when the function is referenced only in
2276 @item visibility ("@var{visibility_type}")
2277 @cindex @code{visibility} attribute
2278 The @code{visibility} attribute on ELF targets causes the declaration
2279 to be emitted with default, hidden, protected or internal visibility.
2282 void __attribute__ ((visibility ("protected")))
2283 f () @{ /* @r{Do something.} */; @}
2284 int i __attribute__ ((visibility ("hidden")));
2287 See the ELF gABI for complete details, but the short story is:
2290 @c keep this list of visibilities in alphabetical order.
2293 Default visibility is the normal case for ELF@. This value is
2294 available for the visibility attribute to override other options
2295 that may change the assumed visibility of symbols.
2298 Hidden visibility indicates that the symbol will not be placed into
2299 the dynamic symbol table, so no other @dfn{module} (executable or
2300 shared library) can reference it directly.
2303 Internal visibility is like hidden visibility, but with additional
2304 processor specific semantics. Unless otherwise specified by the psABI,
2305 GCC defines internal visibility to mean that the function is @emph{never}
2306 called from another module. Note that hidden symbols, while they cannot
2307 be referenced directly by other modules, can be referenced indirectly via
2308 function pointers. By indicating that a symbol cannot be called from
2309 outside the module, GCC may for instance omit the load of a PIC register
2310 since it is known that the calling function loaded the correct value.
2313 Protected visibility indicates that the symbol will be placed in the
2314 dynamic symbol table, but that references within the defining module
2315 will bind to the local symbol. That is, the symbol cannot be overridden
2320 Not all ELF targets support this attribute.
2322 @item warn_unused_result
2323 @cindex @code{warn_unused_result} attribute
2324 The @code{warn_unused_result} attribute causes a warning to be emitted
2325 if a caller of the function with this attribute does not use its
2326 return value. This is useful for functions where not checking
2327 the result is either a security problem or always a bug, such as
2331 int fn () __attribute__ ((warn_unused_result));
2334 if (fn () < 0) return -1;
2340 results in warning on line 5.
2343 @cindex @code{weak} attribute
2344 The @code{weak} attribute causes the declaration to be emitted as a weak
2345 symbol rather than a global. This is primarily useful in defining
2346 library functions which can be overridden in user code, though it can
2347 also be used with non-function declarations. Weak symbols are supported
2348 for ELF targets, and also for a.out targets when using the GNU assembler
2351 @item externally_visible
2352 @cindex @code{externally_visible} attribute.
2353 This attribute, attached to a global variable or function nullify
2354 effect of @option{-fwhole-program} command line option, so the object
2355 remain visible outside the current compilation unit
2359 You can specify multiple attributes in a declaration by separating them
2360 by commas within the double parentheses or by immediately following an
2361 attribute declaration with another attribute declaration.
2363 @cindex @code{#pragma}, reason for not using
2364 @cindex pragma, reason for not using
2365 Some people object to the @code{__attribute__} feature, suggesting that
2366 ISO C's @code{#pragma} should be used instead. At the time
2367 @code{__attribute__} was designed, there were two reasons for not doing
2372 It is impossible to generate @code{#pragma} commands from a macro.
2375 There is no telling what the same @code{#pragma} might mean in another
2379 These two reasons applied to almost any application that might have been
2380 proposed for @code{#pragma}. It was basically a mistake to use
2381 @code{#pragma} for @emph{anything}.
2383 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2384 to be generated from macros. In addition, a @code{#pragma GCC}
2385 namespace is now in use for GCC-specific pragmas. However, it has been
2386 found convenient to use @code{__attribute__} to achieve a natural
2387 attachment of attributes to their corresponding declarations, whereas
2388 @code{#pragma GCC} is of use for constructs that do not naturally form
2389 part of the grammar. @xref{Other Directives,,Miscellaneous
2390 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2392 @node Attribute Syntax
2393 @section Attribute Syntax
2394 @cindex attribute syntax
2396 This section describes the syntax with which @code{__attribute__} may be
2397 used, and the constructs to which attribute specifiers bind, for the C
2398 language. Some details may vary for C++ and Objective-C@. Because of
2399 infelicities in the grammar for attributes, some forms described here
2400 may not be successfully parsed in all cases.
2402 There are some problems with the semantics of attributes in C++. For
2403 example, there are no manglings for attributes, although they may affect
2404 code generation, so problems may arise when attributed types are used in
2405 conjunction with templates or overloading. Similarly, @code{typeid}
2406 does not distinguish between types with different attributes. Support
2407 for attributes in C++ may be restricted in future to attributes on
2408 declarations only, but not on nested declarators.
2410 @xref{Function Attributes}, for details of the semantics of attributes
2411 applying to functions. @xref{Variable Attributes}, for details of the
2412 semantics of attributes applying to variables. @xref{Type Attributes},
2413 for details of the semantics of attributes applying to structure, union
2414 and enumerated types.
2416 An @dfn{attribute specifier} is of the form
2417 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2418 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2419 each attribute is one of the following:
2423 Empty. Empty attributes are ignored.
2426 A word (which may be an identifier such as @code{unused}, or a reserved
2427 word such as @code{const}).
2430 A word, followed by, in parentheses, parameters for the attribute.
2431 These parameters take one of the following forms:
2435 An identifier. For example, @code{mode} attributes use this form.
2438 An identifier followed by a comma and a non-empty comma-separated list
2439 of expressions. For example, @code{format} attributes use this form.
2442 A possibly empty comma-separated list of expressions. For example,
2443 @code{format_arg} attributes use this form with the list being a single
2444 integer constant expression, and @code{alias} attributes use this form
2445 with the list being a single string constant.
2449 An @dfn{attribute specifier list} is a sequence of one or more attribute
2450 specifiers, not separated by any other tokens.
2452 In GNU C, an attribute specifier list may appear after the colon following a
2453 label, other than a @code{case} or @code{default} label. The only
2454 attribute it makes sense to use after a label is @code{unused}. This
2455 feature is intended for code generated by programs which contains labels
2456 that may be unused but which is compiled with @option{-Wall}. It would
2457 not normally be appropriate to use in it human-written code, though it
2458 could be useful in cases where the code that jumps to the label is
2459 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2460 such placement of attribute lists, as it is permissible for a
2461 declaration, which could begin with an attribute list, to be labelled in
2462 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2463 does not arise there.
2465 An attribute specifier list may appear as part of a @code{struct},
2466 @code{union} or @code{enum} specifier. It may go either immediately
2467 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2468 the closing brace. It is ignored if the content of the structure, union
2469 or enumerated type is not defined in the specifier in which the
2470 attribute specifier list is used---that is, in usages such as
2471 @code{struct __attribute__((foo)) bar} with no following opening brace.
2472 Where attribute specifiers follow the closing brace, they are considered
2473 to relate to the structure, union or enumerated type defined, not to any
2474 enclosing declaration the type specifier appears in, and the type
2475 defined is not complete until after the attribute specifiers.
2476 @c Otherwise, there would be the following problems: a shift/reduce
2477 @c conflict between attributes binding the struct/union/enum and
2478 @c binding to the list of specifiers/qualifiers; and "aligned"
2479 @c attributes could use sizeof for the structure, but the size could be
2480 @c changed later by "packed" attributes.
2482 Otherwise, an attribute specifier appears as part of a declaration,
2483 counting declarations of unnamed parameters and type names, and relates
2484 to that declaration (which may be nested in another declaration, for
2485 example in the case of a parameter declaration), or to a particular declarator
2486 within a declaration. Where an
2487 attribute specifier is applied to a parameter declared as a function or
2488 an array, it should apply to the function or array rather than the
2489 pointer to which the parameter is implicitly converted, but this is not
2490 yet correctly implemented.
2492 Any list of specifiers and qualifiers at the start of a declaration may
2493 contain attribute specifiers, whether or not such a list may in that
2494 context contain storage class specifiers. (Some attributes, however,
2495 are essentially in the nature of storage class specifiers, and only make
2496 sense where storage class specifiers may be used; for example,
2497 @code{section}.) There is one necessary limitation to this syntax: the
2498 first old-style parameter declaration in a function definition cannot
2499 begin with an attribute specifier, because such an attribute applies to
2500 the function instead by syntax described below (which, however, is not
2501 yet implemented in this case). In some other cases, attribute
2502 specifiers are permitted by this grammar but not yet supported by the
2503 compiler. All attribute specifiers in this place relate to the
2504 declaration as a whole. In the obsolescent usage where a type of
2505 @code{int} is implied by the absence of type specifiers, such a list of
2506 specifiers and qualifiers may be an attribute specifier list with no
2507 other specifiers or qualifiers.
2509 At present, the first parameter in a function prototype must have some
2510 type specifier which is not an attribute specifier; this resolves an
2511 ambiguity in the interpretation of @code{void f(int
2512 (__attribute__((foo)) x))}, but is subject to change. At present, if
2513 the parentheses of a function declarator contain only attributes then
2514 those attributes are ignored, rather than yielding an error or warning
2515 or implying a single parameter of type int, but this is subject to
2518 An attribute specifier list may appear immediately before a declarator
2519 (other than the first) in a comma-separated list of declarators in a
2520 declaration of more than one identifier using a single list of
2521 specifiers and qualifiers. Such attribute specifiers apply
2522 only to the identifier before whose declarator they appear. For
2526 __attribute__((noreturn)) void d0 (void),
2527 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2532 the @code{noreturn} attribute applies to all the functions
2533 declared; the @code{format} attribute only applies to @code{d1}.
2535 An attribute specifier list may appear immediately before the comma,
2536 @code{=} or semicolon terminating the declaration of an identifier other
2537 than a function definition. At present, such attribute specifiers apply
2538 to the declared object or function, but in future they may attach to the
2539 outermost adjacent declarator. In simple cases there is no difference,
2540 but, for example, in
2543 void (****f)(void) __attribute__((noreturn));
2547 at present the @code{noreturn} attribute applies to @code{f}, which
2548 causes a warning since @code{f} is not a function, but in future it may
2549 apply to the function @code{****f}. The precise semantics of what
2550 attributes in such cases will apply to are not yet specified. Where an
2551 assembler name for an object or function is specified (@pxref{Asm
2552 Labels}), at present the attribute must follow the @code{asm}
2553 specification; in future, attributes before the @code{asm} specification
2554 may apply to the adjacent declarator, and those after it to the declared
2557 An attribute specifier list may, in future, be permitted to appear after
2558 the declarator in a function definition (before any old-style parameter
2559 declarations or the function body).
2561 Attribute specifiers may be mixed with type qualifiers appearing inside
2562 the @code{[]} of a parameter array declarator, in the C99 construct by
2563 which such qualifiers are applied to the pointer to which the array is
2564 implicitly converted. Such attribute specifiers apply to the pointer,
2565 not to the array, but at present this is not implemented and they are
2568 An attribute specifier list may appear at the start of a nested
2569 declarator. At present, there are some limitations in this usage: the
2570 attributes correctly apply to the declarator, but for most individual
2571 attributes the semantics this implies are not implemented.
2572 When attribute specifiers follow the @code{*} of a pointer
2573 declarator, they may be mixed with any type qualifiers present.
2574 The following describes the formal semantics of this syntax. It will make the
2575 most sense if you are familiar with the formal specification of
2576 declarators in the ISO C standard.
2578 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2579 D1}, where @code{T} contains declaration specifiers that specify a type
2580 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2581 contains an identifier @var{ident}. The type specified for @var{ident}
2582 for derived declarators whose type does not include an attribute
2583 specifier is as in the ISO C standard.
2585 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2586 and the declaration @code{T D} specifies the type
2587 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2588 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2589 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2591 If @code{D1} has the form @code{*
2592 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2593 declaration @code{T D} specifies the type
2594 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2595 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2596 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2602 void (__attribute__((noreturn)) ****f) (void);
2606 specifies the type ``pointer to pointer to pointer to pointer to
2607 non-returning function returning @code{void}''. As another example,
2610 char *__attribute__((aligned(8))) *f;
2614 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2615 Note again that this does not work with most attributes; for example,
2616 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2617 is not yet supported.
2619 For compatibility with existing code written for compiler versions that
2620 did not implement attributes on nested declarators, some laxity is
2621 allowed in the placing of attributes. If an attribute that only applies
2622 to types is applied to a declaration, it will be treated as applying to
2623 the type of that declaration. If an attribute that only applies to
2624 declarations is applied to the type of a declaration, it will be treated
2625 as applying to that declaration; and, for compatibility with code
2626 placing the attributes immediately before the identifier declared, such
2627 an attribute applied to a function return type will be treated as
2628 applying to the function type, and such an attribute applied to an array
2629 element type will be treated as applying to the array type. If an
2630 attribute that only applies to function types is applied to a
2631 pointer-to-function type, it will be treated as applying to the pointer
2632 target type; if such an attribute is applied to a function return type
2633 that is not a pointer-to-function type, it will be treated as applying
2634 to the function type.
2636 @node Function Prototypes
2637 @section Prototypes and Old-Style Function Definitions
2638 @cindex function prototype declarations
2639 @cindex old-style function definitions
2640 @cindex promotion of formal parameters
2642 GNU C extends ISO C to allow a function prototype to override a later
2643 old-style non-prototype definition. Consider the following example:
2646 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2653 /* @r{Prototype function declaration.} */
2654 int isroot P((uid_t));
2656 /* @r{Old-style function definition.} */
2658 isroot (x) /* @r{??? lossage here ???} */
2665 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2666 not allow this example, because subword arguments in old-style
2667 non-prototype definitions are promoted. Therefore in this example the
2668 function definition's argument is really an @code{int}, which does not
2669 match the prototype argument type of @code{short}.
2671 This restriction of ISO C makes it hard to write code that is portable
2672 to traditional C compilers, because the programmer does not know
2673 whether the @code{uid_t} type is @code{short}, @code{int}, or
2674 @code{long}. Therefore, in cases like these GNU C allows a prototype
2675 to override a later old-style definition. More precisely, in GNU C, a
2676 function prototype argument type overrides the argument type specified
2677 by a later old-style definition if the former type is the same as the
2678 latter type before promotion. Thus in GNU C the above example is
2679 equivalent to the following:
2692 GNU C++ does not support old-style function definitions, so this
2693 extension is irrelevant.
2696 @section C++ Style Comments
2698 @cindex C++ comments
2699 @cindex comments, C++ style
2701 In GNU C, you may use C++ style comments, which start with @samp{//} and
2702 continue until the end of the line. Many other C implementations allow
2703 such comments, and they are included in the 1999 C standard. However,
2704 C++ style comments are not recognized if you specify an @option{-std}
2705 option specifying a version of ISO C before C99, or @option{-ansi}
2706 (equivalent to @option{-std=c89}).
2709 @section Dollar Signs in Identifier Names
2711 @cindex dollar signs in identifier names
2712 @cindex identifier names, dollar signs in
2714 In GNU C, you may normally use dollar signs in identifier names.
2715 This is because many traditional C implementations allow such identifiers.
2716 However, dollar signs in identifiers are not supported on a few target
2717 machines, typically because the target assembler does not allow them.
2719 @node Character Escapes
2720 @section The Character @key{ESC} in Constants
2722 You can use the sequence @samp{\e} in a string or character constant to
2723 stand for the ASCII character @key{ESC}.
2726 @section Inquiring on Alignment of Types or Variables
2728 @cindex type alignment
2729 @cindex variable alignment
2731 The keyword @code{__alignof__} allows you to inquire about how an object
2732 is aligned, or the minimum alignment usually required by a type. Its
2733 syntax is just like @code{sizeof}.
2735 For example, if the target machine requires a @code{double} value to be
2736 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2737 This is true on many RISC machines. On more traditional machine
2738 designs, @code{__alignof__ (double)} is 4 or even 2.
2740 Some machines never actually require alignment; they allow reference to any
2741 data type even at an odd address. For these machines, @code{__alignof__}
2742 reports the @emph{recommended} alignment of a type.
2744 If the operand of @code{__alignof__} is an lvalue rather than a type,
2745 its value is the required alignment for its type, taking into account
2746 any minimum alignment specified with GCC's @code{__attribute__}
2747 extension (@pxref{Variable Attributes}). For example, after this
2751 struct foo @{ int x; char y; @} foo1;
2755 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2756 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2758 It is an error to ask for the alignment of an incomplete type.
2760 @node Variable Attributes
2761 @section Specifying Attributes of Variables
2762 @cindex attribute of variables
2763 @cindex variable attributes
2765 The keyword @code{__attribute__} allows you to specify special
2766 attributes of variables or structure fields. This keyword is followed
2767 by an attribute specification inside double parentheses. Some
2768 attributes are currently defined generically for variables.
2769 Other attributes are defined for variables on particular target
2770 systems. Other attributes are available for functions
2771 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
2772 Other front ends might define more attributes
2773 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2775 You may also specify attributes with @samp{__} preceding and following
2776 each keyword. This allows you to use them in header files without
2777 being concerned about a possible macro of the same name. For example,
2778 you may use @code{__aligned__} instead of @code{aligned}.
2780 @xref{Attribute Syntax}, for details of the exact syntax for using
2784 @cindex @code{aligned} attribute
2785 @item aligned (@var{alignment})
2786 This attribute specifies a minimum alignment for the variable or
2787 structure field, measured in bytes. For example, the declaration:
2790 int x __attribute__ ((aligned (16))) = 0;
2794 causes the compiler to allocate the global variable @code{x} on a
2795 16-byte boundary. On a 68040, this could be used in conjunction with
2796 an @code{asm} expression to access the @code{move16} instruction which
2797 requires 16-byte aligned operands.
2799 You can also specify the alignment of structure fields. For example, to
2800 create a double-word aligned @code{int} pair, you could write:
2803 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2807 This is an alternative to creating a union with a @code{double} member
2808 that forces the union to be double-word aligned.
2810 As in the preceding examples, you can explicitly specify the alignment
2811 (in bytes) that you wish the compiler to use for a given variable or
2812 structure field. Alternatively, you can leave out the alignment factor
2813 and just ask the compiler to align a variable or field to the maximum
2814 useful alignment for the target machine you are compiling for. For
2815 example, you could write:
2818 short array[3] __attribute__ ((aligned));
2821 Whenever you leave out the alignment factor in an @code{aligned} attribute
2822 specification, the compiler automatically sets the alignment for the declared
2823 variable or field to the largest alignment which is ever used for any data
2824 type on the target machine you are compiling for. Doing this can often make
2825 copy operations more efficient, because the compiler can use whatever
2826 instructions copy the biggest chunks of memory when performing copies to
2827 or from the variables or fields that you have aligned this way.
2829 The @code{aligned} attribute can only increase the alignment; but you
2830 can decrease it by specifying @code{packed} as well. See below.
2832 Note that the effectiveness of @code{aligned} attributes may be limited
2833 by inherent limitations in your linker. On many systems, the linker is
2834 only able to arrange for variables to be aligned up to a certain maximum
2835 alignment. (For some linkers, the maximum supported alignment may
2836 be very very small.) If your linker is only able to align variables
2837 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
2838 in an @code{__attribute__} will still only provide you with 8 byte
2839 alignment. See your linker documentation for further information.
2841 @item cleanup (@var{cleanup_function})
2842 @cindex @code{cleanup} attribute
2843 The @code{cleanup} attribute runs a function when the variable goes
2844 out of scope. This attribute can only be applied to auto function
2845 scope variables; it may not be applied to parameters or variables
2846 with static storage duration. The function must take one parameter,
2847 a pointer to a type compatible with the variable. The return value
2848 of the function (if any) is ignored.
2850 If @option{-fexceptions} is enabled, then @var{cleanup_function}
2851 will be run during the stack unwinding that happens during the
2852 processing of the exception. Note that the @code{cleanup} attribute
2853 does not allow the exception to be caught, only to perform an action.
2854 It is undefined what happens if @var{cleanup_function} does not
2859 @cindex @code{common} attribute
2860 @cindex @code{nocommon} attribute
2863 The @code{common} attribute requests GCC to place a variable in
2864 ``common'' storage. The @code{nocommon} attribute requests the
2865 opposite---to allocate space for it directly.
2867 These attributes override the default chosen by the
2868 @option{-fno-common} and @option{-fcommon} flags respectively.
2871 @cindex @code{deprecated} attribute
2872 The @code{deprecated} attribute results in a warning if the variable
2873 is used anywhere in the source file. This is useful when identifying
2874 variables that are expected to be removed in a future version of a
2875 program. The warning also includes the location of the declaration
2876 of the deprecated variable, to enable users to easily find further
2877 information about why the variable is deprecated, or what they should
2878 do instead. Note that the warning only occurs for uses:
2881 extern int old_var __attribute__ ((deprecated));
2883 int new_fn () @{ return old_var; @}
2886 results in a warning on line 3 but not line 2.
2888 The @code{deprecated} attribute can also be used for functions and
2889 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
2891 @item mode (@var{mode})
2892 @cindex @code{mode} attribute
2893 This attribute specifies the data type for the declaration---whichever
2894 type corresponds to the mode @var{mode}. This in effect lets you
2895 request an integer or floating point type according to its width.
2897 You may also specify a mode of @samp{byte} or @samp{__byte__} to
2898 indicate the mode corresponding to a one-byte integer, @samp{word} or
2899 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
2900 or @samp{__pointer__} for the mode used to represent pointers.
2903 @cindex @code{packed} attribute
2904 The @code{packed} attribute specifies that a variable or structure field
2905 should have the smallest possible alignment---one byte for a variable,
2906 and one bit for a field, unless you specify a larger value with the
2907 @code{aligned} attribute.
2909 Here is a structure in which the field @code{x} is packed, so that it
2910 immediately follows @code{a}:
2916 int x[2] __attribute__ ((packed));
2920 @item section ("@var{section-name}")
2921 @cindex @code{section} variable attribute
2922 Normally, the compiler places the objects it generates in sections like
2923 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
2924 or you need certain particular variables to appear in special sections,
2925 for example to map to special hardware. The @code{section}
2926 attribute specifies that a variable (or function) lives in a particular
2927 section. For example, this small program uses several specific section names:
2930 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
2931 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
2932 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
2933 int init_data __attribute__ ((section ("INITDATA"))) = 0;
2937 /* @r{Initialize stack pointer} */
2938 init_sp (stack + sizeof (stack));
2940 /* @r{Initialize initialized data} */
2941 memcpy (&init_data, &data, &edata - &data);
2943 /* @r{Turn on the serial ports} */
2950 Use the @code{section} attribute with an @emph{initialized} definition
2951 of a @emph{global} variable, as shown in the example. GCC issues
2952 a warning and otherwise ignores the @code{section} attribute in
2953 uninitialized variable declarations.
2955 You may only use the @code{section} attribute with a fully initialized
2956 global definition because of the way linkers work. The linker requires
2957 each object be defined once, with the exception that uninitialized
2958 variables tentatively go in the @code{common} (or @code{bss}) section
2959 and can be multiply ``defined''. You can force a variable to be
2960 initialized with the @option{-fno-common} flag or the @code{nocommon}
2963 Some file formats do not support arbitrary sections so the @code{section}
2964 attribute is not available on all platforms.
2965 If you need to map the entire contents of a module to a particular
2966 section, consider using the facilities of the linker instead.
2969 @cindex @code{shared} variable attribute
2970 On Microsoft Windows, in addition to putting variable definitions in a named
2971 section, the section can also be shared among all running copies of an
2972 executable or DLL@. For example, this small program defines shared data
2973 by putting it in a named section @code{shared} and marking the section
2977 int foo __attribute__((section ("shared"), shared)) = 0;
2982 /* @r{Read and write foo. All running
2983 copies see the same value.} */
2989 You may only use the @code{shared} attribute along with @code{section}
2990 attribute with a fully initialized global definition because of the way
2991 linkers work. See @code{section} attribute for more information.
2993 The @code{shared} attribute is only available on Microsoft Windows@.
2995 @item tls_model ("@var{tls_model}")
2996 @cindex @code{tls_model} attribute
2997 The @code{tls_model} attribute sets thread-local storage model
2998 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
2999 overriding @option{-ftls-model=} command line switch on a per-variable
3001 The @var{tls_model} argument should be one of @code{global-dynamic},
3002 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3004 Not all targets support this attribute.
3006 @item transparent_union
3007 This attribute, attached to a function parameter which is a union, means
3008 that the corresponding argument may have the type of any union member,
3009 but the argument is passed as if its type were that of the first union
3010 member. For more details see @xref{Type Attributes}. You can also use
3011 this attribute on a @code{typedef} for a union data type; then it
3012 applies to all function parameters with that type.
3015 This attribute, attached to a variable, means that the variable is meant
3016 to be possibly unused. GCC will not produce a warning for this
3019 @item vector_size (@var{bytes})
3020 This attribute specifies the vector size for the variable, measured in
3021 bytes. For example, the declaration:
3024 int foo __attribute__ ((vector_size (16)));
3028 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3029 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3030 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3032 This attribute is only applicable to integral and float scalars,
3033 although arrays, pointers, and function return values are allowed in
3034 conjunction with this construct.
3036 Aggregates with this attribute are invalid, even if they are of the same
3037 size as a corresponding scalar. For example, the declaration:
3040 struct S @{ int a; @};
3041 struct S __attribute__ ((vector_size (16))) foo;
3045 is invalid even if the size of the structure is the same as the size of
3049 The @code{selectany} attribute causes an initialized global variable to
3050 have link-once semantics. When multiple definitions of the variable are
3051 encountered by the linker, the first is selected and the remainder are
3052 discarded. Following usage by the Microsoft compiler, the linker is told
3053 @emph{not} to warn about size or content differences of the multiple
3056 Although the primary usage of this attribute is for POD types, the
3057 attribute can also be applied to global C++ objects that are initialized
3058 by a constructor. In this case, the static initialization and destruction
3059 code for the object is emitted in each translation defining the object,
3060 but the calls to the constructor and destructor are protected by a
3061 link-once guard variable.
3063 The @code{selectany} attribute is only available on Microsoft Windows
3064 targets. You can use @code{__declspec (selectany)} as a synonym for
3065 @code{__attribute__ ((selectany))} for compatibility with other
3069 The @code{weak} attribute is described in @xref{Function Attributes}.
3072 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3075 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3079 @subsection M32R/D Variable Attributes
3081 One attribute is currently defined for the M32R/D@.
3084 @item model (@var{model-name})
3085 @cindex variable addressability on the M32R/D
3086 Use this attribute on the M32R/D to set the addressability of an object.
3087 The identifier @var{model-name} is one of @code{small}, @code{medium},
3088 or @code{large}, representing each of the code models.
3090 Small model objects live in the lower 16MB of memory (so that their
3091 addresses can be loaded with the @code{ld24} instruction).
3093 Medium and large model objects may live anywhere in the 32-bit address space
3094 (the compiler will generate @code{seth/add3} instructions to load their
3098 @subsection i386 Variable Attributes
3100 Two attributes are currently defined for i386 configurations:
3101 @code{ms_struct} and @code{gcc_struct}
3106 @cindex @code{ms_struct} attribute
3107 @cindex @code{gcc_struct} attribute
3109 If @code{packed} is used on a structure, or if bit-fields are used
3110 it may be that the Microsoft ABI packs them differently
3111 than GCC would normally pack them. Particularly when moving packed
3112 data between functions compiled with GCC and the native Microsoft compiler
3113 (either via function call or as data in a file), it may be necessary to access
3116 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3117 compilers to match the native Microsoft compiler.
3120 @subsection Xstormy16 Variable Attributes
3122 One attribute is currently defined for xstormy16 configurations:
3127 @cindex @code{below100} attribute
3129 If a variable has the @code{below100} attribute (@code{BELOW100} is
3130 allowed also), GCC will place the variable in the first 0x100 bytes of
3131 memory and use special opcodes to access it. Such variables will be
3132 placed in either the @code{.bss_below100} section or the
3133 @code{.data_below100} section.
3137 @node Type Attributes
3138 @section Specifying Attributes of Types
3139 @cindex attribute of types
3140 @cindex type attributes
3142 The keyword @code{__attribute__} allows you to specify special
3143 attributes of @code{struct} and @code{union} types when you define such
3144 types. This keyword is followed by an attribute specification inside
3145 double parentheses. Six attributes are currently defined for types:
3146 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3147 @code{deprecated} and @code{may_alias}. Other attributes are defined for
3148 functions (@pxref{Function Attributes}) and for variables
3149 (@pxref{Variable Attributes}).
3151 You may also specify any one of these attributes with @samp{__}
3152 preceding and following its keyword. This allows you to use these
3153 attributes in header files without being concerned about a possible
3154 macro of the same name. For example, you may use @code{__aligned__}
3155 instead of @code{aligned}.
3157 You may specify the @code{aligned} and @code{transparent_union}
3158 attributes either in a @code{typedef} declaration or just past the
3159 closing curly brace of a complete enum, struct or union type
3160 @emph{definition} and the @code{packed} attribute only past the closing
3161 brace of a definition.
3163 You may also specify attributes between the enum, struct or union
3164 tag and the name of the type rather than after the closing brace.
3166 @xref{Attribute Syntax}, for details of the exact syntax for using
3170 @cindex @code{aligned} attribute
3171 @item aligned (@var{alignment})
3172 This attribute specifies a minimum alignment (in bytes) for variables
3173 of the specified type. For example, the declarations:
3176 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3177 typedef int more_aligned_int __attribute__ ((aligned (8)));
3181 force the compiler to insure (as far as it can) that each variable whose
3182 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3183 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3184 variables of type @code{struct S} aligned to 8-byte boundaries allows
3185 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3186 store) instructions when copying one variable of type @code{struct S} to
3187 another, thus improving run-time efficiency.
3189 Note that the alignment of any given @code{struct} or @code{union} type
3190 is required by the ISO C standard to be at least a perfect multiple of
3191 the lowest common multiple of the alignments of all of the members of
3192 the @code{struct} or @code{union} in question. This means that you @emph{can}
3193 effectively adjust the alignment of a @code{struct} or @code{union}
3194 type by attaching an @code{aligned} attribute to any one of the members
3195 of such a type, but the notation illustrated in the example above is a
3196 more obvious, intuitive, and readable way to request the compiler to
3197 adjust the alignment of an entire @code{struct} or @code{union} type.
3199 As in the preceding example, you can explicitly specify the alignment
3200 (in bytes) that you wish the compiler to use for a given @code{struct}
3201 or @code{union} type. Alternatively, you can leave out the alignment factor
3202 and just ask the compiler to align a type to the maximum
3203 useful alignment for the target machine you are compiling for. For
3204 example, you could write:
3207 struct S @{ short f[3]; @} __attribute__ ((aligned));
3210 Whenever you leave out the alignment factor in an @code{aligned}
3211 attribute specification, the compiler automatically sets the alignment
3212 for the type to the largest alignment which is ever used for any data
3213 type on the target machine you are compiling for. Doing this can often
3214 make copy operations more efficient, because the compiler can use
3215 whatever instructions copy the biggest chunks of memory when performing
3216 copies to or from the variables which have types that you have aligned
3219 In the example above, if the size of each @code{short} is 2 bytes, then
3220 the size of the entire @code{struct S} type is 6 bytes. The smallest
3221 power of two which is greater than or equal to that is 8, so the
3222 compiler sets the alignment for the entire @code{struct S} type to 8
3225 Note that although you can ask the compiler to select a time-efficient
3226 alignment for a given type and then declare only individual stand-alone
3227 objects of that type, the compiler's ability to select a time-efficient
3228 alignment is primarily useful only when you plan to create arrays of
3229 variables having the relevant (efficiently aligned) type. If you
3230 declare or use arrays of variables of an efficiently-aligned type, then
3231 it is likely that your program will also be doing pointer arithmetic (or
3232 subscripting, which amounts to the same thing) on pointers to the
3233 relevant type, and the code that the compiler generates for these
3234 pointer arithmetic operations will often be more efficient for
3235 efficiently-aligned types than for other types.
3237 The @code{aligned} attribute can only increase the alignment; but you
3238 can decrease it by specifying @code{packed} as well. See below.
3240 Note that the effectiveness of @code{aligned} attributes may be limited
3241 by inherent limitations in your linker. On many systems, the linker is
3242 only able to arrange for variables to be aligned up to a certain maximum
3243 alignment. (For some linkers, the maximum supported alignment may
3244 be very very small.) If your linker is only able to align variables
3245 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3246 in an @code{__attribute__} will still only provide you with 8 byte
3247 alignment. See your linker documentation for further information.
3250 This attribute, attached to @code{struct} or @code{union} type
3251 definition, specifies that each member of the structure or union is
3252 placed to minimize the memory required. When attached to an @code{enum}
3253 definition, it indicates that the smallest integral type should be used.
3255 @opindex fshort-enums
3256 Specifying this attribute for @code{struct} and @code{union} types is
3257 equivalent to specifying the @code{packed} attribute on each of the
3258 structure or union members. Specifying the @option{-fshort-enums}
3259 flag on the line is equivalent to specifying the @code{packed}
3260 attribute on all @code{enum} definitions.
3262 In the following example @code{struct my_packed_struct}'s members are
3263 packed closely together, but the internal layout of its @code{s} member
3264 is not packed---to do that, @code{struct my_unpacked_struct} would need to
3268 struct my_unpacked_struct
3274 struct __attribute__ ((__packed__)) my_packed_struct
3278 struct my_unpacked_struct s;
3282 You may only specify this attribute on the definition of a @code{enum},
3283 @code{struct} or @code{union}, not on a @code{typedef} which does not
3284 also define the enumerated type, structure or union.
3286 @item transparent_union
3287 This attribute, attached to a @code{union} type definition, indicates
3288 that any function parameter having that union type causes calls to that
3289 function to be treated in a special way.
3291 First, the argument corresponding to a transparent union type can be of
3292 any type in the union; no cast is required. Also, if the union contains
3293 a pointer type, the corresponding argument can be a null pointer
3294 constant or a void pointer expression; and if the union contains a void
3295 pointer type, the corresponding argument can be any pointer expression.
3296 If the union member type is a pointer, qualifiers like @code{const} on
3297 the referenced type must be respected, just as with normal pointer
3300 Second, the argument is passed to the function using the calling
3301 conventions of the first member of the transparent union, not the calling
3302 conventions of the union itself. All members of the union must have the
3303 same machine representation; this is necessary for this argument passing
3306 Transparent unions are designed for library functions that have multiple
3307 interfaces for compatibility reasons. For example, suppose the
3308 @code{wait} function must accept either a value of type @code{int *} to
3309 comply with Posix, or a value of type @code{union wait *} to comply with
3310 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3311 @code{wait} would accept both kinds of arguments, but it would also
3312 accept any other pointer type and this would make argument type checking
3313 less useful. Instead, @code{<sys/wait.h>} might define the interface
3321 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3323 pid_t wait (wait_status_ptr_t);
3326 This interface allows either @code{int *} or @code{union wait *}
3327 arguments to be passed, using the @code{int *} calling convention.
3328 The program can call @code{wait} with arguments of either type:
3331 int w1 () @{ int w; return wait (&w); @}
3332 int w2 () @{ union wait w; return wait (&w); @}
3335 With this interface, @code{wait}'s implementation might look like this:
3338 pid_t wait (wait_status_ptr_t p)
3340 return waitpid (-1, p.__ip, 0);
3345 When attached to a type (including a @code{union} or a @code{struct}),
3346 this attribute means that variables of that type are meant to appear
3347 possibly unused. GCC will not produce a warning for any variables of
3348 that type, even if the variable appears to do nothing. This is often
3349 the case with lock or thread classes, which are usually defined and then
3350 not referenced, but contain constructors and destructors that have
3351 nontrivial bookkeeping functions.
3354 The @code{deprecated} attribute results in a warning if the type
3355 is used anywhere in the source file. This is useful when identifying
3356 types that are expected to be removed in a future version of a program.
3357 If possible, the warning also includes the location of the declaration
3358 of the deprecated type, to enable users to easily find further
3359 information about why the type is deprecated, or what they should do
3360 instead. Note that the warnings only occur for uses and then only
3361 if the type is being applied to an identifier that itself is not being
3362 declared as deprecated.
3365 typedef int T1 __attribute__ ((deprecated));
3369 typedef T1 T3 __attribute__ ((deprecated));
3370 T3 z __attribute__ ((deprecated));
3373 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3374 warning is issued for line 4 because T2 is not explicitly
3375 deprecated. Line 5 has no warning because T3 is explicitly
3376 deprecated. Similarly for line 6.
3378 The @code{deprecated} attribute can also be used for functions and
3379 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3382 Accesses to objects with types with this attribute are not subjected to
3383 type-based alias analysis, but are instead assumed to be able to alias
3384 any other type of objects, just like the @code{char} type. See
3385 @option{-fstrict-aliasing} for more information on aliasing issues.
3390 typedef short __attribute__((__may_alias__)) short_a;
3396 short_a *b = (short_a *) &a;
3400 if (a == 0x12345678)
3407 If you replaced @code{short_a} with @code{short} in the variable
3408 declaration, the above program would abort when compiled with
3409 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3410 above in recent GCC versions.
3412 @subsection ARM Type Attributes
3414 On those ARM targets that support @code{dllimport} (such as Symbian
3415 OS), you can use the @code{notshared} attribute to indicate that the
3416 virtual table and other similar data for a class should not be
3417 exported from a DLL@. For example:
3420 class __declspec(notshared) C @{
3422 __declspec(dllimport) C();
3426 __declspec(dllexport)
3430 In this code, @code{C::C} is exported from the current DLL, but the
3431 virtual table for @code{C} is not exported. (You can use
3432 @code{__attribute__} instead of @code{__declspec} if you prefer, but
3433 most Symbian OS code uses @code{__declspec}.)
3435 @subsection i386 Type Attributes
3437 Two attributes are currently defined for i386 configurations:
3438 @code{ms_struct} and @code{gcc_struct}
3442 @cindex @code{ms_struct}
3443 @cindex @code{gcc_struct}
3445 If @code{packed} is used on a structure, or if bit-fields are used
3446 it may be that the Microsoft ABI packs them differently
3447 than GCC would normally pack them. Particularly when moving packed
3448 data between functions compiled with GCC and the native Microsoft compiler
3449 (either via function call or as data in a file), it may be necessary to access
3452 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3453 compilers to match the native Microsoft compiler.
3456 To specify multiple attributes, separate them by commas within the
3457 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3461 @section An Inline Function is As Fast As a Macro
3462 @cindex inline functions
3463 @cindex integrating function code
3465 @cindex macros, inline alternative
3467 By declaring a function @code{inline}, you can direct GCC to
3468 integrate that function's code into the code for its callers. This
3469 makes execution faster by eliminating the function-call overhead; in
3470 addition, if any of the actual argument values are constant, their known
3471 values may permit simplifications at compile time so that not all of the
3472 inline function's code needs to be included. The effect on code size is
3473 less predictable; object code may be larger or smaller with function
3474 inlining, depending on the particular case. Inlining of functions is an
3475 optimization and it really ``works'' only in optimizing compilation. If
3476 you don't use @option{-O}, no function is really inline.
3478 Inline functions are included in the ISO C99 standard, but there are
3479 currently substantial differences between what GCC implements and what
3480 the ISO C99 standard requires.
3482 To declare a function inline, use the @code{inline} keyword in its
3483 declaration, like this:
3493 (If you are writing a header file to be included in ISO C programs, write
3494 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3495 You can also make all ``simple enough'' functions inline with the option
3496 @option{-finline-functions}.
3499 Note that certain usages in a function definition can make it unsuitable
3500 for inline substitution. Among these usages are: use of varargs, use of
3501 alloca, use of variable sized data types (@pxref{Variable Length}),
3502 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3503 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3504 will warn when a function marked @code{inline} could not be substituted,
3505 and will give the reason for the failure.
3507 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3508 does not affect the linkage of the function.
3510 @cindex automatic @code{inline} for C++ member fns
3511 @cindex @code{inline} automatic for C++ member fns
3512 @cindex member fns, automatically @code{inline}
3513 @cindex C++ member fns, automatically @code{inline}
3514 @opindex fno-default-inline
3515 GCC automatically inlines member functions defined within the class
3516 body of C++ programs even if they are not explicitly declared
3517 @code{inline}. (You can override this with @option{-fno-default-inline};
3518 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3520 @cindex inline functions, omission of
3521 @opindex fkeep-inline-functions
3522 When a function is both inline and @code{static}, if all calls to the
3523 function are integrated into the caller, and the function's address is
3524 never used, then the function's own assembler code is never referenced.
3525 In this case, GCC does not actually output assembler code for the
3526 function, unless you specify the option @option{-fkeep-inline-functions}.
3527 Some calls cannot be integrated for various reasons (in particular,
3528 calls that precede the function's definition cannot be integrated, and
3529 neither can recursive calls within the definition). If there is a
3530 nonintegrated call, then the function is compiled to assembler code as
3531 usual. The function must also be compiled as usual if the program
3532 refers to its address, because that can't be inlined.
3534 @cindex non-static inline function
3535 When an inline function is not @code{static}, then the compiler must assume
3536 that there may be calls from other source files; since a global symbol can
3537 be defined only once in any program, the function must not be defined in
3538 the other source files, so the calls therein cannot be integrated.
3539 Therefore, a non-@code{static} inline function is always compiled on its
3540 own in the usual fashion.
3542 If you specify both @code{inline} and @code{extern} in the function
3543 definition, then the definition is used only for inlining. In no case
3544 is the function compiled on its own, not even if you refer to its
3545 address explicitly. Such an address becomes an external reference, as
3546 if you had only declared the function, and had not defined it.
3548 This combination of @code{inline} and @code{extern} has almost the
3549 effect of a macro. The way to use it is to put a function definition in
3550 a header file with these keywords, and put another copy of the
3551 definition (lacking @code{inline} and @code{extern}) in a library file.
3552 The definition in the header file will cause most calls to the function
3553 to be inlined. If any uses of the function remain, they will refer to
3554 the single copy in the library.
3556 Since GCC eventually will implement ISO C99 semantics for
3557 inline functions, it is best to use @code{static inline} only
3558 to guarantee compatibility. (The
3559 existing semantics will remain available when @option{-std=gnu89} is
3560 specified, but eventually the default will be @option{-std=gnu99} and
3561 that will implement the C99 semantics, though it does not do so yet.)
3563 GCC does not inline any functions when not optimizing unless you specify
3564 the @samp{always_inline} attribute for the function, like this:
3567 /* @r{Prototype.} */
3568 inline void foo (const char) __attribute__((always_inline));
3572 @section Assembler Instructions with C Expression Operands
3573 @cindex extended @code{asm}
3574 @cindex @code{asm} expressions
3575 @cindex assembler instructions
3578 In an assembler instruction using @code{asm}, you can specify the
3579 operands of the instruction using C expressions. This means you need not
3580 guess which registers or memory locations will contain the data you want
3583 You must specify an assembler instruction template much like what
3584 appears in a machine description, plus an operand constraint string for
3587 For example, here is how to use the 68881's @code{fsinx} instruction:
3590 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3594 Here @code{angle} is the C expression for the input operand while
3595 @code{result} is that of the output operand. Each has @samp{"f"} as its
3596 operand constraint, saying that a floating point register is required.
3597 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3598 output operands' constraints must use @samp{=}. The constraints use the
3599 same language used in the machine description (@pxref{Constraints}).
3601 Each operand is described by an operand-constraint string followed by
3602 the C expression in parentheses. A colon separates the assembler
3603 template from the first output operand and another separates the last
3604 output operand from the first input, if any. Commas separate the
3605 operands within each group. The total number of operands is currently
3606 limited to 30; this limitation may be lifted in some future version of
3609 If there are no output operands but there are input operands, you must
3610 place two consecutive colons surrounding the place where the output
3613 As of GCC version 3.1, it is also possible to specify input and output
3614 operands using symbolic names which can be referenced within the
3615 assembler code. These names are specified inside square brackets
3616 preceding the constraint string, and can be referenced inside the
3617 assembler code using @code{%[@var{name}]} instead of a percentage sign
3618 followed by the operand number. Using named operands the above example
3622 asm ("fsinx %[angle],%[output]"
3623 : [output] "=f" (result)
3624 : [angle] "f" (angle));
3628 Note that the symbolic operand names have no relation whatsoever to
3629 other C identifiers. You may use any name you like, even those of
3630 existing C symbols, but you must ensure that no two operands within the same
3631 assembler construct use the same symbolic name.
3633 Output operand expressions must be lvalues; the compiler can check this.
3634 The input operands need not be lvalues. The compiler cannot check
3635 whether the operands have data types that are reasonable for the
3636 instruction being executed. It does not parse the assembler instruction
3637 template and does not know what it means or even whether it is valid
3638 assembler input. The extended @code{asm} feature is most often used for
3639 machine instructions the compiler itself does not know exist. If
3640 the output expression cannot be directly addressed (for example, it is a
3641 bit-field), your constraint must allow a register. In that case, GCC
3642 will use the register as the output of the @code{asm}, and then store
3643 that register into the output.
3645 The ordinary output operands must be write-only; GCC will assume that
3646 the values in these operands before the instruction are dead and need
3647 not be generated. Extended asm supports input-output or read-write
3648 operands. Use the constraint character @samp{+} to indicate such an
3649 operand and list it with the output operands. You should only use
3650 read-write operands when the constraints for the operand (or the
3651 operand in which only some of the bits are to be changed) allow a
3654 You may, as an alternative, logically split its function into two
3655 separate operands, one input operand and one write-only output
3656 operand. The connection between them is expressed by constraints
3657 which say they need to be in the same location when the instruction
3658 executes. You can use the same C expression for both operands, or
3659 different expressions. For example, here we write the (fictitious)
3660 @samp{combine} instruction with @code{bar} as its read-only source
3661 operand and @code{foo} as its read-write destination:
3664 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3668 The constraint @samp{"0"} for operand 1 says that it must occupy the
3669 same location as operand 0. A number in constraint is allowed only in
3670 an input operand and it must refer to an output operand.
3672 Only a number in the constraint can guarantee that one operand will be in
3673 the same place as another. The mere fact that @code{foo} is the value
3674 of both operands is not enough to guarantee that they will be in the
3675 same place in the generated assembler code. The following would not
3679 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3682 Various optimizations or reloading could cause operands 0 and 1 to be in
3683 different registers; GCC knows no reason not to do so. For example, the
3684 compiler might find a copy of the value of @code{foo} in one register and
3685 use it for operand 1, but generate the output operand 0 in a different
3686 register (copying it afterward to @code{foo}'s own address). Of course,
3687 since the register for operand 1 is not even mentioned in the assembler
3688 code, the result will not work, but GCC can't tell that.
3690 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3691 the operand number for a matching constraint. For example:
3694 asm ("cmoveq %1,%2,%[result]"
3695 : [result] "=r"(result)
3696 : "r" (test), "r"(new), "[result]"(old));
3699 Sometimes you need to make an @code{asm} operand be a specific register,
3700 but there's no matching constraint letter for that register @emph{by
3701 itself}. To force the operand into that register, use a local variable
3702 for the operand and specify the register in the variable declaration.
3703 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
3704 register constraint letter that matches the register:
3707 register int *p1 asm ("r0") = @dots{};
3708 register int *p2 asm ("r1") = @dots{};
3709 register int *result asm ("r0");
3710 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3713 @anchor{Example of asm with clobbered asm reg}
3714 In the above example, beware that a register that is call-clobbered by
3715 the target ABI will be overwritten by any function call in the
3716 assignment, including library calls for arithmetic operators.
3717 Assuming it is a call-clobbered register, this may happen to @code{r0}
3718 above by the assignment to @code{p2}. If you have to use such a
3719 register, use temporary variables for expressions between the register
3724 register int *p1 asm ("r0") = @dots{};
3725 register int *p2 asm ("r1") = t1;
3726 register int *result asm ("r0");
3727 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3730 Some instructions clobber specific hard registers. To describe this,
3731 write a third colon after the input operands, followed by the names of
3732 the clobbered hard registers (given as strings). Here is a realistic
3733 example for the VAX:
3736 asm volatile ("movc3 %0,%1,%2"
3737 : /* @r{no outputs} */
3738 : "g" (from), "g" (to), "g" (count)
3739 : "r0", "r1", "r2", "r3", "r4", "r5");
3742 You may not write a clobber description in a way that overlaps with an
3743 input or output operand. For example, you may not have an operand
3744 describing a register class with one member if you mention that register
3745 in the clobber list. Variables declared to live in specific registers
3746 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
3747 have no part mentioned in the clobber description.
3748 There is no way for you to specify that an input
3749 operand is modified without also specifying it as an output
3750 operand. Note that if all the output operands you specify are for this
3751 purpose (and hence unused), you will then also need to specify
3752 @code{volatile} for the @code{asm} construct, as described below, to
3753 prevent GCC from deleting the @code{asm} statement as unused.
3755 If you refer to a particular hardware register from the assembler code,
3756 you will probably have to list the register after the third colon to
3757 tell the compiler the register's value is modified. In some assemblers,
3758 the register names begin with @samp{%}; to produce one @samp{%} in the
3759 assembler code, you must write @samp{%%} in the input.
3761 If your assembler instruction can alter the condition code register, add
3762 @samp{cc} to the list of clobbered registers. GCC on some machines
3763 represents the condition codes as a specific hardware register;
3764 @samp{cc} serves to name this register. On other machines, the
3765 condition code is handled differently, and specifying @samp{cc} has no
3766 effect. But it is valid no matter what the machine.
3768 If your assembler instructions access memory in an unpredictable
3769 fashion, add @samp{memory} to the list of clobbered registers. This
3770 will cause GCC to not keep memory values cached in registers across the
3771 assembler instruction and not optimize stores or loads to that memory.
3772 You will also want to add the @code{volatile} keyword if the memory
3773 affected is not listed in the inputs or outputs of the @code{asm}, as
3774 the @samp{memory} clobber does not count as a side-effect of the
3775 @code{asm}. If you know how large the accessed memory is, you can add
3776 it as input or output but if this is not known, you should add
3777 @samp{memory}. As an example, if you access ten bytes of a string, you
3778 can use a memory input like:
3781 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
3784 Note that in the following example the memory input is necessary,
3785 otherwise GCC might optimize the store to @code{x} away:
3792 asm ("magic stuff accessing an 'int' pointed to by '%1'"
3793 "=&d" (r) : "a" (y), "m" (*y));
3798 You can put multiple assembler instructions together in a single
3799 @code{asm} template, separated by the characters normally used in assembly
3800 code for the system. A combination that works in most places is a newline
3801 to break the line, plus a tab character to move to the instruction field
3802 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
3803 assembler allows semicolons as a line-breaking character. Note that some
3804 assembler dialects use semicolons to start a comment.
3805 The input operands are guaranteed not to use any of the clobbered
3806 registers, and neither will the output operands' addresses, so you can
3807 read and write the clobbered registers as many times as you like. Here
3808 is an example of multiple instructions in a template; it assumes the
3809 subroutine @code{_foo} accepts arguments in registers 9 and 10:
3812 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
3814 : "g" (from), "g" (to)
3818 Unless an output operand has the @samp{&} constraint modifier, GCC
3819 may allocate it in the same register as an unrelated input operand, on
3820 the assumption the inputs are consumed before the outputs are produced.
3821 This assumption may be false if the assembler code actually consists of
3822 more than one instruction. In such a case, use @samp{&} for each output
3823 operand that may not overlap an input. @xref{Modifiers}.
3825 If you want to test the condition code produced by an assembler
3826 instruction, you must include a branch and a label in the @code{asm}
3827 construct, as follows:
3830 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
3836 This assumes your assembler supports local labels, as the GNU assembler
3837 and most Unix assemblers do.
3839 Speaking of labels, jumps from one @code{asm} to another are not
3840 supported. The compiler's optimizers do not know about these jumps, and
3841 therefore they cannot take account of them when deciding how to
3844 @cindex macros containing @code{asm}
3845 Usually the most convenient way to use these @code{asm} instructions is to
3846 encapsulate them in macros that look like functions. For example,
3850 (@{ double __value, __arg = (x); \
3851 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
3856 Here the variable @code{__arg} is used to make sure that the instruction
3857 operates on a proper @code{double} value, and to accept only those
3858 arguments @code{x} which can convert automatically to a @code{double}.
3860 Another way to make sure the instruction operates on the correct data
3861 type is to use a cast in the @code{asm}. This is different from using a
3862 variable @code{__arg} in that it converts more different types. For
3863 example, if the desired type were @code{int}, casting the argument to
3864 @code{int} would accept a pointer with no complaint, while assigning the
3865 argument to an @code{int} variable named @code{__arg} would warn about
3866 using a pointer unless the caller explicitly casts it.
3868 If an @code{asm} has output operands, GCC assumes for optimization
3869 purposes the instruction has no side effects except to change the output
3870 operands. This does not mean instructions with a side effect cannot be
3871 used, but you must be careful, because the compiler may eliminate them
3872 if the output operands aren't used, or move them out of loops, or
3873 replace two with one if they constitute a common subexpression. Also,
3874 if your instruction does have a side effect on a variable that otherwise
3875 appears not to change, the old value of the variable may be reused later
3876 if it happens to be found in a register.
3878 You can prevent an @code{asm} instruction from being deleted
3879 by writing the keyword @code{volatile} after
3880 the @code{asm}. For example:
3883 #define get_and_set_priority(new) \
3885 asm volatile ("get_and_set_priority %0, %1" \
3886 : "=g" (__old) : "g" (new)); \
3891 The @code{volatile} keyword indicates that the instruction has
3892 important side-effects. GCC will not delete a volatile @code{asm} if
3893 it is reachable. (The instruction can still be deleted if GCC can
3894 prove that control-flow will never reach the location of the
3895 instruction.) Note that even a volatile @code{asm} instruction
3896 can be moved relative to other code, including across jump
3897 instructions. For example, on many targets there is a system
3898 register which can be set to control the rounding mode of
3899 floating point operations. You might try
3900 setting it with a volatile @code{asm}, like this PowerPC example:
3903 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
3908 This will not work reliably, as the compiler may move the addition back
3909 before the volatile @code{asm}. To make it work you need to add an
3910 artificial dependency to the @code{asm} referencing a variable in the code
3911 you don't want moved, for example:
3914 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
3918 Similarly, you can't expect a
3919 sequence of volatile @code{asm} instructions to remain perfectly
3920 consecutive. If you want consecutive output, use a single @code{asm}.
3921 Also, GCC will perform some optimizations across a volatile @code{asm}
3922 instruction; GCC does not ``forget everything'' when it encounters
3923 a volatile @code{asm} instruction the way some other compilers do.
3925 An @code{asm} instruction without any output operands will be treated
3926 identically to a volatile @code{asm} instruction.
3928 It is a natural idea to look for a way to give access to the condition
3929 code left by the assembler instruction. However, when we attempted to
3930 implement this, we found no way to make it work reliably. The problem
3931 is that output operands might need reloading, which would result in
3932 additional following ``store'' instructions. On most machines, these
3933 instructions would alter the condition code before there was time to
3934 test it. This problem doesn't arise for ordinary ``test'' and
3935 ``compare'' instructions because they don't have any output operands.
3937 For reasons similar to those described above, it is not possible to give
3938 an assembler instruction access to the condition code left by previous
3941 If you are writing a header file that should be includable in ISO C
3942 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
3945 @subsection Size of an @code{asm}
3947 Some targets require that GCC track the size of each instruction used in
3948 order to generate correct code. Because the final length of an
3949 @code{asm} is only known by the assembler, GCC must make an estimate as
3950 to how big it will be. The estimate is formed by counting the number of
3951 statements in the pattern of the @code{asm} and multiplying that by the
3952 length of the longest instruction on that processor. Statements in the
3953 @code{asm} are identified by newline characters and whatever statement
3954 separator characters are supported by the assembler; on most processors
3955 this is the `@code{;}' character.
3957 Normally, GCC's estimate is perfectly adequate to ensure that correct
3958 code is generated, but it is possible to confuse the compiler if you use
3959 pseudo instructions or assembler macros that expand into multiple real
3960 instructions or if you use assembler directives that expand to more
3961 space in the object file than would be needed for a single instruction.
3962 If this happens then the assembler will produce a diagnostic saying that
3963 a label is unreachable.
3965 @subsection i386 floating point asm operands
3967 There are several rules on the usage of stack-like regs in
3968 asm_operands insns. These rules apply only to the operands that are
3973 Given a set of input regs that die in an asm_operands, it is
3974 necessary to know which are implicitly popped by the asm, and
3975 which must be explicitly popped by gcc.
3977 An input reg that is implicitly popped by the asm must be
3978 explicitly clobbered, unless it is constrained to match an
3982 For any input reg that is implicitly popped by an asm, it is
3983 necessary to know how to adjust the stack to compensate for the pop.
3984 If any non-popped input is closer to the top of the reg-stack than
3985 the implicitly popped reg, it would not be possible to know what the
3986 stack looked like---it's not clear how the rest of the stack ``slides
3989 All implicitly popped input regs must be closer to the top of
3990 the reg-stack than any input that is not implicitly popped.
3992 It is possible that if an input dies in an insn, reload might
3993 use the input reg for an output reload. Consider this example:
3996 asm ("foo" : "=t" (a) : "f" (b));
3999 This asm says that input B is not popped by the asm, and that
4000 the asm pushes a result onto the reg-stack, i.e., the stack is one
4001 deeper after the asm than it was before. But, it is possible that
4002 reload will think that it can use the same reg for both the input and
4003 the output, if input B dies in this insn.
4005 If any input operand uses the @code{f} constraint, all output reg
4006 constraints must use the @code{&} earlyclobber.
4008 The asm above would be written as
4011 asm ("foo" : "=&t" (a) : "f" (b));
4015 Some operands need to be in particular places on the stack. All
4016 output operands fall in this category---there is no other way to
4017 know which regs the outputs appear in unless the user indicates
4018 this in the constraints.
4020 Output operands must specifically indicate which reg an output
4021 appears in after an asm. @code{=f} is not allowed: the operand
4022 constraints must select a class with a single reg.
4025 Output operands may not be ``inserted'' between existing stack regs.
4026 Since no 387 opcode uses a read/write operand, all output operands
4027 are dead before the asm_operands, and are pushed by the asm_operands.
4028 It makes no sense to push anywhere but the top of the reg-stack.
4030 Output operands must start at the top of the reg-stack: output
4031 operands may not ``skip'' a reg.
4034 Some asm statements may need extra stack space for internal
4035 calculations. This can be guaranteed by clobbering stack registers
4036 unrelated to the inputs and outputs.
4040 Here are a couple of reasonable asms to want to write. This asm
4041 takes one input, which is internally popped, and produces two outputs.
4044 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4047 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4048 and replaces them with one output. The user must code the @code{st(1)}
4049 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4052 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4058 @section Controlling Names Used in Assembler Code
4059 @cindex assembler names for identifiers
4060 @cindex names used in assembler code
4061 @cindex identifiers, names in assembler code
4063 You can specify the name to be used in the assembler code for a C
4064 function or variable by writing the @code{asm} (or @code{__asm__})
4065 keyword after the declarator as follows:
4068 int foo asm ("myfoo") = 2;
4072 This specifies that the name to be used for the variable @code{foo} in
4073 the assembler code should be @samp{myfoo} rather than the usual
4076 On systems where an underscore is normally prepended to the name of a C
4077 function or variable, this feature allows you to define names for the
4078 linker that do not start with an underscore.
4080 It does not make sense to use this feature with a non-static local
4081 variable since such variables do not have assembler names. If you are
4082 trying to put the variable in a particular register, see @ref{Explicit
4083 Reg Vars}. GCC presently accepts such code with a warning, but will
4084 probably be changed to issue an error, rather than a warning, in the
4087 You cannot use @code{asm} in this way in a function @emph{definition}; but
4088 you can get the same effect by writing a declaration for the function
4089 before its definition and putting @code{asm} there, like this:
4092 extern func () asm ("FUNC");
4099 It is up to you to make sure that the assembler names you choose do not
4100 conflict with any other assembler symbols. Also, you must not use a
4101 register name; that would produce completely invalid assembler code. GCC
4102 does not as yet have the ability to store static variables in registers.
4103 Perhaps that will be added.
4105 @node Explicit Reg Vars
4106 @section Variables in Specified Registers
4107 @cindex explicit register variables
4108 @cindex variables in specified registers
4109 @cindex specified registers
4110 @cindex registers, global allocation
4112 GNU C allows you to put a few global variables into specified hardware
4113 registers. You can also specify the register in which an ordinary
4114 register variable should be allocated.
4118 Global register variables reserve registers throughout the program.
4119 This may be useful in programs such as programming language
4120 interpreters which have a couple of global variables that are accessed
4124 Local register variables in specific registers do not reserve the
4125 registers, except at the point where they are used as input or output
4126 operands in an @code{asm} statement and the @code{asm} statement itself is
4127 not deleted. The compiler's data flow analysis is capable of determining
4128 where the specified registers contain live values, and where they are
4129 available for other uses. Stores into local register variables may be deleted
4130 when they appear to be dead according to dataflow analysis. References
4131 to local register variables may be deleted or moved or simplified.
4133 These local variables are sometimes convenient for use with the extended
4134 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4135 output of the assembler instruction directly into a particular register.
4136 (This will work provided the register you specify fits the constraints
4137 specified for that operand in the @code{asm}.)
4145 @node Global Reg Vars
4146 @subsection Defining Global Register Variables
4147 @cindex global register variables
4148 @cindex registers, global variables in
4150 You can define a global register variable in GNU C like this:
4153 register int *foo asm ("a5");
4157 Here @code{a5} is the name of the register which should be used. Choose a
4158 register which is normally saved and restored by function calls on your
4159 machine, so that library routines will not clobber it.
4161 Naturally the register name is cpu-dependent, so you would need to
4162 conditionalize your program according to cpu type. The register
4163 @code{a5} would be a good choice on a 68000 for a variable of pointer
4164 type. On machines with register windows, be sure to choose a ``global''
4165 register that is not affected magically by the function call mechanism.
4167 In addition, operating systems on one type of cpu may differ in how they
4168 name the registers; then you would need additional conditionals. For
4169 example, some 68000 operating systems call this register @code{%a5}.
4171 Eventually there may be a way of asking the compiler to choose a register
4172 automatically, but first we need to figure out how it should choose and
4173 how to enable you to guide the choice. No solution is evident.
4175 Defining a global register variable in a certain register reserves that
4176 register entirely for this use, at least within the current compilation.
4177 The register will not be allocated for any other purpose in the functions
4178 in the current compilation. The register will not be saved and restored by
4179 these functions. Stores into this register are never deleted even if they
4180 would appear to be dead, but references may be deleted or moved or
4183 It is not safe to access the global register variables from signal
4184 handlers, or from more than one thread of control, because the system
4185 library routines may temporarily use the register for other things (unless
4186 you recompile them specially for the task at hand).
4188 @cindex @code{qsort}, and global register variables
4189 It is not safe for one function that uses a global register variable to
4190 call another such function @code{foo} by way of a third function
4191 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4192 different source file in which the variable wasn't declared). This is
4193 because @code{lose} might save the register and put some other value there.
4194 For example, you can't expect a global register variable to be available in
4195 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4196 might have put something else in that register. (If you are prepared to
4197 recompile @code{qsort} with the same global register variable, you can
4198 solve this problem.)
4200 If you want to recompile @code{qsort} or other source files which do not
4201 actually use your global register variable, so that they will not use that
4202 register for any other purpose, then it suffices to specify the compiler
4203 option @option{-ffixed-@var{reg}}. You need not actually add a global
4204 register declaration to their source code.
4206 A function which can alter the value of a global register variable cannot
4207 safely be called from a function compiled without this variable, because it
4208 could clobber the value the caller expects to find there on return.
4209 Therefore, the function which is the entry point into the part of the
4210 program that uses the global register variable must explicitly save and
4211 restore the value which belongs to its caller.
4213 @cindex register variable after @code{longjmp}
4214 @cindex global register after @code{longjmp}
4215 @cindex value after @code{longjmp}
4218 On most machines, @code{longjmp} will restore to each global register
4219 variable the value it had at the time of the @code{setjmp}. On some
4220 machines, however, @code{longjmp} will not change the value of global
4221 register variables. To be portable, the function that called @code{setjmp}
4222 should make other arrangements to save the values of the global register
4223 variables, and to restore them in a @code{longjmp}. This way, the same
4224 thing will happen regardless of what @code{longjmp} does.
4226 All global register variable declarations must precede all function
4227 definitions. If such a declaration could appear after function
4228 definitions, the declaration would be too late to prevent the register from
4229 being used for other purposes in the preceding functions.
4231 Global register variables may not have initial values, because an
4232 executable file has no means to supply initial contents for a register.
4234 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4235 registers, but certain library functions, such as @code{getwd}, as well
4236 as the subroutines for division and remainder, modify g3 and g4. g1 and
4237 g2 are local temporaries.
4239 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4240 Of course, it will not do to use more than a few of those.
4242 @node Local Reg Vars
4243 @subsection Specifying Registers for Local Variables
4244 @cindex local variables, specifying registers
4245 @cindex specifying registers for local variables
4246 @cindex registers for local variables
4248 You can define a local register variable with a specified register
4252 register int *foo asm ("a5");
4256 Here @code{a5} is the name of the register which should be used. Note
4257 that this is the same syntax used for defining global register
4258 variables, but for a local variable it would appear within a function.
4260 Naturally the register name is cpu-dependent, but this is not a
4261 problem, since specific registers are most often useful with explicit
4262 assembler instructions (@pxref{Extended Asm}). Both of these things
4263 generally require that you conditionalize your program according to
4266 In addition, operating systems on one type of cpu may differ in how they
4267 name the registers; then you would need additional conditionals. For
4268 example, some 68000 operating systems call this register @code{%a5}.
4270 Defining such a register variable does not reserve the register; it
4271 remains available for other uses in places where flow control determines
4272 the variable's value is not live.
4274 This option does not guarantee that GCC will generate code that has
4275 this variable in the register you specify at all times. You may not
4276 code an explicit reference to this register in the @emph{assembler
4277 instruction template} part of an @code{asm} statement and assume it will
4278 always refer to this variable. However, using the variable as an
4279 @code{asm} @emph{operand} guarantees that the specified register is used
4282 Stores into local register variables may be deleted when they appear to be dead
4283 according to dataflow analysis. References to local register variables may
4284 be deleted or moved or simplified.
4286 As for global register variables, it's recommended that you choose a
4287 register which is normally saved and restored by function calls on
4288 your machine, so that library routines will not clobber it. A common
4289 pitfall is to initialize multiple call-clobbered registers with
4290 arbitrary expressions, where a function call or library call for an
4291 arithmetic operator will overwrite a register value from a previous
4292 assignment, for example @code{r0} below:
4294 register int *p1 asm ("r0") = @dots{};
4295 register int *p2 asm ("r1") = @dots{};
4297 In those cases, a solution is to use a temporary variable for
4298 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
4300 @node Alternate Keywords
4301 @section Alternate Keywords
4302 @cindex alternate keywords
4303 @cindex keywords, alternate
4305 @option{-ansi} and the various @option{-std} options disable certain
4306 keywords. This causes trouble when you want to use GNU C extensions, or
4307 a general-purpose header file that should be usable by all programs,
4308 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4309 @code{inline} are not available in programs compiled with
4310 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4311 program compiled with @option{-std=c99}). The ISO C99 keyword
4312 @code{restrict} is only available when @option{-std=gnu99} (which will
4313 eventually be the default) or @option{-std=c99} (or the equivalent
4314 @option{-std=iso9899:1999}) is used.
4316 The way to solve these problems is to put @samp{__} at the beginning and
4317 end of each problematical keyword. For example, use @code{__asm__}
4318 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4320 Other C compilers won't accept these alternative keywords; if you want to
4321 compile with another compiler, you can define the alternate keywords as
4322 macros to replace them with the customary keywords. It looks like this:
4330 @findex __extension__
4332 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4334 prevent such warnings within one expression by writing
4335 @code{__extension__} before the expression. @code{__extension__} has no
4336 effect aside from this.
4338 @node Incomplete Enums
4339 @section Incomplete @code{enum} Types
4341 You can define an @code{enum} tag without specifying its possible values.
4342 This results in an incomplete type, much like what you get if you write
4343 @code{struct foo} without describing the elements. A later declaration
4344 which does specify the possible values completes the type.
4346 You can't allocate variables or storage using the type while it is
4347 incomplete. However, you can work with pointers to that type.
4349 This extension may not be very useful, but it makes the handling of
4350 @code{enum} more consistent with the way @code{struct} and @code{union}
4353 This extension is not supported by GNU C++.
4355 @node Function Names
4356 @section Function Names as Strings
4357 @cindex @code{__func__} identifier
4358 @cindex @code{__FUNCTION__} identifier
4359 @cindex @code{__PRETTY_FUNCTION__} identifier
4361 GCC provides three magic variables which hold the name of the current
4362 function, as a string. The first of these is @code{__func__}, which
4363 is part of the C99 standard:
4366 The identifier @code{__func__} is implicitly declared by the translator
4367 as if, immediately following the opening brace of each function
4368 definition, the declaration
4371 static const char __func__[] = "function-name";
4374 appeared, where function-name is the name of the lexically-enclosing
4375 function. This name is the unadorned name of the function.
4378 @code{__FUNCTION__} is another name for @code{__func__}. Older
4379 versions of GCC recognize only this name. However, it is not
4380 standardized. For maximum portability, we recommend you use
4381 @code{__func__}, but provide a fallback definition with the
4385 #if __STDC_VERSION__ < 199901L
4387 # define __func__ __FUNCTION__
4389 # define __func__ "<unknown>"
4394 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4395 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4396 the type signature of the function as well as its bare name. For
4397 example, this program:
4401 extern int printf (char *, ...);
4408 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4409 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4427 __PRETTY_FUNCTION__ = void a::sub(int)
4430 These identifiers are not preprocessor macros. In GCC 3.3 and
4431 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4432 were treated as string literals; they could be used to initialize
4433 @code{char} arrays, and they could be concatenated with other string
4434 literals. GCC 3.4 and later treat them as variables, like
4435 @code{__func__}. In C++, @code{__FUNCTION__} and
4436 @code{__PRETTY_FUNCTION__} have always been variables.
4438 @node Return Address
4439 @section Getting the Return or Frame Address of a Function
4441 These functions may be used to get information about the callers of a
4444 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4445 This function returns the return address of the current function, or of
4446 one of its callers. The @var{level} argument is number of frames to
4447 scan up the call stack. A value of @code{0} yields the return address
4448 of the current function, a value of @code{1} yields the return address
4449 of the caller of the current function, and so forth. When inlining
4450 the expected behavior is that the function will return the address of
4451 the function that will be returned to. To work around this behavior use
4452 the @code{noinline} function attribute.
4454 The @var{level} argument must be a constant integer.
4456 On some machines it may be impossible to determine the return address of
4457 any function other than the current one; in such cases, or when the top
4458 of the stack has been reached, this function will return @code{0} or a
4459 random value. In addition, @code{__builtin_frame_address} may be used
4460 to determine if the top of the stack has been reached.
4462 This function should only be used with a nonzero argument for debugging
4466 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4467 This function is similar to @code{__builtin_return_address}, but it
4468 returns the address of the function frame rather than the return address
4469 of the function. Calling @code{__builtin_frame_address} with a value of
4470 @code{0} yields the frame address of the current function, a value of
4471 @code{1} yields the frame address of the caller of the current function,
4474 The frame is the area on the stack which holds local variables and saved
4475 registers. The frame address is normally the address of the first word
4476 pushed on to the stack by the function. However, the exact definition
4477 depends upon the processor and the calling convention. If the processor
4478 has a dedicated frame pointer register, and the function has a frame,
4479 then @code{__builtin_frame_address} will return the value of the frame
4482 On some machines it may be impossible to determine the frame address of
4483 any function other than the current one; in such cases, or when the top
4484 of the stack has been reached, this function will return @code{0} if
4485 the first frame pointer is properly initialized by the startup code.
4487 This function should only be used with a nonzero argument for debugging
4491 @node Vector Extensions
4492 @section Using vector instructions through built-in functions
4494 On some targets, the instruction set contains SIMD vector instructions that
4495 operate on multiple values contained in one large register at the same time.
4496 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4499 The first step in using these extensions is to provide the necessary data
4500 types. This should be done using an appropriate @code{typedef}:
4503 typedef int v4si __attribute__ ((vector_size (16)));
4506 The @code{int} type specifies the base type, while the attribute specifies
4507 the vector size for the variable, measured in bytes. For example, the
4508 declaration above causes the compiler to set the mode for the @code{v4si}
4509 type to be 16 bytes wide and divided into @code{int} sized units. For
4510 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4511 corresponding mode of @code{foo} will be @acronym{V4SI}.
4513 The @code{vector_size} attribute is only applicable to integral and
4514 float scalars, although arrays, pointers, and function return values
4515 are allowed in conjunction with this construct.
4517 All the basic integer types can be used as base types, both as signed
4518 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4519 @code{long long}. In addition, @code{float} and @code{double} can be
4520 used to build floating-point vector types.
4522 Specifying a combination that is not valid for the current architecture
4523 will cause GCC to synthesize the instructions using a narrower mode.
4524 For example, if you specify a variable of type @code{V4SI} and your
4525 architecture does not allow for this specific SIMD type, GCC will
4526 produce code that uses 4 @code{SIs}.
4528 The types defined in this manner can be used with a subset of normal C
4529 operations. Currently, GCC will allow using the following operators
4530 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4532 The operations behave like C++ @code{valarrays}. Addition is defined as
4533 the addition of the corresponding elements of the operands. For
4534 example, in the code below, each of the 4 elements in @var{a} will be
4535 added to the corresponding 4 elements in @var{b} and the resulting
4536 vector will be stored in @var{c}.
4539 typedef int v4si __attribute__ ((vector_size (16)));
4546 Subtraction, multiplication, division, and the logical operations
4547 operate in a similar manner. Likewise, the result of using the unary
4548 minus or complement operators on a vector type is a vector whose
4549 elements are the negative or complemented values of the corresponding
4550 elements in the operand.
4552 You can declare variables and use them in function calls and returns, as
4553 well as in assignments and some casts. You can specify a vector type as
4554 a return type for a function. Vector types can also be used as function
4555 arguments. It is possible to cast from one vector type to another,
4556 provided they are of the same size (in fact, you can also cast vectors
4557 to and from other datatypes of the same size).
4559 You cannot operate between vectors of different lengths or different
4560 signedness without a cast.
4562 A port that supports hardware vector operations, usually provides a set
4563 of built-in functions that can be used to operate on vectors. For
4564 example, a function to add two vectors and multiply the result by a
4565 third could look like this:
4568 v4si f (v4si a, v4si b, v4si c)
4570 v4si tmp = __builtin_addv4si (a, b);
4571 return __builtin_mulv4si (tmp, c);
4578 @findex __builtin_offsetof
4580 GCC implements for both C and C++ a syntactic extension to implement
4581 the @code{offsetof} macro.
4585 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
4587 offsetof_member_designator:
4589 | offsetof_member_designator "." @code{identifier}
4590 | offsetof_member_designator "[" @code{expr} "]"
4593 This extension is sufficient such that
4596 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
4599 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
4600 may be dependent. In either case, @var{member} may consist of a single
4601 identifier, or a sequence of member accesses and array references.
4603 @node Atomic Builtins
4604 @section Built-in functions for atomic memory access
4606 The following builtins are intended to be compatible with those described
4607 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
4608 section 7.4. As such, they depart from the normal GCC practice of using
4609 the ``__builtin_'' prefix, and further that they are overloaded such that
4610 they work on multiple types.
4612 The definition given in the Intel documentation allows only for the use of
4613 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
4614 counterparts. GCC will allow any integral scalar or pointer type that is
4615 1, 2, 4 or 8 bytes in length.
4617 Not all operations are supported by all target processors. If a particular
4618 operation cannot be implemented on the target processor, a warning will be
4619 generated and a call an external function will be generated. The external
4620 function will carry the same name as the builtin, with an additional suffix
4621 @samp{_@var{n}} where @var{n} is the size of the data type.
4623 @c ??? Should we have a mechanism to suppress this warning? This is almost
4624 @c useful for implementing the operation under the control of an external
4627 In most cases, these builtins are considered a @dfn{full barrier}. That is,
4628 no memory operand will be moved across the operation, either forward or
4629 backward. Further, instructions will be issued as necessary to prevent the
4630 processor from speculating loads across the operation and from queuing stores
4631 after the operation.
4633 All of the routines are are described in the Intel documentation to take
4634 ``an optional list of variables protected by the memory barrier''. It's
4635 not clear what is meant by that; it could mean that @emph{only} the
4636 following variables are protected, or it could mean that these variables
4637 should in addition be protected. At present GCC ignores this list and
4638 protects all variables which are globally accessible. If in the future
4639 we make some use of this list, an empty list will continue to mean all
4640 globally accessible variables.
4643 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
4644 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
4645 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
4646 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
4647 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
4648 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
4649 @findex __sync_fetch_and_add
4650 @findex __sync_fetch_and_sub
4651 @findex __sync_fetch_and_or
4652 @findex __sync_fetch_and_and
4653 @findex __sync_fetch_and_xor
4654 @findex __sync_fetch_and_nand
4655 These builtins perform the operation suggested by the name, and
4656 returns the value that had previously been in memory. That is,
4659 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
4660 @{ tmp = *ptr; *ptr = ~tmp & value; return tmp; @} // nand
4663 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
4664 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
4665 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
4666 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
4667 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
4668 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
4669 @findex __sync_add_and_fetch
4670 @findex __sync_sub_and_fetch
4671 @findex __sync_or_and_fetch
4672 @findex __sync_and_and_fetch
4673 @findex __sync_xor_and_fetch
4674 @findex __sync_nand_and_fetch
4675 These builtins perform the operation suggested by the name, and
4676 return the new value. That is,
4679 @{ *ptr @var{op}= value; return *ptr; @}
4680 @{ *ptr = ~*ptr & value; return *ptr; @} // nand
4683 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
4684 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
4685 @findex __sync_bool_compare_and_swap
4686 @findex __sync_val_compare_and_swap
4687 These builtins perform an atomic compare and swap. That is, if the current
4688 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
4691 The ``bool'' version returns true if the comparison is successful and
4692 @var{newval} was written. The ``val'' version returns the contents
4693 of @code{*@var{ptr}} before the operation.
4695 @item __sync_synchronize (...)
4696 @findex __sync_synchronize
4697 This builtin issues a full memory barrier.
4699 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
4700 @findex __sync_lock_test_and_set
4701 This builtin, as described by Intel, is not a traditional test-and-set
4702 operation, but rather an atomic exchange operation. It writes @var{value}
4703 into @code{*@var{ptr}}, and returns the previous contents of
4706 Many targets have only minimal support for such locks, and do not support
4707 a full exchange operation. In this case, a target may support reduced
4708 functionality here by which the @emph{only} valid value to store is the
4709 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
4710 is implementation defined.
4712 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
4713 This means that references after the builtin cannot move to (or be
4714 speculated to) before the builtin, but previous memory stores may not
4715 be globally visible yet, and previous memory loads may not yet be
4718 @item void __sync_lock_release (@var{type} *ptr, ...)
4719 @findex __sync_lock_release
4720 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
4721 Normally this means writing the constant 0 to @code{*@var{ptr}}.
4723 This builtin is not a full barrier, but rather a @dfn{release barrier}.
4724 This means that all previous memory stores are globally visible, and all
4725 previous memory loads have been satisfied, but following memory reads
4726 are not prevented from being speculated to before the barrier.
4729 @node Object Size Checking
4730 @section Object Size Checking Builtins
4731 @findex __builtin_object_size
4732 @findex __builtin___memcpy_chk
4733 @findex __builtin___mempcpy_chk
4734 @findex __builtin___memmove_chk
4735 @findex __builtin___memset_chk
4736 @findex __builtin___strcpy_chk
4737 @findex __builtin___stpcpy_chk
4738 @findex __builtin___strncpy_chk
4739 @findex __builtin___strcat_chk
4740 @findex __builtin___strncat_chk
4741 @findex __builtin___sprintf_chk
4742 @findex __builtin___snprintf_chk
4743 @findex __builtin___vsprintf_chk
4744 @findex __builtin___vsnprintf_chk
4745 @findex __builtin___printf_chk
4746 @findex __builtin___vprintf_chk
4747 @findex __builtin___fprintf_chk
4748 @findex __builtin___vfprintf_chk
4750 GCC implements a limited buffer overflow protection mechanism
4751 that can prevent some buffer overflow attacks.
4753 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
4754 is a built-in construct that returns a constant number of bytes from
4755 @var{ptr} to the end of the object @var{ptr} pointer points to
4756 (if known at compile time). @code{__builtin_object_size} never evaluates
4757 its arguments for side-effects. If there are any side-effects in them, it
4758 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
4759 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
4760 point to and all of them are known at compile time, the returned number
4761 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
4762 0 and minimum if non-zero. If it is not possible to determine which objects
4763 @var{ptr} points to at compile time, @code{__builtin_object_size} should
4764 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
4765 for @var{type} 2 or 3.
4767 @var{type} is an integer constant from 0 to 3. If the least significant
4768 bit is clear, objects are whole variables, if it is set, a closest
4769 surrounding subobject is considered the object a pointer points to.
4770 The second bit determines if maximum or minimum of remaining bytes
4774 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
4775 char *p = &var.buf1[1], *q = &var.b;
4777 /* Here the object p points to is var. */
4778 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
4779 /* The subobject p points to is var.buf1. */
4780 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
4781 /* The object q points to is var. */
4782 assert (__builtin_object_size (q, 0)
4783 == (char *) (&var + 1) - (char *) &var.b);
4784 /* The subobject q points to is var.b. */
4785 assert (__builtin_object_size (q, 1) == sizeof (var.b));
4789 There are built-in functions added for many common string operation
4790 functions, e.g. for @code{memcpy} @code{__builtin___memcpy_chk}
4791 built-in is provided. This built-in has an additional last argument,
4792 which is the number of bytes remaining in object the @var{dest}
4793 argument points to or @code{(size_t) -1} if the size is not known.
4795 The built-in functions are optimized into the normal string functions
4796 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
4797 it is known at compile time that the destination object will not
4798 be overflown. If the compiler can determine at compile time the
4799 object will be always overflown, it issues a warning.
4801 The intended use can be e.g.
4805 #define bos0(dest) __builtin_object_size (dest, 0)
4806 #define memcpy(dest, src, n) \
4807 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
4811 /* It is unknown what object p points to, so this is optimized
4812 into plain memcpy - no checking is possible. */
4813 memcpy (p, "abcde", n);
4814 /* Destination is known and length too. It is known at compile
4815 time there will be no overflow. */
4816 memcpy (&buf[5], "abcde", 5);
4817 /* Destination is known, but the length is not known at compile time.
4818 This will result in __memcpy_chk call that can check for overflow
4820 memcpy (&buf[5], "abcde", n);
4821 /* Destination is known and it is known at compile time there will
4822 be overflow. There will be a warning and __memcpy_chk call that
4823 will abort the program at runtime. */
4824 memcpy (&buf[6], "abcde", 5);
4827 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
4828 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
4829 @code{strcat} and @code{strncat}.
4831 There are also checking built-in functions for formatted output functions.
4833 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
4834 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
4835 const char *fmt, ...);
4836 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
4838 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
4839 const char *fmt, va_list ap);
4842 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
4843 etc. functions and can contain implementation specific flags on what
4844 additional security measures the checking function might take, such as
4845 handling @code{%n} differently.
4847 The @var{os} argument is the object size @var{s} points to, like in the
4848 other built-in functions. There is a small difference in the behaviour
4849 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
4850 optimized into the non-checking functions only if @var{flag} is 0, otherwise
4851 the checking function is called with @var{os} argument set to
4854 In addition to this, there are checking built-in functions
4855 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
4856 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
4857 These have just one additional argument, @var{flag}, right before
4858 format string @var{fmt}. If the compiler is able to optimize them to
4859 @code{fputc} etc. functions, it will, otherwise the checking function
4860 should be called and the @var{flag} argument passed to it.
4862 @node Other Builtins
4863 @section Other built-in functions provided by GCC
4864 @cindex built-in functions
4865 @findex __builtin_isgreater
4866 @findex __builtin_isgreaterequal
4867 @findex __builtin_isless
4868 @findex __builtin_islessequal
4869 @findex __builtin_islessgreater
4870 @findex __builtin_isunordered
4871 @findex __builtin_powi
4872 @findex __builtin_powif
4873 @findex __builtin_powil
5031 @findex fprintf_unlocked
5033 @findex fputs_unlocked
5143 @findex printf_unlocked
5172 @findex significandf
5173 @findex significandl
5244 GCC provides a large number of built-in functions other than the ones
5245 mentioned above. Some of these are for internal use in the processing
5246 of exceptions or variable-length argument lists and will not be
5247 documented here because they may change from time to time; we do not
5248 recommend general use of these functions.
5250 The remaining functions are provided for optimization purposes.
5252 @opindex fno-builtin
5253 GCC includes built-in versions of many of the functions in the standard
5254 C library. The versions prefixed with @code{__builtin_} will always be
5255 treated as having the same meaning as the C library function even if you
5256 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
5257 Many of these functions are only optimized in certain cases; if they are
5258 not optimized in a particular case, a call to the library function will
5263 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
5264 @option{-std=c99}), the functions
5265 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
5266 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
5267 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
5268 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
5269 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
5270 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
5271 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
5272 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
5273 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
5274 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
5275 @code{significandf}, @code{significandl}, @code{significand},
5276 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
5277 @code{stpncpy}, @code{strcasecmp}, @code{strdup}, @code{strfmon},
5278 @code{strncasecmp}, @code{strndup}, @code{toascii}, @code{y0f},
5279 @code{y0l}, @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf},
5280 @code{ynl} and @code{yn}
5281 may be handled as built-in functions.
5282 All these functions have corresponding versions
5283 prefixed with @code{__builtin_}, which may be used even in strict C89
5286 The ISO C99 functions
5287 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
5288 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
5289 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
5290 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
5291 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
5292 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
5293 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
5294 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
5295 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
5296 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
5297 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
5298 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
5299 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
5300 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
5301 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
5302 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
5303 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
5304 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
5305 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
5306 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
5307 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
5308 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
5309 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
5310 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
5311 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
5312 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
5313 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
5314 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
5315 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
5316 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
5317 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
5318 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
5319 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
5320 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
5321 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
5322 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
5323 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
5324 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
5325 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
5326 are handled as built-in functions
5327 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5329 There are also built-in versions of the ISO C99 functions
5330 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5331 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5332 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5333 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5334 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5335 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5336 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5337 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5338 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5339 that are recognized in any mode since ISO C90 reserves these names for
5340 the purpose to which ISO C99 puts them. All these functions have
5341 corresponding versions prefixed with @code{__builtin_}.
5343 The ISO C94 functions
5344 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
5345 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
5346 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
5348 are handled as built-in functions
5349 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5351 The ISO C90 functions
5352 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5353 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5354 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5355 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
5356 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
5357 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
5358 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
5359 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
5360 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
5361 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
5362 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
5363 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
5364 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
5365 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
5366 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
5367 @code{vprintf} and @code{vsprintf}
5368 are all recognized as built-in functions unless
5369 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5370 is specified for an individual function). All of these functions have
5371 corresponding versions prefixed with @code{__builtin_}.
5373 GCC provides built-in versions of the ISO C99 floating point comparison
5374 macros that avoid raising exceptions for unordered operands. They have
5375 the same names as the standard macros ( @code{isgreater},
5376 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5377 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5378 prefixed. We intend for a library implementor to be able to simply
5379 @code{#define} each standard macro to its built-in equivalent.
5381 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
5383 You can use the built-in function @code{__builtin_types_compatible_p} to
5384 determine whether two types are the same.
5386 This built-in function returns 1 if the unqualified versions of the
5387 types @var{type1} and @var{type2} (which are types, not expressions) are
5388 compatible, 0 otherwise. The result of this built-in function can be
5389 used in integer constant expressions.
5391 This built-in function ignores top level qualifiers (e.g., @code{const},
5392 @code{volatile}). For example, @code{int} is equivalent to @code{const
5395 The type @code{int[]} and @code{int[5]} are compatible. On the other
5396 hand, @code{int} and @code{char *} are not compatible, even if the size
5397 of their types, on the particular architecture are the same. Also, the
5398 amount of pointer indirection is taken into account when determining
5399 similarity. Consequently, @code{short *} is not similar to
5400 @code{short **}. Furthermore, two types that are typedefed are
5401 considered compatible if their underlying types are compatible.
5403 An @code{enum} type is not considered to be compatible with another
5404 @code{enum} type even if both are compatible with the same integer
5405 type; this is what the C standard specifies.
5406 For example, @code{enum @{foo, bar@}} is not similar to
5407 @code{enum @{hot, dog@}}.
5409 You would typically use this function in code whose execution varies
5410 depending on the arguments' types. For example:
5416 if (__builtin_types_compatible_p (typeof (x), long double)) \
5417 tmp = foo_long_double (tmp); \
5418 else if (__builtin_types_compatible_p (typeof (x), double)) \
5419 tmp = foo_double (tmp); \
5420 else if (__builtin_types_compatible_p (typeof (x), float)) \
5421 tmp = foo_float (tmp); \
5428 @emph{Note:} This construct is only available for C@.
5432 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
5434 You can use the built-in function @code{__builtin_choose_expr} to
5435 evaluate code depending on the value of a constant expression. This
5436 built-in function returns @var{exp1} if @var{const_exp}, which is a
5437 constant expression that must be able to be determined at compile time,
5438 is nonzero. Otherwise it returns 0.
5440 This built-in function is analogous to the @samp{? :} operator in C,
5441 except that the expression returned has its type unaltered by promotion
5442 rules. Also, the built-in function does not evaluate the expression
5443 that was not chosen. For example, if @var{const_exp} evaluates to true,
5444 @var{exp2} is not evaluated even if it has side-effects.
5446 This built-in function can return an lvalue if the chosen argument is an
5449 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5450 type. Similarly, if @var{exp2} is returned, its return type is the same
5457 __builtin_choose_expr ( \
5458 __builtin_types_compatible_p (typeof (x), double), \
5460 __builtin_choose_expr ( \
5461 __builtin_types_compatible_p (typeof (x), float), \
5463 /* @r{The void expression results in a compile-time error} \
5464 @r{when assigning the result to something.} */ \
5468 @emph{Note:} This construct is only available for C@. Furthermore, the
5469 unused expression (@var{exp1} or @var{exp2} depending on the value of
5470 @var{const_exp}) may still generate syntax errors. This may change in
5475 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5476 You can use the built-in function @code{__builtin_constant_p} to
5477 determine if a value is known to be constant at compile-time and hence
5478 that GCC can perform constant-folding on expressions involving that
5479 value. The argument of the function is the value to test. The function
5480 returns the integer 1 if the argument is known to be a compile-time
5481 constant and 0 if it is not known to be a compile-time constant. A
5482 return of 0 does not indicate that the value is @emph{not} a constant,
5483 but merely that GCC cannot prove it is a constant with the specified
5484 value of the @option{-O} option.
5486 You would typically use this function in an embedded application where
5487 memory was a critical resource. If you have some complex calculation,
5488 you may want it to be folded if it involves constants, but need to call
5489 a function if it does not. For example:
5492 #define Scale_Value(X) \
5493 (__builtin_constant_p (X) \
5494 ? ((X) * SCALE + OFFSET) : Scale (X))
5497 You may use this built-in function in either a macro or an inline
5498 function. However, if you use it in an inlined function and pass an
5499 argument of the function as the argument to the built-in, GCC will
5500 never return 1 when you call the inline function with a string constant
5501 or compound literal (@pxref{Compound Literals}) and will not return 1
5502 when you pass a constant numeric value to the inline function unless you
5503 specify the @option{-O} option.
5505 You may also use @code{__builtin_constant_p} in initializers for static
5506 data. For instance, you can write
5509 static const int table[] = @{
5510 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5516 This is an acceptable initializer even if @var{EXPRESSION} is not a
5517 constant expression. GCC must be more conservative about evaluating the
5518 built-in in this case, because it has no opportunity to perform
5521 Previous versions of GCC did not accept this built-in in data
5522 initializers. The earliest version where it is completely safe is
5526 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5527 @opindex fprofile-arcs
5528 You may use @code{__builtin_expect} to provide the compiler with
5529 branch prediction information. In general, you should prefer to
5530 use actual profile feedback for this (@option{-fprofile-arcs}), as
5531 programmers are notoriously bad at predicting how their programs
5532 actually perform. However, there are applications in which this
5533 data is hard to collect.
5535 The return value is the value of @var{exp}, which should be an
5536 integral expression. The value of @var{c} must be a compile-time
5537 constant. The semantics of the built-in are that it is expected
5538 that @var{exp} == @var{c}. For example:
5541 if (__builtin_expect (x, 0))
5546 would indicate that we do not expect to call @code{foo}, since
5547 we expect @code{x} to be zero. Since you are limited to integral
5548 expressions for @var{exp}, you should use constructions such as
5551 if (__builtin_expect (ptr != NULL, 1))
5556 when testing pointer or floating-point values.
5559 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5560 This function is used to minimize cache-miss latency by moving data into
5561 a cache before it is accessed.
5562 You can insert calls to @code{__builtin_prefetch} into code for which
5563 you know addresses of data in memory that is likely to be accessed soon.
5564 If the target supports them, data prefetch instructions will be generated.
5565 If the prefetch is done early enough before the access then the data will
5566 be in the cache by the time it is accessed.
5568 The value of @var{addr} is the address of the memory to prefetch.
5569 There are two optional arguments, @var{rw} and @var{locality}.
5570 The value of @var{rw} is a compile-time constant one or zero; one
5571 means that the prefetch is preparing for a write to the memory address
5572 and zero, the default, means that the prefetch is preparing for a read.
5573 The value @var{locality} must be a compile-time constant integer between
5574 zero and three. A value of zero means that the data has no temporal
5575 locality, so it need not be left in the cache after the access. A value
5576 of three means that the data has a high degree of temporal locality and
5577 should be left in all levels of cache possible. Values of one and two
5578 mean, respectively, a low or moderate degree of temporal locality. The
5582 for (i = 0; i < n; i++)
5585 __builtin_prefetch (&a[i+j], 1, 1);
5586 __builtin_prefetch (&b[i+j], 0, 1);
5591 Data prefetch does not generate faults if @var{addr} is invalid, but
5592 the address expression itself must be valid. For example, a prefetch
5593 of @code{p->next} will not fault if @code{p->next} is not a valid
5594 address, but evaluation will fault if @code{p} is not a valid address.
5596 If the target does not support data prefetch, the address expression
5597 is evaluated if it includes side effects but no other code is generated
5598 and GCC does not issue a warning.
5601 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5602 Returns a positive infinity, if supported by the floating-point format,
5603 else @code{DBL_MAX}. This function is suitable for implementing the
5604 ISO C macro @code{HUGE_VAL}.
5607 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5608 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5611 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5612 Similar to @code{__builtin_huge_val}, except the return
5613 type is @code{long double}.
5616 @deftypefn {Built-in Function} double __builtin_inf (void)
5617 Similar to @code{__builtin_huge_val}, except a warning is generated
5618 if the target floating-point format does not support infinities.
5621 @deftypefn {Built-in Function} float __builtin_inff (void)
5622 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5623 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5626 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5627 Similar to @code{__builtin_inf}, except the return
5628 type is @code{long double}.
5631 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5632 This is an implementation of the ISO C99 function @code{nan}.
5634 Since ISO C99 defines this function in terms of @code{strtod}, which we
5635 do not implement, a description of the parsing is in order. The string
5636 is parsed as by @code{strtol}; that is, the base is recognized by
5637 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5638 in the significand such that the least significant bit of the number
5639 is at the least significant bit of the significand. The number is
5640 truncated to fit the significand field provided. The significand is
5641 forced to be a quiet NaN@.
5643 This function, if given a string literal, is evaluated early enough
5644 that it is considered a compile-time constant.
5647 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
5648 Similar to @code{__builtin_nan}, except the return type is @code{float}.
5651 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
5652 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
5655 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
5656 Similar to @code{__builtin_nan}, except the significand is forced
5657 to be a signaling NaN@. The @code{nans} function is proposed by
5658 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
5661 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
5662 Similar to @code{__builtin_nans}, except the return type is @code{float}.
5665 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
5666 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
5669 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
5670 Returns one plus the index of the least significant 1-bit of @var{x}, or
5671 if @var{x} is zero, returns zero.
5674 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
5675 Returns the number of leading 0-bits in @var{x}, starting at the most
5676 significant bit position. If @var{x} is 0, the result is undefined.
5679 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
5680 Returns the number of trailing 0-bits in @var{x}, starting at the least
5681 significant bit position. If @var{x} is 0, the result is undefined.
5684 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
5685 Returns the number of 1-bits in @var{x}.
5688 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
5689 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
5693 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
5694 Similar to @code{__builtin_ffs}, except the argument type is
5695 @code{unsigned long}.
5698 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
5699 Similar to @code{__builtin_clz}, except the argument type is
5700 @code{unsigned long}.
5703 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
5704 Similar to @code{__builtin_ctz}, except the argument type is
5705 @code{unsigned long}.
5708 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
5709 Similar to @code{__builtin_popcount}, except the argument type is
5710 @code{unsigned long}.
5713 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
5714 Similar to @code{__builtin_parity}, except the argument type is
5715 @code{unsigned long}.
5718 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
5719 Similar to @code{__builtin_ffs}, except the argument type is
5720 @code{unsigned long long}.
5723 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
5724 Similar to @code{__builtin_clz}, except the argument type is
5725 @code{unsigned long long}.
5728 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
5729 Similar to @code{__builtin_ctz}, except the argument type is
5730 @code{unsigned long long}.
5733 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
5734 Similar to @code{__builtin_popcount}, except the argument type is
5735 @code{unsigned long long}.
5738 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
5739 Similar to @code{__builtin_parity}, except the argument type is
5740 @code{unsigned long long}.
5743 @deftypefn {Built-in Function} double __builtin_powi (double, int)
5744 Returns the first argument raised to the power of the second. Unlike the
5745 @code{pow} function no guarantees about precision and rounding are made.
5748 @deftypefn {Built-in Function} float __builtin_powif (float, int)
5749 Similar to @code{__builtin_powi}, except the argument and return types
5753 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
5754 Similar to @code{__builtin_powi}, except the argument and return types
5755 are @code{long double}.
5759 @node Target Builtins
5760 @section Built-in Functions Specific to Particular Target Machines
5762 On some target machines, GCC supports many built-in functions specific
5763 to those machines. Generally these generate calls to specific machine
5764 instructions, but allow the compiler to schedule those calls.
5767 * Alpha Built-in Functions::
5768 * ARM Built-in Functions::
5769 * Blackfin Built-in Functions::
5770 * FR-V Built-in Functions::
5771 * X86 Built-in Functions::
5772 * MIPS Paired-Single Support::
5773 * PowerPC AltiVec Built-in Functions::
5774 * SPARC VIS Built-in Functions::
5777 @node Alpha Built-in Functions
5778 @subsection Alpha Built-in Functions
5780 These built-in functions are available for the Alpha family of
5781 processors, depending on the command-line switches used.
5783 The following built-in functions are always available. They
5784 all generate the machine instruction that is part of the name.
5787 long __builtin_alpha_implver (void)
5788 long __builtin_alpha_rpcc (void)
5789 long __builtin_alpha_amask (long)
5790 long __builtin_alpha_cmpbge (long, long)
5791 long __builtin_alpha_extbl (long, long)
5792 long __builtin_alpha_extwl (long, long)
5793 long __builtin_alpha_extll (long, long)
5794 long __builtin_alpha_extql (long, long)
5795 long __builtin_alpha_extwh (long, long)
5796 long __builtin_alpha_extlh (long, long)
5797 long __builtin_alpha_extqh (long, long)
5798 long __builtin_alpha_insbl (long, long)
5799 long __builtin_alpha_inswl (long, long)
5800 long __builtin_alpha_insll (long, long)
5801 long __builtin_alpha_insql (long, long)
5802 long __builtin_alpha_inswh (long, long)
5803 long __builtin_alpha_inslh (long, long)
5804 long __builtin_alpha_insqh (long, long)
5805 long __builtin_alpha_mskbl (long, long)
5806 long __builtin_alpha_mskwl (long, long)
5807 long __builtin_alpha_mskll (long, long)
5808 long __builtin_alpha_mskql (long, long)
5809 long __builtin_alpha_mskwh (long, long)
5810 long __builtin_alpha_msklh (long, long)
5811 long __builtin_alpha_mskqh (long, long)
5812 long __builtin_alpha_umulh (long, long)
5813 long __builtin_alpha_zap (long, long)
5814 long __builtin_alpha_zapnot (long, long)
5817 The following built-in functions are always with @option{-mmax}
5818 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
5819 later. They all generate the machine instruction that is part
5823 long __builtin_alpha_pklb (long)
5824 long __builtin_alpha_pkwb (long)
5825 long __builtin_alpha_unpkbl (long)
5826 long __builtin_alpha_unpkbw (long)
5827 long __builtin_alpha_minub8 (long, long)
5828 long __builtin_alpha_minsb8 (long, long)
5829 long __builtin_alpha_minuw4 (long, long)
5830 long __builtin_alpha_minsw4 (long, long)
5831 long __builtin_alpha_maxub8 (long, long)
5832 long __builtin_alpha_maxsb8 (long, long)
5833 long __builtin_alpha_maxuw4 (long, long)
5834 long __builtin_alpha_maxsw4 (long, long)
5835 long __builtin_alpha_perr (long, long)
5838 The following built-in functions are always with @option{-mcix}
5839 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
5840 later. They all generate the machine instruction that is part
5844 long __builtin_alpha_cttz (long)
5845 long __builtin_alpha_ctlz (long)
5846 long __builtin_alpha_ctpop (long)
5849 The following builtins are available on systems that use the OSF/1
5850 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
5851 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
5852 @code{rdval} and @code{wrval}.
5855 void *__builtin_thread_pointer (void)
5856 void __builtin_set_thread_pointer (void *)
5859 @node ARM Built-in Functions
5860 @subsection ARM Built-in Functions
5862 These built-in functions are available for the ARM family of
5863 processors, when the @option{-mcpu=iwmmxt} switch is used:
5866 typedef int v2si __attribute__ ((vector_size (8)));
5867 typedef short v4hi __attribute__ ((vector_size (8)));
5868 typedef char v8qi __attribute__ ((vector_size (8)));
5870 int __builtin_arm_getwcx (int)
5871 void __builtin_arm_setwcx (int, int)
5872 int __builtin_arm_textrmsb (v8qi, int)
5873 int __builtin_arm_textrmsh (v4hi, int)
5874 int __builtin_arm_textrmsw (v2si, int)
5875 int __builtin_arm_textrmub (v8qi, int)
5876 int __builtin_arm_textrmuh (v4hi, int)
5877 int __builtin_arm_textrmuw (v2si, int)
5878 v8qi __builtin_arm_tinsrb (v8qi, int)
5879 v4hi __builtin_arm_tinsrh (v4hi, int)
5880 v2si __builtin_arm_tinsrw (v2si, int)
5881 long long __builtin_arm_tmia (long long, int, int)
5882 long long __builtin_arm_tmiabb (long long, int, int)
5883 long long __builtin_arm_tmiabt (long long, int, int)
5884 long long __builtin_arm_tmiaph (long long, int, int)
5885 long long __builtin_arm_tmiatb (long long, int, int)
5886 long long __builtin_arm_tmiatt (long long, int, int)
5887 int __builtin_arm_tmovmskb (v8qi)
5888 int __builtin_arm_tmovmskh (v4hi)
5889 int __builtin_arm_tmovmskw (v2si)
5890 long long __builtin_arm_waccb (v8qi)
5891 long long __builtin_arm_wacch (v4hi)
5892 long long __builtin_arm_waccw (v2si)
5893 v8qi __builtin_arm_waddb (v8qi, v8qi)
5894 v8qi __builtin_arm_waddbss (v8qi, v8qi)
5895 v8qi __builtin_arm_waddbus (v8qi, v8qi)
5896 v4hi __builtin_arm_waddh (v4hi, v4hi)
5897 v4hi __builtin_arm_waddhss (v4hi, v4hi)
5898 v4hi __builtin_arm_waddhus (v4hi, v4hi)
5899 v2si __builtin_arm_waddw (v2si, v2si)
5900 v2si __builtin_arm_waddwss (v2si, v2si)
5901 v2si __builtin_arm_waddwus (v2si, v2si)
5902 v8qi __builtin_arm_walign (v8qi, v8qi, int)
5903 long long __builtin_arm_wand(long long, long long)
5904 long long __builtin_arm_wandn (long long, long long)
5905 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
5906 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
5907 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
5908 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
5909 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
5910 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
5911 v2si __builtin_arm_wcmpeqw (v2si, v2si)
5912 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
5913 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
5914 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
5915 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
5916 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
5917 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
5918 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
5919 long long __builtin_arm_wmacsz (v4hi, v4hi)
5920 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
5921 long long __builtin_arm_wmacuz (v4hi, v4hi)
5922 v4hi __builtin_arm_wmadds (v4hi, v4hi)
5923 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
5924 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
5925 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
5926 v2si __builtin_arm_wmaxsw (v2si, v2si)
5927 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
5928 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
5929 v2si __builtin_arm_wmaxuw (v2si, v2si)
5930 v8qi __builtin_arm_wminsb (v8qi, v8qi)
5931 v4hi __builtin_arm_wminsh (v4hi, v4hi)
5932 v2si __builtin_arm_wminsw (v2si, v2si)
5933 v8qi __builtin_arm_wminub (v8qi, v8qi)
5934 v4hi __builtin_arm_wminuh (v4hi, v4hi)
5935 v2si __builtin_arm_wminuw (v2si, v2si)
5936 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
5937 v4hi __builtin_arm_wmulul (v4hi, v4hi)
5938 v4hi __builtin_arm_wmulum (v4hi, v4hi)
5939 long long __builtin_arm_wor (long long, long long)
5940 v2si __builtin_arm_wpackdss (long long, long long)
5941 v2si __builtin_arm_wpackdus (long long, long long)
5942 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
5943 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
5944 v4hi __builtin_arm_wpackwss (v2si, v2si)
5945 v4hi __builtin_arm_wpackwus (v2si, v2si)
5946 long long __builtin_arm_wrord (long long, long long)
5947 long long __builtin_arm_wrordi (long long, int)
5948 v4hi __builtin_arm_wrorh (v4hi, long long)
5949 v4hi __builtin_arm_wrorhi (v4hi, int)
5950 v2si __builtin_arm_wrorw (v2si, long long)
5951 v2si __builtin_arm_wrorwi (v2si, int)
5952 v2si __builtin_arm_wsadb (v8qi, v8qi)
5953 v2si __builtin_arm_wsadbz (v8qi, v8qi)
5954 v2si __builtin_arm_wsadh (v4hi, v4hi)
5955 v2si __builtin_arm_wsadhz (v4hi, v4hi)
5956 v4hi __builtin_arm_wshufh (v4hi, int)
5957 long long __builtin_arm_wslld (long long, long long)
5958 long long __builtin_arm_wslldi (long long, int)
5959 v4hi __builtin_arm_wsllh (v4hi, long long)
5960 v4hi __builtin_arm_wsllhi (v4hi, int)
5961 v2si __builtin_arm_wsllw (v2si, long long)
5962 v2si __builtin_arm_wsllwi (v2si, int)
5963 long long __builtin_arm_wsrad (long long, long long)
5964 long long __builtin_arm_wsradi (long long, int)
5965 v4hi __builtin_arm_wsrah (v4hi, long long)
5966 v4hi __builtin_arm_wsrahi (v4hi, int)
5967 v2si __builtin_arm_wsraw (v2si, long long)
5968 v2si __builtin_arm_wsrawi (v2si, int)
5969 long long __builtin_arm_wsrld (long long, long long)
5970 long long __builtin_arm_wsrldi (long long, int)
5971 v4hi __builtin_arm_wsrlh (v4hi, long long)
5972 v4hi __builtin_arm_wsrlhi (v4hi, int)
5973 v2si __builtin_arm_wsrlw (v2si, long long)
5974 v2si __builtin_arm_wsrlwi (v2si, int)
5975 v8qi __builtin_arm_wsubb (v8qi, v8qi)
5976 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
5977 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
5978 v4hi __builtin_arm_wsubh (v4hi, v4hi)
5979 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
5980 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
5981 v2si __builtin_arm_wsubw (v2si, v2si)
5982 v2si __builtin_arm_wsubwss (v2si, v2si)
5983 v2si __builtin_arm_wsubwus (v2si, v2si)
5984 v4hi __builtin_arm_wunpckehsb (v8qi)
5985 v2si __builtin_arm_wunpckehsh (v4hi)
5986 long long __builtin_arm_wunpckehsw (v2si)
5987 v4hi __builtin_arm_wunpckehub (v8qi)
5988 v2si __builtin_arm_wunpckehuh (v4hi)
5989 long long __builtin_arm_wunpckehuw (v2si)
5990 v4hi __builtin_arm_wunpckelsb (v8qi)
5991 v2si __builtin_arm_wunpckelsh (v4hi)
5992 long long __builtin_arm_wunpckelsw (v2si)
5993 v4hi __builtin_arm_wunpckelub (v8qi)
5994 v2si __builtin_arm_wunpckeluh (v4hi)
5995 long long __builtin_arm_wunpckeluw (v2si)
5996 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
5997 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
5998 v2si __builtin_arm_wunpckihw (v2si, v2si)
5999 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
6000 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
6001 v2si __builtin_arm_wunpckilw (v2si, v2si)
6002 long long __builtin_arm_wxor (long long, long long)
6003 long long __builtin_arm_wzero ()
6006 @node Blackfin Built-in Functions
6007 @subsection Blackfin Built-in Functions
6009 Currently, there are two Blackfin-specific built-in functions. These are
6010 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
6011 using inline assembly; by using these built-in functions the compiler can
6012 automatically add workarounds for hardware errata involving these
6013 instructions. These functions are named as follows:
6016 void __builtin_bfin_csync (void)
6017 void __builtin_bfin_ssync (void)
6020 @node FR-V Built-in Functions
6021 @subsection FR-V Built-in Functions
6023 GCC provides many FR-V-specific built-in functions. In general,
6024 these functions are intended to be compatible with those described
6025 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
6026 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
6027 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
6028 pointer rather than by value.
6030 Most of the functions are named after specific FR-V instructions.
6031 Such functions are said to be ``directly mapped'' and are summarized
6032 here in tabular form.
6036 * Directly-mapped Integer Functions::
6037 * Directly-mapped Media Functions::
6038 * Other Built-in Functions::
6041 @node Argument Types
6042 @subsubsection Argument Types
6044 The arguments to the built-in functions can be divided into three groups:
6045 register numbers, compile-time constants and run-time values. In order
6046 to make this classification clear at a glance, the arguments and return
6047 values are given the following pseudo types:
6049 @multitable @columnfractions .20 .30 .15 .35
6050 @item Pseudo type @tab Real C type @tab Constant? @tab Description
6051 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
6052 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
6053 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
6054 @item @code{uw2} @tab @code{unsigned long long} @tab No
6055 @tab an unsigned doubleword
6056 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
6057 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
6058 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
6059 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
6062 These pseudo types are not defined by GCC, they are simply a notational
6063 convenience used in this manual.
6065 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
6066 and @code{sw2} are evaluated at run time. They correspond to
6067 register operands in the underlying FR-V instructions.
6069 @code{const} arguments represent immediate operands in the underlying
6070 FR-V instructions. They must be compile-time constants.
6072 @code{acc} arguments are evaluated at compile time and specify the number
6073 of an accumulator register. For example, an @code{acc} argument of 2
6074 will select the ACC2 register.
6076 @code{iacc} arguments are similar to @code{acc} arguments but specify the
6077 number of an IACC register. See @pxref{Other Built-in Functions}
6080 @node Directly-mapped Integer Functions
6081 @subsubsection Directly-mapped Integer Functions
6083 The functions listed below map directly to FR-V I-type instructions.
6085 @multitable @columnfractions .45 .32 .23
6086 @item Function prototype @tab Example usage @tab Assembly output
6087 @item @code{sw1 __ADDSS (sw1, sw1)}
6088 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
6089 @tab @code{ADDSS @var{a},@var{b},@var{c}}
6090 @item @code{sw1 __SCAN (sw1, sw1)}
6091 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
6092 @tab @code{SCAN @var{a},@var{b},@var{c}}
6093 @item @code{sw1 __SCUTSS (sw1)}
6094 @tab @code{@var{b} = __SCUTSS (@var{a})}
6095 @tab @code{SCUTSS @var{a},@var{b}}
6096 @item @code{sw1 __SLASS (sw1, sw1)}
6097 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
6098 @tab @code{SLASS @var{a},@var{b},@var{c}}
6099 @item @code{void __SMASS (sw1, sw1)}
6100 @tab @code{__SMASS (@var{a}, @var{b})}
6101 @tab @code{SMASS @var{a},@var{b}}
6102 @item @code{void __SMSSS (sw1, sw1)}
6103 @tab @code{__SMSSS (@var{a}, @var{b})}
6104 @tab @code{SMSSS @var{a},@var{b}}
6105 @item @code{void __SMU (sw1, sw1)}
6106 @tab @code{__SMU (@var{a}, @var{b})}
6107 @tab @code{SMU @var{a},@var{b}}
6108 @item @code{sw2 __SMUL (sw1, sw1)}
6109 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
6110 @tab @code{SMUL @var{a},@var{b},@var{c}}
6111 @item @code{sw1 __SUBSS (sw1, sw1)}
6112 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
6113 @tab @code{SUBSS @var{a},@var{b},@var{c}}
6114 @item @code{uw2 __UMUL (uw1, uw1)}
6115 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
6116 @tab @code{UMUL @var{a},@var{b},@var{c}}
6119 @node Directly-mapped Media Functions
6120 @subsubsection Directly-mapped Media Functions
6122 The functions listed below map directly to FR-V M-type instructions.
6124 @multitable @columnfractions .45 .32 .23
6125 @item Function prototype @tab Example usage @tab Assembly output
6126 @item @code{uw1 __MABSHS (sw1)}
6127 @tab @code{@var{b} = __MABSHS (@var{a})}
6128 @tab @code{MABSHS @var{a},@var{b}}
6129 @item @code{void __MADDACCS (acc, acc)}
6130 @tab @code{__MADDACCS (@var{b}, @var{a})}
6131 @tab @code{MADDACCS @var{a},@var{b}}
6132 @item @code{sw1 __MADDHSS (sw1, sw1)}
6133 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
6134 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
6135 @item @code{uw1 __MADDHUS (uw1, uw1)}
6136 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
6137 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
6138 @item @code{uw1 __MAND (uw1, uw1)}
6139 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
6140 @tab @code{MAND @var{a},@var{b},@var{c}}
6141 @item @code{void __MASACCS (acc, acc)}
6142 @tab @code{__MASACCS (@var{b}, @var{a})}
6143 @tab @code{MASACCS @var{a},@var{b}}
6144 @item @code{uw1 __MAVEH (uw1, uw1)}
6145 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
6146 @tab @code{MAVEH @var{a},@var{b},@var{c}}
6147 @item @code{uw2 __MBTOH (uw1)}
6148 @tab @code{@var{b} = __MBTOH (@var{a})}
6149 @tab @code{MBTOH @var{a},@var{b}}
6150 @item @code{void __MBTOHE (uw1 *, uw1)}
6151 @tab @code{__MBTOHE (&@var{b}, @var{a})}
6152 @tab @code{MBTOHE @var{a},@var{b}}
6153 @item @code{void __MCLRACC (acc)}
6154 @tab @code{__MCLRACC (@var{a})}
6155 @tab @code{MCLRACC @var{a}}
6156 @item @code{void __MCLRACCA (void)}
6157 @tab @code{__MCLRACCA ()}
6158 @tab @code{MCLRACCA}
6159 @item @code{uw1 __Mcop1 (uw1, uw1)}
6160 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
6161 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
6162 @item @code{uw1 __Mcop2 (uw1, uw1)}
6163 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
6164 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
6165 @item @code{uw1 __MCPLHI (uw2, const)}
6166 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
6167 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
6168 @item @code{uw1 __MCPLI (uw2, const)}
6169 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
6170 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
6171 @item @code{void __MCPXIS (acc, sw1, sw1)}
6172 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
6173 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
6174 @item @code{void __MCPXIU (acc, uw1, uw1)}
6175 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
6176 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
6177 @item @code{void __MCPXRS (acc, sw1, sw1)}
6178 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
6179 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
6180 @item @code{void __MCPXRU (acc, uw1, uw1)}
6181 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
6182 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
6183 @item @code{uw1 __MCUT (acc, uw1)}
6184 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
6185 @tab @code{MCUT @var{a},@var{b},@var{c}}
6186 @item @code{uw1 __MCUTSS (acc, sw1)}
6187 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
6188 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
6189 @item @code{void __MDADDACCS (acc, acc)}
6190 @tab @code{__MDADDACCS (@var{b}, @var{a})}
6191 @tab @code{MDADDACCS @var{a},@var{b}}
6192 @item @code{void __MDASACCS (acc, acc)}
6193 @tab @code{__MDASACCS (@var{b}, @var{a})}
6194 @tab @code{MDASACCS @var{a},@var{b}}
6195 @item @code{uw2 __MDCUTSSI (acc, const)}
6196 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
6197 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
6198 @item @code{uw2 __MDPACKH (uw2, uw2)}
6199 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
6200 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
6201 @item @code{uw2 __MDROTLI (uw2, const)}
6202 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
6203 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
6204 @item @code{void __MDSUBACCS (acc, acc)}
6205 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
6206 @tab @code{MDSUBACCS @var{a},@var{b}}
6207 @item @code{void __MDUNPACKH (uw1 *, uw2)}
6208 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
6209 @tab @code{MDUNPACKH @var{a},@var{b}}
6210 @item @code{uw2 __MEXPDHD (uw1, const)}
6211 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
6212 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
6213 @item @code{uw1 __MEXPDHW (uw1, const)}
6214 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
6215 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
6216 @item @code{uw1 __MHDSETH (uw1, const)}
6217 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
6218 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
6219 @item @code{sw1 __MHDSETS (const)}
6220 @tab @code{@var{b} = __MHDSETS (@var{a})}
6221 @tab @code{MHDSETS #@var{a},@var{b}}
6222 @item @code{uw1 __MHSETHIH (uw1, const)}
6223 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
6224 @tab @code{MHSETHIH #@var{a},@var{b}}
6225 @item @code{sw1 __MHSETHIS (sw1, const)}
6226 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
6227 @tab @code{MHSETHIS #@var{a},@var{b}}
6228 @item @code{uw1 __MHSETLOH (uw1, const)}
6229 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
6230 @tab @code{MHSETLOH #@var{a},@var{b}}
6231 @item @code{sw1 __MHSETLOS (sw1, const)}
6232 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
6233 @tab @code{MHSETLOS #@var{a},@var{b}}
6234 @item @code{uw1 __MHTOB (uw2)}
6235 @tab @code{@var{b} = __MHTOB (@var{a})}
6236 @tab @code{MHTOB @var{a},@var{b}}
6237 @item @code{void __MMACHS (acc, sw1, sw1)}
6238 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
6239 @tab @code{MMACHS @var{a},@var{b},@var{c}}
6240 @item @code{void __MMACHU (acc, uw1, uw1)}
6241 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
6242 @tab @code{MMACHU @var{a},@var{b},@var{c}}
6243 @item @code{void __MMRDHS (acc, sw1, sw1)}
6244 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
6245 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
6246 @item @code{void __MMRDHU (acc, uw1, uw1)}
6247 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
6248 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
6249 @item @code{void __MMULHS (acc, sw1, sw1)}
6250 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
6251 @tab @code{MMULHS @var{a},@var{b},@var{c}}
6252 @item @code{void __MMULHU (acc, uw1, uw1)}
6253 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
6254 @tab @code{MMULHU @var{a},@var{b},@var{c}}
6255 @item @code{void __MMULXHS (acc, sw1, sw1)}
6256 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
6257 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
6258 @item @code{void __MMULXHU (acc, uw1, uw1)}
6259 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
6260 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
6261 @item @code{uw1 __MNOT (uw1)}
6262 @tab @code{@var{b} = __MNOT (@var{a})}
6263 @tab @code{MNOT @var{a},@var{b}}
6264 @item @code{uw1 __MOR (uw1, uw1)}
6265 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
6266 @tab @code{MOR @var{a},@var{b},@var{c}}
6267 @item @code{uw1 __MPACKH (uh, uh)}
6268 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
6269 @tab @code{MPACKH @var{a},@var{b},@var{c}}
6270 @item @code{sw2 __MQADDHSS (sw2, sw2)}
6271 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
6272 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
6273 @item @code{uw2 __MQADDHUS (uw2, uw2)}
6274 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
6275 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
6276 @item @code{void __MQCPXIS (acc, sw2, sw2)}
6277 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
6278 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
6279 @item @code{void __MQCPXIU (acc, uw2, uw2)}
6280 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
6281 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
6282 @item @code{void __MQCPXRS (acc, sw2, sw2)}
6283 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
6284 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
6285 @item @code{void __MQCPXRU (acc, uw2, uw2)}
6286 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
6287 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
6288 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
6289 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
6290 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
6291 @item @code{sw2 __MQLMTHS (sw2, sw2)}
6292 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
6293 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
6294 @item @code{void __MQMACHS (acc, sw2, sw2)}
6295 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
6296 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
6297 @item @code{void __MQMACHU (acc, uw2, uw2)}
6298 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
6299 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
6300 @item @code{void __MQMACXHS (acc, sw2, sw2)}
6301 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
6302 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
6303 @item @code{void __MQMULHS (acc, sw2, sw2)}
6304 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
6305 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
6306 @item @code{void __MQMULHU (acc, uw2, uw2)}
6307 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
6308 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
6309 @item @code{void __MQMULXHS (acc, sw2, sw2)}
6310 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
6311 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
6312 @item @code{void __MQMULXHU (acc, uw2, uw2)}
6313 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
6314 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
6315 @item @code{sw2 __MQSATHS (sw2, sw2)}
6316 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
6317 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
6318 @item @code{uw2 __MQSLLHI (uw2, int)}
6319 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
6320 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
6321 @item @code{sw2 __MQSRAHI (sw2, int)}
6322 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
6323 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
6324 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
6325 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
6326 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
6327 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
6328 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
6329 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
6330 @item @code{void __MQXMACHS (acc, sw2, sw2)}
6331 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
6332 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
6333 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
6334 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
6335 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
6336 @item @code{uw1 __MRDACC (acc)}
6337 @tab @code{@var{b} = __MRDACC (@var{a})}
6338 @tab @code{MRDACC @var{a},@var{b}}
6339 @item @code{uw1 __MRDACCG (acc)}
6340 @tab @code{@var{b} = __MRDACCG (@var{a})}
6341 @tab @code{MRDACCG @var{a},@var{b}}
6342 @item @code{uw1 __MROTLI (uw1, const)}
6343 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
6344 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
6345 @item @code{uw1 __MROTRI (uw1, const)}
6346 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
6347 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
6348 @item @code{sw1 __MSATHS (sw1, sw1)}
6349 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
6350 @tab @code{MSATHS @var{a},@var{b},@var{c}}
6351 @item @code{uw1 __MSATHU (uw1, uw1)}
6352 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
6353 @tab @code{MSATHU @var{a},@var{b},@var{c}}
6354 @item @code{uw1 __MSLLHI (uw1, const)}
6355 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
6356 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
6357 @item @code{sw1 __MSRAHI (sw1, const)}
6358 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
6359 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
6360 @item @code{uw1 __MSRLHI (uw1, const)}
6361 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
6362 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
6363 @item @code{void __MSUBACCS (acc, acc)}
6364 @tab @code{__MSUBACCS (@var{b}, @var{a})}
6365 @tab @code{MSUBACCS @var{a},@var{b}}
6366 @item @code{sw1 __MSUBHSS (sw1, sw1)}
6367 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
6368 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
6369 @item @code{uw1 __MSUBHUS (uw1, uw1)}
6370 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
6371 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
6372 @item @code{void __MTRAP (void)}
6373 @tab @code{__MTRAP ()}
6375 @item @code{uw2 __MUNPACKH (uw1)}
6376 @tab @code{@var{b} = __MUNPACKH (@var{a})}
6377 @tab @code{MUNPACKH @var{a},@var{b}}
6378 @item @code{uw1 __MWCUT (uw2, uw1)}
6379 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
6380 @tab @code{MWCUT @var{a},@var{b},@var{c}}
6381 @item @code{void __MWTACC (acc, uw1)}
6382 @tab @code{__MWTACC (@var{b}, @var{a})}
6383 @tab @code{MWTACC @var{a},@var{b}}
6384 @item @code{void __MWTACCG (acc, uw1)}
6385 @tab @code{__MWTACCG (@var{b}, @var{a})}
6386 @tab @code{MWTACCG @var{a},@var{b}}
6387 @item @code{uw1 __MXOR (uw1, uw1)}
6388 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
6389 @tab @code{MXOR @var{a},@var{b},@var{c}}
6392 @node Other Built-in Functions
6393 @subsubsection Other Built-in Functions
6395 This section describes built-in functions that are not named after
6396 a specific FR-V instruction.
6399 @item sw2 __IACCreadll (iacc @var{reg})
6400 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
6401 for future expansion and must be 0.
6403 @item sw1 __IACCreadl (iacc @var{reg})
6404 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
6405 Other values of @var{reg} are rejected as invalid.
6407 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
6408 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
6409 is reserved for future expansion and must be 0.
6411 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
6412 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
6413 is 1. Other values of @var{reg} are rejected as invalid.
6415 @item void __data_prefetch0 (const void *@var{x})
6416 Use the @code{dcpl} instruction to load the contents of address @var{x}
6417 into the data cache.
6419 @item void __data_prefetch (const void *@var{x})
6420 Use the @code{nldub} instruction to load the contents of address @var{x}
6421 into the data cache. The instruction will be issued in slot I1@.
6424 @node X86 Built-in Functions
6425 @subsection X86 Built-in Functions
6427 These built-in functions are available for the i386 and x86-64 family
6428 of computers, depending on the command-line switches used.
6430 The following machine modes are available for use with MMX built-in functions
6431 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
6432 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
6433 vector of eight 8-bit integers. Some of the built-in functions operate on
6434 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
6436 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
6437 of two 32-bit floating point values.
6439 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
6440 floating point values. Some instructions use a vector of four 32-bit
6441 integers, these use @code{V4SI}. Finally, some instructions operate on an
6442 entire vector register, interpreting it as a 128-bit integer, these use mode
6445 The following built-in functions are made available by @option{-mmmx}.
6446 All of them generate the machine instruction that is part of the name.
6449 v8qi __builtin_ia32_paddb (v8qi, v8qi)
6450 v4hi __builtin_ia32_paddw (v4hi, v4hi)
6451 v2si __builtin_ia32_paddd (v2si, v2si)
6452 v8qi __builtin_ia32_psubb (v8qi, v8qi)
6453 v4hi __builtin_ia32_psubw (v4hi, v4hi)
6454 v2si __builtin_ia32_psubd (v2si, v2si)
6455 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
6456 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
6457 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
6458 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
6459 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
6460 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
6461 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
6462 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
6463 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
6464 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
6465 di __builtin_ia32_pand (di, di)
6466 di __builtin_ia32_pandn (di,di)
6467 di __builtin_ia32_por (di, di)
6468 di __builtin_ia32_pxor (di, di)
6469 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
6470 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
6471 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
6472 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
6473 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
6474 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
6475 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
6476 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
6477 v2si __builtin_ia32_punpckhdq (v2si, v2si)
6478 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
6479 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
6480 v2si __builtin_ia32_punpckldq (v2si, v2si)
6481 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
6482 v4hi __builtin_ia32_packssdw (v2si, v2si)
6483 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
6486 The following built-in functions are made available either with
6487 @option{-msse}, or with a combination of @option{-m3dnow} and
6488 @option{-march=athlon}. All of them generate the machine
6489 instruction that is part of the name.
6492 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
6493 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
6494 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
6495 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
6496 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
6497 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
6498 v8qi __builtin_ia32_pminub (v8qi, v8qi)
6499 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
6500 int __builtin_ia32_pextrw (v4hi, int)
6501 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
6502 int __builtin_ia32_pmovmskb (v8qi)
6503 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
6504 void __builtin_ia32_movntq (di *, di)
6505 void __builtin_ia32_sfence (void)
6508 The following built-in functions are available when @option{-msse} is used.
6509 All of them generate the machine instruction that is part of the name.
6512 int __builtin_ia32_comieq (v4sf, v4sf)
6513 int __builtin_ia32_comineq (v4sf, v4sf)
6514 int __builtin_ia32_comilt (v4sf, v4sf)
6515 int __builtin_ia32_comile (v4sf, v4sf)
6516 int __builtin_ia32_comigt (v4sf, v4sf)
6517 int __builtin_ia32_comige (v4sf, v4sf)
6518 int __builtin_ia32_ucomieq (v4sf, v4sf)
6519 int __builtin_ia32_ucomineq (v4sf, v4sf)
6520 int __builtin_ia32_ucomilt (v4sf, v4sf)
6521 int __builtin_ia32_ucomile (v4sf, v4sf)
6522 int __builtin_ia32_ucomigt (v4sf, v4sf)
6523 int __builtin_ia32_ucomige (v4sf, v4sf)
6524 v4sf __builtin_ia32_addps (v4sf, v4sf)
6525 v4sf __builtin_ia32_subps (v4sf, v4sf)
6526 v4sf __builtin_ia32_mulps (v4sf, v4sf)
6527 v4sf __builtin_ia32_divps (v4sf, v4sf)
6528 v4sf __builtin_ia32_addss (v4sf, v4sf)
6529 v4sf __builtin_ia32_subss (v4sf, v4sf)
6530 v4sf __builtin_ia32_mulss (v4sf, v4sf)
6531 v4sf __builtin_ia32_divss (v4sf, v4sf)
6532 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
6533 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
6534 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
6535 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
6536 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
6537 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
6538 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
6539 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
6540 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
6541 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
6542 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
6543 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
6544 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
6545 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
6546 v4si __builtin_ia32_cmpless (v4sf, v4sf)
6547 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
6548 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
6549 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
6550 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
6551 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
6552 v4sf __builtin_ia32_maxps (v4sf, v4sf)
6553 v4sf __builtin_ia32_maxss (v4sf, v4sf)
6554 v4sf __builtin_ia32_minps (v4sf, v4sf)
6555 v4sf __builtin_ia32_minss (v4sf, v4sf)
6556 v4sf __builtin_ia32_andps (v4sf, v4sf)
6557 v4sf __builtin_ia32_andnps (v4sf, v4sf)
6558 v4sf __builtin_ia32_orps (v4sf, v4sf)
6559 v4sf __builtin_ia32_xorps (v4sf, v4sf)
6560 v4sf __builtin_ia32_movss (v4sf, v4sf)
6561 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
6562 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
6563 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
6564 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
6565 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
6566 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
6567 v2si __builtin_ia32_cvtps2pi (v4sf)
6568 int __builtin_ia32_cvtss2si (v4sf)
6569 v2si __builtin_ia32_cvttps2pi (v4sf)
6570 int __builtin_ia32_cvttss2si (v4sf)
6571 v4sf __builtin_ia32_rcpps (v4sf)
6572 v4sf __builtin_ia32_rsqrtps (v4sf)
6573 v4sf __builtin_ia32_sqrtps (v4sf)
6574 v4sf __builtin_ia32_rcpss (v4sf)
6575 v4sf __builtin_ia32_rsqrtss (v4sf)
6576 v4sf __builtin_ia32_sqrtss (v4sf)
6577 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
6578 void __builtin_ia32_movntps (float *, v4sf)
6579 int __builtin_ia32_movmskps (v4sf)
6582 The following built-in functions are available when @option{-msse} is used.
6585 @item v4sf __builtin_ia32_loadaps (float *)
6586 Generates the @code{movaps} machine instruction as a load from memory.
6587 @item void __builtin_ia32_storeaps (float *, v4sf)
6588 Generates the @code{movaps} machine instruction as a store to memory.
6589 @item v4sf __builtin_ia32_loadups (float *)
6590 Generates the @code{movups} machine instruction as a load from memory.
6591 @item void __builtin_ia32_storeups (float *, v4sf)
6592 Generates the @code{movups} machine instruction as a store to memory.
6593 @item v4sf __builtin_ia32_loadsss (float *)
6594 Generates the @code{movss} machine instruction as a load from memory.
6595 @item void __builtin_ia32_storess (float *, v4sf)
6596 Generates the @code{movss} machine instruction as a store to memory.
6597 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
6598 Generates the @code{movhps} machine instruction as a load from memory.
6599 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
6600 Generates the @code{movlps} machine instruction as a load from memory
6601 @item void __builtin_ia32_storehps (v4sf, v2si *)
6602 Generates the @code{movhps} machine instruction as a store to memory.
6603 @item void __builtin_ia32_storelps (v4sf, v2si *)
6604 Generates the @code{movlps} machine instruction as a store to memory.
6607 The following built-in functions are available when @option{-msse3} is used.
6608 All of them generate the machine instruction that is part of the name.
6611 v2df __builtin_ia32_addsubpd (v2df, v2df)
6612 v2df __builtin_ia32_addsubps (v2df, v2df)
6613 v2df __builtin_ia32_haddpd (v2df, v2df)
6614 v2df __builtin_ia32_haddps (v2df, v2df)
6615 v2df __builtin_ia32_hsubpd (v2df, v2df)
6616 v2df __builtin_ia32_hsubps (v2df, v2df)
6617 v16qi __builtin_ia32_lddqu (char const *)
6618 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
6619 v2df __builtin_ia32_movddup (v2df)
6620 v4sf __builtin_ia32_movshdup (v4sf)
6621 v4sf __builtin_ia32_movsldup (v4sf)
6622 void __builtin_ia32_mwait (unsigned int, unsigned int)
6625 The following built-in functions are available when @option{-msse3} is used.
6628 @item v2df __builtin_ia32_loadddup (double const *)
6629 Generates the @code{movddup} machine instruction as a load from memory.
6632 The following built-in functions are available when @option{-m3dnow} is used.
6633 All of them generate the machine instruction that is part of the name.
6636 void __builtin_ia32_femms (void)
6637 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
6638 v2si __builtin_ia32_pf2id (v2sf)
6639 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
6640 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
6641 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
6642 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
6643 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
6644 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
6645 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
6646 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
6647 v2sf __builtin_ia32_pfrcp (v2sf)
6648 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
6649 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
6650 v2sf __builtin_ia32_pfrsqrt (v2sf)
6651 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
6652 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
6653 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
6654 v2sf __builtin_ia32_pi2fd (v2si)
6655 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
6658 The following built-in functions are available when both @option{-m3dnow}
6659 and @option{-march=athlon} are used. All of them generate the machine
6660 instruction that is part of the name.
6663 v2si __builtin_ia32_pf2iw (v2sf)
6664 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
6665 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
6666 v2sf __builtin_ia32_pi2fw (v2si)
6667 v2sf __builtin_ia32_pswapdsf (v2sf)
6668 v2si __builtin_ia32_pswapdsi (v2si)
6671 @node MIPS Paired-Single Support
6672 @subsection MIPS Paired-Single Support
6674 The MIPS64 architecture includes a number of instructions that
6675 operate on pairs of single-precision floating-point values.
6676 Each pair is packed into a 64-bit floating-point register,
6677 with one element being designated the ``upper half'' and
6678 the other being designated the ``lower half''.
6680 GCC supports paired-single operations using both the generic
6681 vector extensions (@pxref{Vector Extensions}) and a collection of
6682 MIPS-specific built-in functions. Both kinds of support are
6683 enabled by the @option{-mpaired-single} command-line option.
6685 The vector type associated with paired-single values is usually
6686 called @code{v2sf}. It can be defined in C as follows:
6689 typedef float v2sf __attribute__ ((vector_size (8)));
6692 @code{v2sf} values are initialized in the same way as aggregates.
6696 v2sf a = @{1.5, 9.1@};
6699 b = (v2sf) @{e, f@};
6702 @emph{Note:} The CPU's endianness determines which value is stored in
6703 the upper half of a register and which value is stored in the lower half.
6704 On little-endian targets, the first value is the lower one and the second
6705 value is the upper one. The opposite order applies to big-endian targets.
6706 For example, the code above will set the lower half of @code{a} to
6707 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
6710 * Paired-Single Arithmetic::
6711 * Paired-Single Built-in Functions::
6712 * MIPS-3D Built-in Functions::
6715 @node Paired-Single Arithmetic
6716 @subsubsection Paired-Single Arithmetic
6718 The table below lists the @code{v2sf} operations for which hardware
6719 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
6720 values and @code{x} is an integral value.
6722 @multitable @columnfractions .50 .50
6723 @item C code @tab MIPS instruction
6724 @item @code{a + b} @tab @code{add.ps}
6725 @item @code{a - b} @tab @code{sub.ps}
6726 @item @code{-a} @tab @code{neg.ps}
6727 @item @code{a * b} @tab @code{mul.ps}
6728 @item @code{a * b + c} @tab @code{madd.ps}
6729 @item @code{a * b - c} @tab @code{msub.ps}
6730 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
6731 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
6732 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
6735 Note that the multiply-accumulate instructions can be disabled
6736 using the command-line option @code{-mno-fused-madd}.
6738 @node Paired-Single Built-in Functions
6739 @subsubsection Paired-Single Built-in Functions
6741 The following paired-single functions map directly to a particular
6742 MIPS instruction. Please refer to the architecture specification
6743 for details on what each instruction does.
6746 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
6747 Pair lower lower (@code{pll.ps}).
6749 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
6750 Pair upper lower (@code{pul.ps}).
6752 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
6753 Pair lower upper (@code{plu.ps}).
6755 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
6756 Pair upper upper (@code{puu.ps}).
6758 @item v2sf __builtin_mips_cvt_ps_s (float, float)
6759 Convert pair to paired single (@code{cvt.ps.s}).
6761 @item float __builtin_mips_cvt_s_pl (v2sf)
6762 Convert pair lower to single (@code{cvt.s.pl}).
6764 @item float __builtin_mips_cvt_s_pu (v2sf)
6765 Convert pair upper to single (@code{cvt.s.pu}).
6767 @item v2sf __builtin_mips_abs_ps (v2sf)
6768 Absolute value (@code{abs.ps}).
6770 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
6771 Align variable (@code{alnv.ps}).
6773 @emph{Note:} The value of the third parameter must be 0 or 4
6774 modulo 8, otherwise the result will be unpredictable. Please read the
6775 instruction description for details.
6778 The following multi-instruction functions are also available.
6779 In each case, @var{cond} can be any of the 16 floating-point conditions:
6780 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
6781 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
6782 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
6785 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6786 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6787 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
6788 @code{movt.ps}/@code{movf.ps}).
6790 The @code{movt} functions return the value @var{x} computed by:
6793 c.@var{cond}.ps @var{cc},@var{a},@var{b}
6794 mov.ps @var{x},@var{c}
6795 movt.ps @var{x},@var{d},@var{cc}
6798 The @code{movf} functions are similar but use @code{movf.ps} instead
6801 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6802 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6803 Comparison of two paired-single values (@code{c.@var{cond}.ps},
6804 @code{bc1t}/@code{bc1f}).
6806 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
6807 and return either the upper or lower half of the result. For example:
6811 if (__builtin_mips_upper_c_eq_ps (a, b))
6812 upper_halves_are_equal ();
6814 upper_halves_are_unequal ();
6816 if (__builtin_mips_lower_c_eq_ps (a, b))
6817 lower_halves_are_equal ();
6819 lower_halves_are_unequal ();
6823 @node MIPS-3D Built-in Functions
6824 @subsubsection MIPS-3D Built-in Functions
6826 The MIPS-3D Application-Specific Extension (ASE) includes additional
6827 paired-single instructions that are designed to improve the performance
6828 of 3D graphics operations. Support for these instructions is controlled
6829 by the @option{-mips3d} command-line option.
6831 The functions listed below map directly to a particular MIPS-3D
6832 instruction. Please refer to the architecture specification for
6833 more details on what each instruction does.
6836 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
6837 Reduction add (@code{addr.ps}).
6839 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
6840 Reduction multiply (@code{mulr.ps}).
6842 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
6843 Convert paired single to paired word (@code{cvt.pw.ps}).
6845 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
6846 Convert paired word to paired single (@code{cvt.ps.pw}).
6848 @item float __builtin_mips_recip1_s (float)
6849 @itemx double __builtin_mips_recip1_d (double)
6850 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
6851 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
6853 @item float __builtin_mips_recip2_s (float, float)
6854 @itemx double __builtin_mips_recip2_d (double, double)
6855 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
6856 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
6858 @item float __builtin_mips_rsqrt1_s (float)
6859 @itemx double __builtin_mips_rsqrt1_d (double)
6860 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
6861 Reduced precision reciprocal square root (sequence step 1)
6862 (@code{rsqrt1.@var{fmt}}).
6864 @item float __builtin_mips_rsqrt2_s (float, float)
6865 @itemx double __builtin_mips_rsqrt2_d (double, double)
6866 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
6867 Reduced precision reciprocal square root (sequence step 2)
6868 (@code{rsqrt2.@var{fmt}}).
6871 The following multi-instruction functions are also available.
6872 In each case, @var{cond} can be any of the 16 floating-point conditions:
6873 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
6874 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
6875 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
6878 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
6879 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
6880 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
6881 @code{bc1t}/@code{bc1f}).
6883 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
6884 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
6889 if (__builtin_mips_cabs_eq_s (a, b))
6895 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6896 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6897 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
6898 @code{bc1t}/@code{bc1f}).
6900 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
6901 and return either the upper or lower half of the result. For example:
6905 if (__builtin_mips_upper_cabs_eq_ps (a, b))
6906 upper_halves_are_equal ();
6908 upper_halves_are_unequal ();
6910 if (__builtin_mips_lower_cabs_eq_ps (a, b))
6911 lower_halves_are_equal ();
6913 lower_halves_are_unequal ();
6916 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6917 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6918 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
6919 @code{movt.ps}/@code{movf.ps}).
6921 The @code{movt} functions return the value @var{x} computed by:
6924 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
6925 mov.ps @var{x},@var{c}
6926 movt.ps @var{x},@var{d},@var{cc}
6929 The @code{movf} functions are similar but use @code{movf.ps} instead
6932 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6933 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6934 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6935 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6936 Comparison of two paired-single values
6937 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
6938 @code{bc1any2t}/@code{bc1any2f}).
6940 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
6941 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
6942 result is true and the @code{all} forms return true if both results are true.
6947 if (__builtin_mips_any_c_eq_ps (a, b))
6952 if (__builtin_mips_all_c_eq_ps (a, b))
6958 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6959 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6960 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6961 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6962 Comparison of four paired-single values
6963 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
6964 @code{bc1any4t}/@code{bc1any4f}).
6966 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
6967 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
6968 The @code{any} forms return true if any of the four results are true
6969 and the @code{all} forms return true if all four results are true.
6974 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
6979 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
6986 @node PowerPC AltiVec Built-in Functions
6987 @subsection PowerPC AltiVec Built-in Functions
6989 GCC provides an interface for the PowerPC family of processors to access
6990 the AltiVec operations described in Motorola's AltiVec Programming
6991 Interface Manual. The interface is made available by including
6992 @code{<altivec.h>} and using @option{-maltivec} and
6993 @option{-mabi=altivec}. The interface supports the following vector
6997 vector unsigned char
7001 vector unsigned short
7012 GCC's implementation of the high-level language interface available from
7013 C and C++ code differs from Motorola's documentation in several ways.
7018 A vector constant is a list of constant expressions within curly braces.
7021 A vector initializer requires no cast if the vector constant is of the
7022 same type as the variable it is initializing.
7025 If @code{signed} or @code{unsigned} is omitted, the signedness of the
7026 vector type is the default signedness of the base type. The default
7027 varies depending on the operating system, so a portable program should
7028 always specify the signedness.
7031 Compiling with @option{-maltivec} adds keywords @code{__vector},
7032 @code{__pixel}, and @code{__bool}. Macros @option{vector},
7033 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
7037 GCC allows using a @code{typedef} name as the type specifier for a
7041 For C, overloaded functions are implemented with macros so the following
7045 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
7048 Since @code{vec_add} is a macro, the vector constant in the example
7049 is treated as four separate arguments. Wrap the entire argument in
7050 parentheses for this to work.
7053 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
7054 Internally, GCC uses built-in functions to achieve the functionality in
7055 the aforementioned header file, but they are not supported and are
7056 subject to change without notice.
7058 The following interfaces are supported for the generic and specific
7059 AltiVec operations and the AltiVec predicates. In cases where there
7060 is a direct mapping between generic and specific operations, only the
7061 generic names are shown here, although the specific operations can also
7064 Arguments that are documented as @code{const int} require literal
7065 integral values within the range required for that operation.
7068 vector signed char vec_abs (vector signed char);
7069 vector signed short vec_abs (vector signed short);
7070 vector signed int vec_abs (vector signed int);
7071 vector float vec_abs (vector float);
7073 vector signed char vec_abss (vector signed char);
7074 vector signed short vec_abss (vector signed short);
7075 vector signed int vec_abss (vector signed int);
7077 vector signed char vec_add (vector bool char, vector signed char);
7078 vector signed char vec_add (vector signed char, vector bool char);
7079 vector signed char vec_add (vector signed char, vector signed char);
7080 vector unsigned char vec_add (vector bool char, vector unsigned char);
7081 vector unsigned char vec_add (vector unsigned char, vector bool char);
7082 vector unsigned char vec_add (vector unsigned char,
7083 vector unsigned char);
7084 vector signed short vec_add (vector bool short, vector signed short);
7085 vector signed short vec_add (vector signed short, vector bool short);
7086 vector signed short vec_add (vector signed short, vector signed short);
7087 vector unsigned short vec_add (vector bool short,
7088 vector unsigned short);
7089 vector unsigned short vec_add (vector unsigned short,
7091 vector unsigned short vec_add (vector unsigned short,
7092 vector unsigned short);
7093 vector signed int vec_add (vector bool int, vector signed int);
7094 vector signed int vec_add (vector signed int, vector bool int);
7095 vector signed int vec_add (vector signed int, vector signed int);
7096 vector unsigned int vec_add (vector bool int, vector unsigned int);
7097 vector unsigned int vec_add (vector unsigned int, vector bool int);
7098 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
7099 vector float vec_add (vector float, vector float);
7101 vector float vec_vaddfp (vector float, vector float);
7103 vector signed int vec_vadduwm (vector bool int, vector signed int);
7104 vector signed int vec_vadduwm (vector signed int, vector bool int);
7105 vector signed int vec_vadduwm (vector signed int, vector signed int);
7106 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
7107 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
7108 vector unsigned int vec_vadduwm (vector unsigned int,
7109 vector unsigned int);
7111 vector signed short vec_vadduhm (vector bool short,
7112 vector signed short);
7113 vector signed short vec_vadduhm (vector signed short,
7115 vector signed short vec_vadduhm (vector signed short,
7116 vector signed short);
7117 vector unsigned short vec_vadduhm (vector bool short,
7118 vector unsigned short);
7119 vector unsigned short vec_vadduhm (vector unsigned short,
7121 vector unsigned short vec_vadduhm (vector unsigned short,
7122 vector unsigned short);
7124 vector signed char vec_vaddubm (vector bool char, vector signed char);
7125 vector signed char vec_vaddubm (vector signed char, vector bool char);
7126 vector signed char vec_vaddubm (vector signed char, vector signed char);
7127 vector unsigned char vec_vaddubm (vector bool char,
7128 vector unsigned char);
7129 vector unsigned char vec_vaddubm (vector unsigned char,
7131 vector unsigned char vec_vaddubm (vector unsigned char,
7132 vector unsigned char);
7134 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
7136 vector unsigned char vec_adds (vector bool char, vector unsigned char);
7137 vector unsigned char vec_adds (vector unsigned char, vector bool char);
7138 vector unsigned char vec_adds (vector unsigned char,
7139 vector unsigned char);
7140 vector signed char vec_adds (vector bool char, vector signed char);
7141 vector signed char vec_adds (vector signed char, vector bool char);
7142 vector signed char vec_adds (vector signed char, vector signed char);
7143 vector unsigned short vec_adds (vector bool short,
7144 vector unsigned short);
7145 vector unsigned short vec_adds (vector unsigned short,
7147 vector unsigned short vec_adds (vector unsigned short,
7148 vector unsigned short);
7149 vector signed short vec_adds (vector bool short, vector signed short);
7150 vector signed short vec_adds (vector signed short, vector bool short);
7151 vector signed short vec_adds (vector signed short, vector signed short);
7152 vector unsigned int vec_adds (vector bool int, vector unsigned int);
7153 vector unsigned int vec_adds (vector unsigned int, vector bool int);
7154 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
7155 vector signed int vec_adds (vector bool int, vector signed int);
7156 vector signed int vec_adds (vector signed int, vector bool int);
7157 vector signed int vec_adds (vector signed int, vector signed int);
7159 vector signed int vec_vaddsws (vector bool int, vector signed int);
7160 vector signed int vec_vaddsws (vector signed int, vector bool int);
7161 vector signed int vec_vaddsws (vector signed int, vector signed int);
7163 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
7164 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
7165 vector unsigned int vec_vadduws (vector unsigned int,
7166 vector unsigned int);
7168 vector signed short vec_vaddshs (vector bool short,
7169 vector signed short);
7170 vector signed short vec_vaddshs (vector signed short,
7172 vector signed short vec_vaddshs (vector signed short,
7173 vector signed short);
7175 vector unsigned short vec_vadduhs (vector bool short,
7176 vector unsigned short);
7177 vector unsigned short vec_vadduhs (vector unsigned short,
7179 vector unsigned short vec_vadduhs (vector unsigned short,
7180 vector unsigned short);
7182 vector signed char vec_vaddsbs (vector bool char, vector signed char);
7183 vector signed char vec_vaddsbs (vector signed char, vector bool char);
7184 vector signed char vec_vaddsbs (vector signed char, vector signed char);
7186 vector unsigned char vec_vaddubs (vector bool char,
7187 vector unsigned char);
7188 vector unsigned char vec_vaddubs (vector unsigned char,
7190 vector unsigned char vec_vaddubs (vector unsigned char,
7191 vector unsigned char);
7193 vector float vec_and (vector float, vector float);
7194 vector float vec_and (vector float, vector bool int);
7195 vector float vec_and (vector bool int, vector float);
7196 vector bool int vec_and (vector bool int, vector bool int);
7197 vector signed int vec_and (vector bool int, vector signed int);
7198 vector signed int vec_and (vector signed int, vector bool int);
7199 vector signed int vec_and (vector signed int, vector signed int);
7200 vector unsigned int vec_and (vector bool int, vector unsigned int);
7201 vector unsigned int vec_and (vector unsigned int, vector bool int);
7202 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
7203 vector bool short vec_and (vector bool short, vector bool short);
7204 vector signed short vec_and (vector bool short, vector signed short);
7205 vector signed short vec_and (vector signed short, vector bool short);
7206 vector signed short vec_and (vector signed short, vector signed short);
7207 vector unsigned short vec_and (vector bool short,
7208 vector unsigned short);
7209 vector unsigned short vec_and (vector unsigned short,
7211 vector unsigned short vec_and (vector unsigned short,
7212 vector unsigned short);
7213 vector signed char vec_and (vector bool char, vector signed char);
7214 vector bool char vec_and (vector bool char, vector bool char);
7215 vector signed char vec_and (vector signed char, vector bool char);
7216 vector signed char vec_and (vector signed char, vector signed char);
7217 vector unsigned char vec_and (vector bool char, vector unsigned char);
7218 vector unsigned char vec_and (vector unsigned char, vector bool char);
7219 vector unsigned char vec_and (vector unsigned char,
7220 vector unsigned char);
7222 vector float vec_andc (vector float, vector float);
7223 vector float vec_andc (vector float, vector bool int);
7224 vector float vec_andc (vector bool int, vector float);
7225 vector bool int vec_andc (vector bool int, vector bool int);
7226 vector signed int vec_andc (vector bool int, vector signed int);
7227 vector signed int vec_andc (vector signed int, vector bool int);
7228 vector signed int vec_andc (vector signed int, vector signed int);
7229 vector unsigned int vec_andc (vector bool int, vector unsigned int);
7230 vector unsigned int vec_andc (vector unsigned int, vector bool int);
7231 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
7232 vector bool short vec_andc (vector bool short, vector bool short);
7233 vector signed short vec_andc (vector bool short, vector signed short);
7234 vector signed short vec_andc (vector signed short, vector bool short);
7235 vector signed short vec_andc (vector signed short, vector signed short);
7236 vector unsigned short vec_andc (vector bool short,
7237 vector unsigned short);
7238 vector unsigned short vec_andc (vector unsigned short,
7240 vector unsigned short vec_andc (vector unsigned short,
7241 vector unsigned short);
7242 vector signed char vec_andc (vector bool char, vector signed char);
7243 vector bool char vec_andc (vector bool char, vector bool char);
7244 vector signed char vec_andc (vector signed char, vector bool char);
7245 vector signed char vec_andc (vector signed char, vector signed char);
7246 vector unsigned char vec_andc (vector bool char, vector unsigned char);
7247 vector unsigned char vec_andc (vector unsigned char, vector bool char);
7248 vector unsigned char vec_andc (vector unsigned char,
7249 vector unsigned char);
7251 vector unsigned char vec_avg (vector unsigned char,
7252 vector unsigned char);
7253 vector signed char vec_avg (vector signed char, vector signed char);
7254 vector unsigned short vec_avg (vector unsigned short,
7255 vector unsigned short);
7256 vector signed short vec_avg (vector signed short, vector signed short);
7257 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
7258 vector signed int vec_avg (vector signed int, vector signed int);
7260 vector signed int vec_vavgsw (vector signed int, vector signed int);
7262 vector unsigned int vec_vavguw (vector unsigned int,
7263 vector unsigned int);
7265 vector signed short vec_vavgsh (vector signed short,
7266 vector signed short);
7268 vector unsigned short vec_vavguh (vector unsigned short,
7269 vector unsigned short);
7271 vector signed char vec_vavgsb (vector signed char, vector signed char);
7273 vector unsigned char vec_vavgub (vector unsigned char,
7274 vector unsigned char);
7276 vector float vec_ceil (vector float);
7278 vector signed int vec_cmpb (vector float, vector float);
7280 vector bool char vec_cmpeq (vector signed char, vector signed char);
7281 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
7282 vector bool short vec_cmpeq (vector signed short, vector signed short);
7283 vector bool short vec_cmpeq (vector unsigned short,
7284 vector unsigned short);
7285 vector bool int vec_cmpeq (vector signed int, vector signed int);
7286 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
7287 vector bool int vec_cmpeq (vector float, vector float);
7289 vector bool int vec_vcmpeqfp (vector float, vector float);
7291 vector bool int vec_vcmpequw (vector signed int, vector signed int);
7292 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
7294 vector bool short vec_vcmpequh (vector signed short,
7295 vector signed short);
7296 vector bool short vec_vcmpequh (vector unsigned short,
7297 vector unsigned short);
7299 vector bool char vec_vcmpequb (vector signed char, vector signed char);
7300 vector bool char vec_vcmpequb (vector unsigned char,
7301 vector unsigned char);
7303 vector bool int vec_cmpge (vector float, vector float);
7305 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
7306 vector bool char vec_cmpgt (vector signed char, vector signed char);
7307 vector bool short vec_cmpgt (vector unsigned short,
7308 vector unsigned short);
7309 vector bool short vec_cmpgt (vector signed short, vector signed short);
7310 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
7311 vector bool int vec_cmpgt (vector signed int, vector signed int);
7312 vector bool int vec_cmpgt (vector float, vector float);
7314 vector bool int vec_vcmpgtfp (vector float, vector float);
7316 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
7318 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
7320 vector bool short vec_vcmpgtsh (vector signed short,
7321 vector signed short);
7323 vector bool short vec_vcmpgtuh (vector unsigned short,
7324 vector unsigned short);
7326 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
7328 vector bool char vec_vcmpgtub (vector unsigned char,
7329 vector unsigned char);
7331 vector bool int vec_cmple (vector float, vector float);
7333 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
7334 vector bool char vec_cmplt (vector signed char, vector signed char);
7335 vector bool short vec_cmplt (vector unsigned short,
7336 vector unsigned short);
7337 vector bool short vec_cmplt (vector signed short, vector signed short);
7338 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
7339 vector bool int vec_cmplt (vector signed int, vector signed int);
7340 vector bool int vec_cmplt (vector float, vector float);
7342 vector float vec_ctf (vector unsigned int, const int);
7343 vector float vec_ctf (vector signed int, const int);
7345 vector float vec_vcfsx (vector signed int, const int);
7347 vector float vec_vcfux (vector unsigned int, const int);
7349 vector signed int vec_cts (vector float, const int);
7351 vector unsigned int vec_ctu (vector float, const int);
7353 void vec_dss (const int);
7355 void vec_dssall (void);
7357 void vec_dst (const vector unsigned char *, int, const int);
7358 void vec_dst (const vector signed char *, int, const int);
7359 void vec_dst (const vector bool char *, int, const int);
7360 void vec_dst (const vector unsigned short *, int, const int);
7361 void vec_dst (const vector signed short *, int, const int);
7362 void vec_dst (const vector bool short *, int, const int);
7363 void vec_dst (const vector pixel *, int, const int);
7364 void vec_dst (const vector unsigned int *, int, const int);
7365 void vec_dst (const vector signed int *, int, const int);
7366 void vec_dst (const vector bool int *, int, const int);
7367 void vec_dst (const vector float *, int, const int);
7368 void vec_dst (const unsigned char *, int, const int);
7369 void vec_dst (const signed char *, int, const int);
7370 void vec_dst (const unsigned short *, int, const int);
7371 void vec_dst (const short *, int, const int);
7372 void vec_dst (const unsigned int *, int, const int);
7373 void vec_dst (const int *, int, const int);
7374 void vec_dst (const unsigned long *, int, const int);
7375 void vec_dst (const long *, int, const int);
7376 void vec_dst (const float *, int, const int);
7378 void vec_dstst (const vector unsigned char *, int, const int);
7379 void vec_dstst (const vector signed char *, int, const int);
7380 void vec_dstst (const vector bool char *, int, const int);
7381 void vec_dstst (const vector unsigned short *, int, const int);
7382 void vec_dstst (const vector signed short *, int, const int);
7383 void vec_dstst (const vector bool short *, int, const int);
7384 void vec_dstst (const vector pixel *, int, const int);
7385 void vec_dstst (const vector unsigned int *, int, const int);
7386 void vec_dstst (const vector signed int *, int, const int);
7387 void vec_dstst (const vector bool int *, int, const int);
7388 void vec_dstst (const vector float *, int, const int);
7389 void vec_dstst (const unsigned char *, int, const int);
7390 void vec_dstst (const signed char *, int, const int);
7391 void vec_dstst (const unsigned short *, int, const int);
7392 void vec_dstst (const short *, int, const int);
7393 void vec_dstst (const unsigned int *, int, const int);
7394 void vec_dstst (const int *, int, const int);
7395 void vec_dstst (const unsigned long *, int, const int);
7396 void vec_dstst (const long *, int, const int);
7397 void vec_dstst (const float *, int, const int);
7399 void vec_dststt (const vector unsigned char *, int, const int);
7400 void vec_dststt (const vector signed char *, int, const int);
7401 void vec_dststt (const vector bool char *, int, const int);
7402 void vec_dststt (const vector unsigned short *, int, const int);
7403 void vec_dststt (const vector signed short *, int, const int);
7404 void vec_dststt (const vector bool short *, int, const int);
7405 void vec_dststt (const vector pixel *, int, const int);
7406 void vec_dststt (const vector unsigned int *, int, const int);
7407 void vec_dststt (const vector signed int *, int, const int);
7408 void vec_dststt (const vector bool int *, int, const int);
7409 void vec_dststt (const vector float *, int, const int);
7410 void vec_dststt (const unsigned char *, int, const int);
7411 void vec_dststt (const signed char *, int, const int);
7412 void vec_dststt (const unsigned short *, int, const int);
7413 void vec_dststt (const short *, int, const int);
7414 void vec_dststt (const unsigned int *, int, const int);
7415 void vec_dststt (const int *, int, const int);
7416 void vec_dststt (const unsigned long *, int, const int);
7417 void vec_dststt (const long *, int, const int);
7418 void vec_dststt (const float *, int, const int);
7420 void vec_dstt (const vector unsigned char *, int, const int);
7421 void vec_dstt (const vector signed char *, int, const int);
7422 void vec_dstt (const vector bool char *, int, const int);
7423 void vec_dstt (const vector unsigned short *, int, const int);
7424 void vec_dstt (const vector signed short *, int, const int);
7425 void vec_dstt (const vector bool short *, int, const int);
7426 void vec_dstt (const vector pixel *, int, const int);
7427 void vec_dstt (const vector unsigned int *, int, const int);
7428 void vec_dstt (const vector signed int *, int, const int);
7429 void vec_dstt (const vector bool int *, int, const int);
7430 void vec_dstt (const vector float *, int, const int);
7431 void vec_dstt (const unsigned char *, int, const int);
7432 void vec_dstt (const signed char *, int, const int);
7433 void vec_dstt (const unsigned short *, int, const int);
7434 void vec_dstt (const short *, int, const int);
7435 void vec_dstt (const unsigned int *, int, const int);
7436 void vec_dstt (const int *, int, const int);
7437 void vec_dstt (const unsigned long *, int, const int);
7438 void vec_dstt (const long *, int, const int);
7439 void vec_dstt (const float *, int, const int);
7441 vector float vec_expte (vector float);
7443 vector float vec_floor (vector float);
7445 vector float vec_ld (int, const vector float *);
7446 vector float vec_ld (int, const float *);
7447 vector bool int vec_ld (int, const vector bool int *);
7448 vector signed int vec_ld (int, const vector signed int *);
7449 vector signed int vec_ld (int, const int *);
7450 vector signed int vec_ld (int, const long *);
7451 vector unsigned int vec_ld (int, const vector unsigned int *);
7452 vector unsigned int vec_ld (int, const unsigned int *);
7453 vector unsigned int vec_ld (int, const unsigned long *);
7454 vector bool short vec_ld (int, const vector bool short *);
7455 vector pixel vec_ld (int, const vector pixel *);
7456 vector signed short vec_ld (int, const vector signed short *);
7457 vector signed short vec_ld (int, const short *);
7458 vector unsigned short vec_ld (int, const vector unsigned short *);
7459 vector unsigned short vec_ld (int, const unsigned short *);
7460 vector bool char vec_ld (int, const vector bool char *);
7461 vector signed char vec_ld (int, const vector signed char *);
7462 vector signed char vec_ld (int, const signed char *);
7463 vector unsigned char vec_ld (int, const vector unsigned char *);
7464 vector unsigned char vec_ld (int, const unsigned char *);
7466 vector signed char vec_lde (int, const signed char *);
7467 vector unsigned char vec_lde (int, const unsigned char *);
7468 vector signed short vec_lde (int, const short *);
7469 vector unsigned short vec_lde (int, const unsigned short *);
7470 vector float vec_lde (int, const float *);
7471 vector signed int vec_lde (int, const int *);
7472 vector unsigned int vec_lde (int, const unsigned int *);
7473 vector signed int vec_lde (int, const long *);
7474 vector unsigned int vec_lde (int, const unsigned long *);
7476 vector float vec_lvewx (int, float *);
7477 vector signed int vec_lvewx (int, int *);
7478 vector unsigned int vec_lvewx (int, unsigned int *);
7479 vector signed int vec_lvewx (int, long *);
7480 vector unsigned int vec_lvewx (int, unsigned long *);
7482 vector signed short vec_lvehx (int, short *);
7483 vector unsigned short vec_lvehx (int, unsigned short *);
7485 vector signed char vec_lvebx (int, char *);
7486 vector unsigned char vec_lvebx (int, unsigned char *);
7488 vector float vec_ldl (int, const vector float *);
7489 vector float vec_ldl (int, const float *);
7490 vector bool int vec_ldl (int, const vector bool int *);
7491 vector signed int vec_ldl (int, const vector signed int *);
7492 vector signed int vec_ldl (int, const int *);
7493 vector signed int vec_ldl (int, const long *);
7494 vector unsigned int vec_ldl (int, const vector unsigned int *);
7495 vector unsigned int vec_ldl (int, const unsigned int *);
7496 vector unsigned int vec_ldl (int, const unsigned long *);
7497 vector bool short vec_ldl (int, const vector bool short *);
7498 vector pixel vec_ldl (int, const vector pixel *);
7499 vector signed short vec_ldl (int, const vector signed short *);
7500 vector signed short vec_ldl (int, const short *);
7501 vector unsigned short vec_ldl (int, const vector unsigned short *);
7502 vector unsigned short vec_ldl (int, const unsigned short *);
7503 vector bool char vec_ldl (int, const vector bool char *);
7504 vector signed char vec_ldl (int, const vector signed char *);
7505 vector signed char vec_ldl (int, const signed char *);
7506 vector unsigned char vec_ldl (int, const vector unsigned char *);
7507 vector unsigned char vec_ldl (int, const unsigned char *);
7509 vector float vec_loge (vector float);
7511 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
7512 vector unsigned char vec_lvsl (int, const volatile signed char *);
7513 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
7514 vector unsigned char vec_lvsl (int, const volatile short *);
7515 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
7516 vector unsigned char vec_lvsl (int, const volatile int *);
7517 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
7518 vector unsigned char vec_lvsl (int, const volatile long *);
7519 vector unsigned char vec_lvsl (int, const volatile float *);
7521 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
7522 vector unsigned char vec_lvsr (int, const volatile signed char *);
7523 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
7524 vector unsigned char vec_lvsr (int, const volatile short *);
7525 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
7526 vector unsigned char vec_lvsr (int, const volatile int *);
7527 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
7528 vector unsigned char vec_lvsr (int, const volatile long *);
7529 vector unsigned char vec_lvsr (int, const volatile float *);
7531 vector float vec_madd (vector float, vector float, vector float);
7533 vector signed short vec_madds (vector signed short,
7534 vector signed short,
7535 vector signed short);
7537 vector unsigned char vec_max (vector bool char, vector unsigned char);
7538 vector unsigned char vec_max (vector unsigned char, vector bool char);
7539 vector unsigned char vec_max (vector unsigned char,
7540 vector unsigned char);
7541 vector signed char vec_max (vector bool char, vector signed char);
7542 vector signed char vec_max (vector signed char, vector bool char);
7543 vector signed char vec_max (vector signed char, vector signed char);
7544 vector unsigned short vec_max (vector bool short,
7545 vector unsigned short);
7546 vector unsigned short vec_max (vector unsigned short,
7548 vector unsigned short vec_max (vector unsigned short,
7549 vector unsigned short);
7550 vector signed short vec_max (vector bool short, vector signed short);
7551 vector signed short vec_max (vector signed short, vector bool short);
7552 vector signed short vec_max (vector signed short, vector signed short);
7553 vector unsigned int vec_max (vector bool int, vector unsigned int);
7554 vector unsigned int vec_max (vector unsigned int, vector bool int);
7555 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
7556 vector signed int vec_max (vector bool int, vector signed int);
7557 vector signed int vec_max (vector signed int, vector bool int);
7558 vector signed int vec_max (vector signed int, vector signed int);
7559 vector float vec_max (vector float, vector float);
7561 vector float vec_vmaxfp (vector float, vector float);
7563 vector signed int vec_vmaxsw (vector bool int, vector signed int);
7564 vector signed int vec_vmaxsw (vector signed int, vector bool int);
7565 vector signed int vec_vmaxsw (vector signed int, vector signed int);
7567 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
7568 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
7569 vector unsigned int vec_vmaxuw (vector unsigned int,
7570 vector unsigned int);
7572 vector signed short vec_vmaxsh (vector bool short, vector signed short);
7573 vector signed short vec_vmaxsh (vector signed short, vector bool short);
7574 vector signed short vec_vmaxsh (vector signed short,
7575 vector signed short);
7577 vector unsigned short vec_vmaxuh (vector bool short,
7578 vector unsigned short);
7579 vector unsigned short vec_vmaxuh (vector unsigned short,
7581 vector unsigned short vec_vmaxuh (vector unsigned short,
7582 vector unsigned short);
7584 vector signed char vec_vmaxsb (vector bool char, vector signed char);
7585 vector signed char vec_vmaxsb (vector signed char, vector bool char);
7586 vector signed char vec_vmaxsb (vector signed char, vector signed char);
7588 vector unsigned char vec_vmaxub (vector bool char,
7589 vector unsigned char);
7590 vector unsigned char vec_vmaxub (vector unsigned char,
7592 vector unsigned char vec_vmaxub (vector unsigned char,
7593 vector unsigned char);
7595 vector bool char vec_mergeh (vector bool char, vector bool char);
7596 vector signed char vec_mergeh (vector signed char, vector signed char);
7597 vector unsigned char vec_mergeh (vector unsigned char,
7598 vector unsigned char);
7599 vector bool short vec_mergeh (vector bool short, vector bool short);
7600 vector pixel vec_mergeh (vector pixel, vector pixel);
7601 vector signed short vec_mergeh (vector signed short,
7602 vector signed short);
7603 vector unsigned short vec_mergeh (vector unsigned short,
7604 vector unsigned short);
7605 vector float vec_mergeh (vector float, vector float);
7606 vector bool int vec_mergeh (vector bool int, vector bool int);
7607 vector signed int vec_mergeh (vector signed int, vector signed int);
7608 vector unsigned int vec_mergeh (vector unsigned int,
7609 vector unsigned int);
7611 vector float vec_vmrghw (vector float, vector float);
7612 vector bool int vec_vmrghw (vector bool int, vector bool int);
7613 vector signed int vec_vmrghw (vector signed int, vector signed int);
7614 vector unsigned int vec_vmrghw (vector unsigned int,
7615 vector unsigned int);
7617 vector bool short vec_vmrghh (vector bool short, vector bool short);
7618 vector signed short vec_vmrghh (vector signed short,
7619 vector signed short);
7620 vector unsigned short vec_vmrghh (vector unsigned short,
7621 vector unsigned short);
7622 vector pixel vec_vmrghh (vector pixel, vector pixel);
7624 vector bool char vec_vmrghb (vector bool char, vector bool char);
7625 vector signed char vec_vmrghb (vector signed char, vector signed char);
7626 vector unsigned char vec_vmrghb (vector unsigned char,
7627 vector unsigned char);
7629 vector bool char vec_mergel (vector bool char, vector bool char);
7630 vector signed char vec_mergel (vector signed char, vector signed char);
7631 vector unsigned char vec_mergel (vector unsigned char,
7632 vector unsigned char);
7633 vector bool short vec_mergel (vector bool short, vector bool short);
7634 vector pixel vec_mergel (vector pixel, vector pixel);
7635 vector signed short vec_mergel (vector signed short,
7636 vector signed short);
7637 vector unsigned short vec_mergel (vector unsigned short,
7638 vector unsigned short);
7639 vector float vec_mergel (vector float, vector float);
7640 vector bool int vec_mergel (vector bool int, vector bool int);
7641 vector signed int vec_mergel (vector signed int, vector signed int);
7642 vector unsigned int vec_mergel (vector unsigned int,
7643 vector unsigned int);
7645 vector float vec_vmrglw (vector float, vector float);
7646 vector signed int vec_vmrglw (vector signed int, vector signed int);
7647 vector unsigned int vec_vmrglw (vector unsigned int,
7648 vector unsigned int);
7649 vector bool int vec_vmrglw (vector bool int, vector bool int);
7651 vector bool short vec_vmrglh (vector bool short, vector bool short);
7652 vector signed short vec_vmrglh (vector signed short,
7653 vector signed short);
7654 vector unsigned short vec_vmrglh (vector unsigned short,
7655 vector unsigned short);
7656 vector pixel vec_vmrglh (vector pixel, vector pixel);
7658 vector bool char vec_vmrglb (vector bool char, vector bool char);
7659 vector signed char vec_vmrglb (vector signed char, vector signed char);
7660 vector unsigned char vec_vmrglb (vector unsigned char,
7661 vector unsigned char);
7663 vector unsigned short vec_mfvscr (void);
7665 vector unsigned char vec_min (vector bool char, vector unsigned char);
7666 vector unsigned char vec_min (vector unsigned char, vector bool char);
7667 vector unsigned char vec_min (vector unsigned char,
7668 vector unsigned char);
7669 vector signed char vec_min (vector bool char, vector signed char);
7670 vector signed char vec_min (vector signed char, vector bool char);
7671 vector signed char vec_min (vector signed char, vector signed char);
7672 vector unsigned short vec_min (vector bool short,
7673 vector unsigned short);
7674 vector unsigned short vec_min (vector unsigned short,
7676 vector unsigned short vec_min (vector unsigned short,
7677 vector unsigned short);
7678 vector signed short vec_min (vector bool short, vector signed short);
7679 vector signed short vec_min (vector signed short, vector bool short);
7680 vector signed short vec_min (vector signed short, vector signed short);
7681 vector unsigned int vec_min (vector bool int, vector unsigned int);
7682 vector unsigned int vec_min (vector unsigned int, vector bool int);
7683 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
7684 vector signed int vec_min (vector bool int, vector signed int);
7685 vector signed int vec_min (vector signed int, vector bool int);
7686 vector signed int vec_min (vector signed int, vector signed int);
7687 vector float vec_min (vector float, vector float);
7689 vector float vec_vminfp (vector float, vector float);
7691 vector signed int vec_vminsw (vector bool int, vector signed int);
7692 vector signed int vec_vminsw (vector signed int, vector bool int);
7693 vector signed int vec_vminsw (vector signed int, vector signed int);
7695 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
7696 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
7697 vector unsigned int vec_vminuw (vector unsigned int,
7698 vector unsigned int);
7700 vector signed short vec_vminsh (vector bool short, vector signed short);
7701 vector signed short vec_vminsh (vector signed short, vector bool short);
7702 vector signed short vec_vminsh (vector signed short,
7703 vector signed short);
7705 vector unsigned short vec_vminuh (vector bool short,
7706 vector unsigned short);
7707 vector unsigned short vec_vminuh (vector unsigned short,
7709 vector unsigned short vec_vminuh (vector unsigned short,
7710 vector unsigned short);
7712 vector signed char vec_vminsb (vector bool char, vector signed char);
7713 vector signed char vec_vminsb (vector signed char, vector bool char);
7714 vector signed char vec_vminsb (vector signed char, vector signed char);
7716 vector unsigned char vec_vminub (vector bool char,
7717 vector unsigned char);
7718 vector unsigned char vec_vminub (vector unsigned char,
7720 vector unsigned char vec_vminub (vector unsigned char,
7721 vector unsigned char);
7723 vector signed short vec_mladd (vector signed short,
7724 vector signed short,
7725 vector signed short);
7726 vector signed short vec_mladd (vector signed short,
7727 vector unsigned short,
7728 vector unsigned short);
7729 vector signed short vec_mladd (vector unsigned short,
7730 vector signed short,
7731 vector signed short);
7732 vector unsigned short vec_mladd (vector unsigned short,
7733 vector unsigned short,
7734 vector unsigned short);
7736 vector signed short vec_mradds (vector signed short,
7737 vector signed short,
7738 vector signed short);
7740 vector unsigned int vec_msum (vector unsigned char,
7741 vector unsigned char,
7742 vector unsigned int);
7743 vector signed int vec_msum (vector signed char,
7744 vector unsigned char,
7746 vector unsigned int vec_msum (vector unsigned short,
7747 vector unsigned short,
7748 vector unsigned int);
7749 vector signed int vec_msum (vector signed short,
7750 vector signed short,
7753 vector signed int vec_vmsumshm (vector signed short,
7754 vector signed short,
7757 vector unsigned int vec_vmsumuhm (vector unsigned short,
7758 vector unsigned short,
7759 vector unsigned int);
7761 vector signed int vec_vmsummbm (vector signed char,
7762 vector unsigned char,
7765 vector unsigned int vec_vmsumubm (vector unsigned char,
7766 vector unsigned char,
7767 vector unsigned int);
7769 vector unsigned int vec_msums (vector unsigned short,
7770 vector unsigned short,
7771 vector unsigned int);
7772 vector signed int vec_msums (vector signed short,
7773 vector signed short,
7776 vector signed int vec_vmsumshs (vector signed short,
7777 vector signed short,
7780 vector unsigned int vec_vmsumuhs (vector unsigned short,
7781 vector unsigned short,
7782 vector unsigned int);
7784 void vec_mtvscr (vector signed int);
7785 void vec_mtvscr (vector unsigned int);
7786 void vec_mtvscr (vector bool int);
7787 void vec_mtvscr (vector signed short);
7788 void vec_mtvscr (vector unsigned short);
7789 void vec_mtvscr (vector bool short);
7790 void vec_mtvscr (vector pixel);
7791 void vec_mtvscr (vector signed char);
7792 void vec_mtvscr (vector unsigned char);
7793 void vec_mtvscr (vector bool char);
7795 vector unsigned short vec_mule (vector unsigned char,
7796 vector unsigned char);
7797 vector signed short vec_mule (vector signed char,
7798 vector signed char);
7799 vector unsigned int vec_mule (vector unsigned short,
7800 vector unsigned short);
7801 vector signed int vec_mule (vector signed short, vector signed short);
7803 vector signed int vec_vmulesh (vector signed short,
7804 vector signed short);
7806 vector unsigned int vec_vmuleuh (vector unsigned short,
7807 vector unsigned short);
7809 vector signed short vec_vmulesb (vector signed char,
7810 vector signed char);
7812 vector unsigned short vec_vmuleub (vector unsigned char,
7813 vector unsigned char);
7815 vector unsigned short vec_mulo (vector unsigned char,
7816 vector unsigned char);
7817 vector signed short vec_mulo (vector signed char, vector signed char);
7818 vector unsigned int vec_mulo (vector unsigned short,
7819 vector unsigned short);
7820 vector signed int vec_mulo (vector signed short, vector signed short);
7822 vector signed int vec_vmulosh (vector signed short,
7823 vector signed short);
7825 vector unsigned int vec_vmulouh (vector unsigned short,
7826 vector unsigned short);
7828 vector signed short vec_vmulosb (vector signed char,
7829 vector signed char);
7831 vector unsigned short vec_vmuloub (vector unsigned char,
7832 vector unsigned char);
7834 vector float vec_nmsub (vector float, vector float, vector float);
7836 vector float vec_nor (vector float, vector float);
7837 vector signed int vec_nor (vector signed int, vector signed int);
7838 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
7839 vector bool int vec_nor (vector bool int, vector bool int);
7840 vector signed short vec_nor (vector signed short, vector signed short);
7841 vector unsigned short vec_nor (vector unsigned short,
7842 vector unsigned short);
7843 vector bool short vec_nor (vector bool short, vector bool short);
7844 vector signed char vec_nor (vector signed char, vector signed char);
7845 vector unsigned char vec_nor (vector unsigned char,
7846 vector unsigned char);
7847 vector bool char vec_nor (vector bool char, vector bool char);
7849 vector float vec_or (vector float, vector float);
7850 vector float vec_or (vector float, vector bool int);
7851 vector float vec_or (vector bool int, vector float);
7852 vector bool int vec_or (vector bool int, vector bool int);
7853 vector signed int vec_or (vector bool int, vector signed int);
7854 vector signed int vec_or (vector signed int, vector bool int);
7855 vector signed int vec_or (vector signed int, vector signed int);
7856 vector unsigned int vec_or (vector bool int, vector unsigned int);
7857 vector unsigned int vec_or (vector unsigned int, vector bool int);
7858 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
7859 vector bool short vec_or (vector bool short, vector bool short);
7860 vector signed short vec_or (vector bool short, vector signed short);
7861 vector signed short vec_or (vector signed short, vector bool short);
7862 vector signed short vec_or (vector signed short, vector signed short);
7863 vector unsigned short vec_or (vector bool short, vector unsigned short);
7864 vector unsigned short vec_or (vector unsigned short, vector bool short);
7865 vector unsigned short vec_or (vector unsigned short,
7866 vector unsigned short);
7867 vector signed char vec_or (vector bool char, vector signed char);
7868 vector bool char vec_or (vector bool char, vector bool char);
7869 vector signed char vec_or (vector signed char, vector bool char);
7870 vector signed char vec_or (vector signed char, vector signed char);
7871 vector unsigned char vec_or (vector bool char, vector unsigned char);
7872 vector unsigned char vec_or (vector unsigned char, vector bool char);
7873 vector unsigned char vec_or (vector unsigned char,
7874 vector unsigned char);
7876 vector signed char vec_pack (vector signed short, vector signed short);
7877 vector unsigned char vec_pack (vector unsigned short,
7878 vector unsigned short);
7879 vector bool char vec_pack (vector bool short, vector bool short);
7880 vector signed short vec_pack (vector signed int, vector signed int);
7881 vector unsigned short vec_pack (vector unsigned int,
7882 vector unsigned int);
7883 vector bool short vec_pack (vector bool int, vector bool int);
7885 vector bool short vec_vpkuwum (vector bool int, vector bool int);
7886 vector signed short vec_vpkuwum (vector signed int, vector signed int);
7887 vector unsigned short vec_vpkuwum (vector unsigned int,
7888 vector unsigned int);
7890 vector bool char vec_vpkuhum (vector bool short, vector bool short);
7891 vector signed char vec_vpkuhum (vector signed short,
7892 vector signed short);
7893 vector unsigned char vec_vpkuhum (vector unsigned short,
7894 vector unsigned short);
7896 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
7898 vector unsigned char vec_packs (vector unsigned short,
7899 vector unsigned short);
7900 vector signed char vec_packs (vector signed short, vector signed short);
7901 vector unsigned short vec_packs (vector unsigned int,
7902 vector unsigned int);
7903 vector signed short vec_packs (vector signed int, vector signed int);
7905 vector signed short vec_vpkswss (vector signed int, vector signed int);
7907 vector unsigned short vec_vpkuwus (vector unsigned int,
7908 vector unsigned int);
7910 vector signed char vec_vpkshss (vector signed short,
7911 vector signed short);
7913 vector unsigned char vec_vpkuhus (vector unsigned short,
7914 vector unsigned short);
7916 vector unsigned char vec_packsu (vector unsigned short,
7917 vector unsigned short);
7918 vector unsigned char vec_packsu (vector signed short,
7919 vector signed short);
7920 vector unsigned short vec_packsu (vector unsigned int,
7921 vector unsigned int);
7922 vector unsigned short vec_packsu (vector signed int, vector signed int);
7924 vector unsigned short vec_vpkswus (vector signed int,
7927 vector unsigned char vec_vpkshus (vector signed short,
7928 vector signed short);
7930 vector float vec_perm (vector float,
7932 vector unsigned char);
7933 vector signed int vec_perm (vector signed int,
7935 vector unsigned char);
7936 vector unsigned int vec_perm (vector unsigned int,
7937 vector unsigned int,
7938 vector unsigned char);
7939 vector bool int vec_perm (vector bool int,
7941 vector unsigned char);
7942 vector signed short vec_perm (vector signed short,
7943 vector signed short,
7944 vector unsigned char);
7945 vector unsigned short vec_perm (vector unsigned short,
7946 vector unsigned short,
7947 vector unsigned char);
7948 vector bool short vec_perm (vector bool short,
7950 vector unsigned char);
7951 vector pixel vec_perm (vector pixel,
7953 vector unsigned char);
7954 vector signed char vec_perm (vector signed char,
7956 vector unsigned char);
7957 vector unsigned char vec_perm (vector unsigned char,
7958 vector unsigned char,
7959 vector unsigned char);
7960 vector bool char vec_perm (vector bool char,
7962 vector unsigned char);
7964 vector float vec_re (vector float);
7966 vector signed char vec_rl (vector signed char,
7967 vector unsigned char);
7968 vector unsigned char vec_rl (vector unsigned char,
7969 vector unsigned char);
7970 vector signed short vec_rl (vector signed short, vector unsigned short);
7971 vector unsigned short vec_rl (vector unsigned short,
7972 vector unsigned short);
7973 vector signed int vec_rl (vector signed int, vector unsigned int);
7974 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
7976 vector signed int vec_vrlw (vector signed int, vector unsigned int);
7977 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
7979 vector signed short vec_vrlh (vector signed short,
7980 vector unsigned short);
7981 vector unsigned short vec_vrlh (vector unsigned short,
7982 vector unsigned short);
7984 vector signed char vec_vrlb (vector signed char, vector unsigned char);
7985 vector unsigned char vec_vrlb (vector unsigned char,
7986 vector unsigned char);
7988 vector float vec_round (vector float);
7990 vector float vec_rsqrte (vector float);
7992 vector float vec_sel (vector float, vector float, vector bool int);
7993 vector float vec_sel (vector float, vector float, vector unsigned int);
7994 vector signed int vec_sel (vector signed int,
7997 vector signed int vec_sel (vector signed int,
7999 vector unsigned int);
8000 vector unsigned int vec_sel (vector unsigned int,
8001 vector unsigned int,
8003 vector unsigned int vec_sel (vector unsigned int,
8004 vector unsigned int,
8005 vector unsigned int);
8006 vector bool int vec_sel (vector bool int,
8009 vector bool int vec_sel (vector bool int,
8011 vector unsigned int);
8012 vector signed short vec_sel (vector signed short,
8013 vector signed short,
8015 vector signed short vec_sel (vector signed short,
8016 vector signed short,
8017 vector unsigned short);
8018 vector unsigned short vec_sel (vector unsigned short,
8019 vector unsigned short,
8021 vector unsigned short vec_sel (vector unsigned short,
8022 vector unsigned short,
8023 vector unsigned short);
8024 vector bool short vec_sel (vector bool short,
8027 vector bool short vec_sel (vector bool short,
8029 vector unsigned short);
8030 vector signed char vec_sel (vector signed char,
8033 vector signed char vec_sel (vector signed char,
8035 vector unsigned char);
8036 vector unsigned char vec_sel (vector unsigned char,
8037 vector unsigned char,
8039 vector unsigned char vec_sel (vector unsigned char,
8040 vector unsigned char,
8041 vector unsigned char);
8042 vector bool char vec_sel (vector bool char,
8045 vector bool char vec_sel (vector bool char,
8047 vector unsigned char);
8049 vector signed char vec_sl (vector signed char,
8050 vector unsigned char);
8051 vector unsigned char vec_sl (vector unsigned char,
8052 vector unsigned char);
8053 vector signed short vec_sl (vector signed short, vector unsigned short);
8054 vector unsigned short vec_sl (vector unsigned short,
8055 vector unsigned short);
8056 vector signed int vec_sl (vector signed int, vector unsigned int);
8057 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
8059 vector signed int vec_vslw (vector signed int, vector unsigned int);
8060 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
8062 vector signed short vec_vslh (vector signed short,
8063 vector unsigned short);
8064 vector unsigned short vec_vslh (vector unsigned short,
8065 vector unsigned short);
8067 vector signed char vec_vslb (vector signed char, vector unsigned char);
8068 vector unsigned char vec_vslb (vector unsigned char,
8069 vector unsigned char);
8071 vector float vec_sld (vector float, vector float, const int);
8072 vector signed int vec_sld (vector signed int,
8075 vector unsigned int vec_sld (vector unsigned int,
8076 vector unsigned int,
8078 vector bool int vec_sld (vector bool int,
8081 vector signed short vec_sld (vector signed short,
8082 vector signed short,
8084 vector unsigned short vec_sld (vector unsigned short,
8085 vector unsigned short,
8087 vector bool short vec_sld (vector bool short,
8090 vector pixel vec_sld (vector pixel,
8093 vector signed char vec_sld (vector signed char,
8096 vector unsigned char vec_sld (vector unsigned char,
8097 vector unsigned char,
8099 vector bool char vec_sld (vector bool char,
8103 vector signed int vec_sll (vector signed int,
8104 vector unsigned int);
8105 vector signed int vec_sll (vector signed int,
8106 vector unsigned short);
8107 vector signed int vec_sll (vector signed int,
8108 vector unsigned char);
8109 vector unsigned int vec_sll (vector unsigned int,
8110 vector unsigned int);
8111 vector unsigned int vec_sll (vector unsigned int,
8112 vector unsigned short);
8113 vector unsigned int vec_sll (vector unsigned int,
8114 vector unsigned char);
8115 vector bool int vec_sll (vector bool int,
8116 vector unsigned int);
8117 vector bool int vec_sll (vector bool int,
8118 vector unsigned short);
8119 vector bool int vec_sll (vector bool int,
8120 vector unsigned char);
8121 vector signed short vec_sll (vector signed short,
8122 vector unsigned int);
8123 vector signed short vec_sll (vector signed short,
8124 vector unsigned short);
8125 vector signed short vec_sll (vector signed short,
8126 vector unsigned char);
8127 vector unsigned short vec_sll (vector unsigned short,
8128 vector unsigned int);
8129 vector unsigned short vec_sll (vector unsigned short,
8130 vector unsigned short);
8131 vector unsigned short vec_sll (vector unsigned short,
8132 vector unsigned char);
8133 vector bool short vec_sll (vector bool short, vector unsigned int);
8134 vector bool short vec_sll (vector bool short, vector unsigned short);
8135 vector bool short vec_sll (vector bool short, vector unsigned char);
8136 vector pixel vec_sll (vector pixel, vector unsigned int);
8137 vector pixel vec_sll (vector pixel, vector unsigned short);
8138 vector pixel vec_sll (vector pixel, vector unsigned char);
8139 vector signed char vec_sll (vector signed char, vector unsigned int);
8140 vector signed char vec_sll (vector signed char, vector unsigned short);
8141 vector signed char vec_sll (vector signed char, vector unsigned char);
8142 vector unsigned char vec_sll (vector unsigned char,
8143 vector unsigned int);
8144 vector unsigned char vec_sll (vector unsigned char,
8145 vector unsigned short);
8146 vector unsigned char vec_sll (vector unsigned char,
8147 vector unsigned char);
8148 vector bool char vec_sll (vector bool char, vector unsigned int);
8149 vector bool char vec_sll (vector bool char, vector unsigned short);
8150 vector bool char vec_sll (vector bool char, vector unsigned char);
8152 vector float vec_slo (vector float, vector signed char);
8153 vector float vec_slo (vector float, vector unsigned char);
8154 vector signed int vec_slo (vector signed int, vector signed char);
8155 vector signed int vec_slo (vector signed int, vector unsigned char);
8156 vector unsigned int vec_slo (vector unsigned int, vector signed char);
8157 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
8158 vector signed short vec_slo (vector signed short, vector signed char);
8159 vector signed short vec_slo (vector signed short, vector unsigned char);
8160 vector unsigned short vec_slo (vector unsigned short,
8161 vector signed char);
8162 vector unsigned short vec_slo (vector unsigned short,
8163 vector unsigned char);
8164 vector pixel vec_slo (vector pixel, vector signed char);
8165 vector pixel vec_slo (vector pixel, vector unsigned char);
8166 vector signed char vec_slo (vector signed char, vector signed char);
8167 vector signed char vec_slo (vector signed char, vector unsigned char);
8168 vector unsigned char vec_slo (vector unsigned char, vector signed char);
8169 vector unsigned char vec_slo (vector unsigned char,
8170 vector unsigned char);
8172 vector signed char vec_splat (vector signed char, const int);
8173 vector unsigned char vec_splat (vector unsigned char, const int);
8174 vector bool char vec_splat (vector bool char, const int);
8175 vector signed short vec_splat (vector signed short, const int);
8176 vector unsigned short vec_splat (vector unsigned short, const int);
8177 vector bool short vec_splat (vector bool short, const int);
8178 vector pixel vec_splat (vector pixel, const int);
8179 vector float vec_splat (vector float, const int);
8180 vector signed int vec_splat (vector signed int, const int);
8181 vector unsigned int vec_splat (vector unsigned int, const int);
8182 vector bool int vec_splat (vector bool int, const int);
8184 vector float vec_vspltw (vector float, const int);
8185 vector signed int vec_vspltw (vector signed int, const int);
8186 vector unsigned int vec_vspltw (vector unsigned int, const int);
8187 vector bool int vec_vspltw (vector bool int, const int);
8189 vector bool short vec_vsplth (vector bool short, const int);
8190 vector signed short vec_vsplth (vector signed short, const int);
8191 vector unsigned short vec_vsplth (vector unsigned short, const int);
8192 vector pixel vec_vsplth (vector pixel, const int);
8194 vector signed char vec_vspltb (vector signed char, const int);
8195 vector unsigned char vec_vspltb (vector unsigned char, const int);
8196 vector bool char vec_vspltb (vector bool char, const int);
8198 vector signed char vec_splat_s8 (const int);
8200 vector signed short vec_splat_s16 (const int);
8202 vector signed int vec_splat_s32 (const int);
8204 vector unsigned char vec_splat_u8 (const int);
8206 vector unsigned short vec_splat_u16 (const int);
8208 vector unsigned int vec_splat_u32 (const int);
8210 vector signed char vec_sr (vector signed char, vector unsigned char);
8211 vector unsigned char vec_sr (vector unsigned char,
8212 vector unsigned char);
8213 vector signed short vec_sr (vector signed short,
8214 vector unsigned short);
8215 vector unsigned short vec_sr (vector unsigned short,
8216 vector unsigned short);
8217 vector signed int vec_sr (vector signed int, vector unsigned int);
8218 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
8220 vector signed int vec_vsrw (vector signed int, vector unsigned int);
8221 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
8223 vector signed short vec_vsrh (vector signed short,
8224 vector unsigned short);
8225 vector unsigned short vec_vsrh (vector unsigned short,
8226 vector unsigned short);
8228 vector signed char vec_vsrb (vector signed char, vector unsigned char);
8229 vector unsigned char vec_vsrb (vector unsigned char,
8230 vector unsigned char);
8232 vector signed char vec_sra (vector signed char, vector unsigned char);
8233 vector unsigned char vec_sra (vector unsigned char,
8234 vector unsigned char);
8235 vector signed short vec_sra (vector signed short,
8236 vector unsigned short);
8237 vector unsigned short vec_sra (vector unsigned short,
8238 vector unsigned short);
8239 vector signed int vec_sra (vector signed int, vector unsigned int);
8240 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
8242 vector signed int vec_vsraw (vector signed int, vector unsigned int);
8243 vector unsigned int vec_vsraw (vector unsigned int,
8244 vector unsigned int);
8246 vector signed short vec_vsrah (vector signed short,
8247 vector unsigned short);
8248 vector unsigned short vec_vsrah (vector unsigned short,
8249 vector unsigned short);
8251 vector signed char vec_vsrab (vector signed char, vector unsigned char);
8252 vector unsigned char vec_vsrab (vector unsigned char,
8253 vector unsigned char);
8255 vector signed int vec_srl (vector signed int, vector unsigned int);
8256 vector signed int vec_srl (vector signed int, vector unsigned short);
8257 vector signed int vec_srl (vector signed int, vector unsigned char);
8258 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
8259 vector unsigned int vec_srl (vector unsigned int,
8260 vector unsigned short);
8261 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
8262 vector bool int vec_srl (vector bool int, vector unsigned int);
8263 vector bool int vec_srl (vector bool int, vector unsigned short);
8264 vector bool int vec_srl (vector bool int, vector unsigned char);
8265 vector signed short vec_srl (vector signed short, vector unsigned int);
8266 vector signed short vec_srl (vector signed short,
8267 vector unsigned short);
8268 vector signed short vec_srl (vector signed short, vector unsigned char);
8269 vector unsigned short vec_srl (vector unsigned short,
8270 vector unsigned int);
8271 vector unsigned short vec_srl (vector unsigned short,
8272 vector unsigned short);
8273 vector unsigned short vec_srl (vector unsigned short,
8274 vector unsigned char);
8275 vector bool short vec_srl (vector bool short, vector unsigned int);
8276 vector bool short vec_srl (vector bool short, vector unsigned short);
8277 vector bool short vec_srl (vector bool short, vector unsigned char);
8278 vector pixel vec_srl (vector pixel, vector unsigned int);
8279 vector pixel vec_srl (vector pixel, vector unsigned short);
8280 vector pixel vec_srl (vector pixel, vector unsigned char);
8281 vector signed char vec_srl (vector signed char, vector unsigned int);
8282 vector signed char vec_srl (vector signed char, vector unsigned short);
8283 vector signed char vec_srl (vector signed char, vector unsigned char);
8284 vector unsigned char vec_srl (vector unsigned char,
8285 vector unsigned int);
8286 vector unsigned char vec_srl (vector unsigned char,
8287 vector unsigned short);
8288 vector unsigned char vec_srl (vector unsigned char,
8289 vector unsigned char);
8290 vector bool char vec_srl (vector bool char, vector unsigned int);
8291 vector bool char vec_srl (vector bool char, vector unsigned short);
8292 vector bool char vec_srl (vector bool char, vector unsigned char);
8294 vector float vec_sro (vector float, vector signed char);
8295 vector float vec_sro (vector float, vector unsigned char);
8296 vector signed int vec_sro (vector signed int, vector signed char);
8297 vector signed int vec_sro (vector signed int, vector unsigned char);
8298 vector unsigned int vec_sro (vector unsigned int, vector signed char);
8299 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
8300 vector signed short vec_sro (vector signed short, vector signed char);
8301 vector signed short vec_sro (vector signed short, vector unsigned char);
8302 vector unsigned short vec_sro (vector unsigned short,
8303 vector signed char);
8304 vector unsigned short vec_sro (vector unsigned short,
8305 vector unsigned char);
8306 vector pixel vec_sro (vector pixel, vector signed char);
8307 vector pixel vec_sro (vector pixel, vector unsigned char);
8308 vector signed char vec_sro (vector signed char, vector signed char);
8309 vector signed char vec_sro (vector signed char, vector unsigned char);
8310 vector unsigned char vec_sro (vector unsigned char, vector signed char);
8311 vector unsigned char vec_sro (vector unsigned char,
8312 vector unsigned char);
8314 void vec_st (vector float, int, vector float *);
8315 void vec_st (vector float, int, float *);
8316 void vec_st (vector signed int, int, vector signed int *);
8317 void vec_st (vector signed int, int, int *);
8318 void vec_st (vector unsigned int, int, vector unsigned int *);
8319 void vec_st (vector unsigned int, int, unsigned int *);
8320 void vec_st (vector bool int, int, vector bool int *);
8321 void vec_st (vector bool int, int, unsigned int *);
8322 void vec_st (vector bool int, int, int *);
8323 void vec_st (vector signed short, int, vector signed short *);
8324 void vec_st (vector signed short, int, short *);
8325 void vec_st (vector unsigned short, int, vector unsigned short *);
8326 void vec_st (vector unsigned short, int, unsigned short *);
8327 void vec_st (vector bool short, int, vector bool short *);
8328 void vec_st (vector bool short, int, unsigned short *);
8329 void vec_st (vector pixel, int, vector pixel *);
8330 void vec_st (vector pixel, int, unsigned short *);
8331 void vec_st (vector pixel, int, short *);
8332 void vec_st (vector bool short, int, short *);
8333 void vec_st (vector signed char, int, vector signed char *);
8334 void vec_st (vector signed char, int, signed char *);
8335 void vec_st (vector unsigned char, int, vector unsigned char *);
8336 void vec_st (vector unsigned char, int, unsigned char *);
8337 void vec_st (vector bool char, int, vector bool char *);
8338 void vec_st (vector bool char, int, unsigned char *);
8339 void vec_st (vector bool char, int, signed char *);
8341 void vec_ste (vector signed char, int, signed char *);
8342 void vec_ste (vector unsigned char, int, unsigned char *);
8343 void vec_ste (vector bool char, int, signed char *);
8344 void vec_ste (vector bool char, int, unsigned char *);
8345 void vec_ste (vector signed short, int, short *);
8346 void vec_ste (vector unsigned short, int, unsigned short *);
8347 void vec_ste (vector bool short, int, short *);
8348 void vec_ste (vector bool short, int, unsigned short *);
8349 void vec_ste (vector pixel, int, short *);
8350 void vec_ste (vector pixel, int, unsigned short *);
8351 void vec_ste (vector float, int, float *);
8352 void vec_ste (vector signed int, int, int *);
8353 void vec_ste (vector unsigned int, int, unsigned int *);
8354 void vec_ste (vector bool int, int, int *);
8355 void vec_ste (vector bool int, int, unsigned int *);
8357 void vec_stvewx (vector float, int, float *);
8358 void vec_stvewx (vector signed int, int, int *);
8359 void vec_stvewx (vector unsigned int, int, unsigned int *);
8360 void vec_stvewx (vector bool int, int, int *);
8361 void vec_stvewx (vector bool int, int, unsigned int *);
8363 void vec_stvehx (vector signed short, int, short *);
8364 void vec_stvehx (vector unsigned short, int, unsigned short *);
8365 void vec_stvehx (vector bool short, int, short *);
8366 void vec_stvehx (vector bool short, int, unsigned short *);
8367 void vec_stvehx (vector pixel, int, short *);
8368 void vec_stvehx (vector pixel, int, unsigned short *);
8370 void vec_stvebx (vector signed char, int, signed char *);
8371 void vec_stvebx (vector unsigned char, int, unsigned char *);
8372 void vec_stvebx (vector bool char, int, signed char *);
8373 void vec_stvebx (vector bool char, int, unsigned char *);
8375 void vec_stl (vector float, int, vector float *);
8376 void vec_stl (vector float, int, float *);
8377 void vec_stl (vector signed int, int, vector signed int *);
8378 void vec_stl (vector signed int, int, int *);
8379 void vec_stl (vector unsigned int, int, vector unsigned int *);
8380 void vec_stl (vector unsigned int, int, unsigned int *);
8381 void vec_stl (vector bool int, int, vector bool int *);
8382 void vec_stl (vector bool int, int, unsigned int *);
8383 void vec_stl (vector bool int, int, int *);
8384 void vec_stl (vector signed short, int, vector signed short *);
8385 void vec_stl (vector signed short, int, short *);
8386 void vec_stl (vector unsigned short, int, vector unsigned short *);
8387 void vec_stl (vector unsigned short, int, unsigned short *);
8388 void vec_stl (vector bool short, int, vector bool short *);
8389 void vec_stl (vector bool short, int, unsigned short *);
8390 void vec_stl (vector bool short, int, short *);
8391 void vec_stl (vector pixel, int, vector pixel *);
8392 void vec_stl (vector pixel, int, unsigned short *);
8393 void vec_stl (vector pixel, int, short *);
8394 void vec_stl (vector signed char, int, vector signed char *);
8395 void vec_stl (vector signed char, int, signed char *);
8396 void vec_stl (vector unsigned char, int, vector unsigned char *);
8397 void vec_stl (vector unsigned char, int, unsigned char *);
8398 void vec_stl (vector bool char, int, vector bool char *);
8399 void vec_stl (vector bool char, int, unsigned char *);
8400 void vec_stl (vector bool char, int, signed char *);
8402 vector signed char vec_sub (vector bool char, vector signed char);
8403 vector signed char vec_sub (vector signed char, vector bool char);
8404 vector signed char vec_sub (vector signed char, vector signed char);
8405 vector unsigned char vec_sub (vector bool char, vector unsigned char);
8406 vector unsigned char vec_sub (vector unsigned char, vector bool char);
8407 vector unsigned char vec_sub (vector unsigned char,
8408 vector unsigned char);
8409 vector signed short vec_sub (vector bool short, vector signed short);
8410 vector signed short vec_sub (vector signed short, vector bool short);
8411 vector signed short vec_sub (vector signed short, vector signed short);
8412 vector unsigned short vec_sub (vector bool short,
8413 vector unsigned short);
8414 vector unsigned short vec_sub (vector unsigned short,
8416 vector unsigned short vec_sub (vector unsigned short,
8417 vector unsigned short);
8418 vector signed int vec_sub (vector bool int, vector signed int);
8419 vector signed int vec_sub (vector signed int, vector bool int);
8420 vector signed int vec_sub (vector signed int, vector signed int);
8421 vector unsigned int vec_sub (vector bool int, vector unsigned int);
8422 vector unsigned int vec_sub (vector unsigned int, vector bool int);
8423 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
8424 vector float vec_sub (vector float, vector float);
8426 vector float vec_vsubfp (vector float, vector float);
8428 vector signed int vec_vsubuwm (vector bool int, vector signed int);
8429 vector signed int vec_vsubuwm (vector signed int, vector bool int);
8430 vector signed int vec_vsubuwm (vector signed int, vector signed int);
8431 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
8432 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
8433 vector unsigned int vec_vsubuwm (vector unsigned int,
8434 vector unsigned int);
8436 vector signed short vec_vsubuhm (vector bool short,
8437 vector signed short);
8438 vector signed short vec_vsubuhm (vector signed short,
8440 vector signed short vec_vsubuhm (vector signed short,
8441 vector signed short);
8442 vector unsigned short vec_vsubuhm (vector bool short,
8443 vector unsigned short);
8444 vector unsigned short vec_vsubuhm (vector unsigned short,
8446 vector unsigned short vec_vsubuhm (vector unsigned short,
8447 vector unsigned short);
8449 vector signed char vec_vsububm (vector bool char, vector signed char);
8450 vector signed char vec_vsububm (vector signed char, vector bool char);
8451 vector signed char vec_vsububm (vector signed char, vector signed char);
8452 vector unsigned char vec_vsububm (vector bool char,
8453 vector unsigned char);
8454 vector unsigned char vec_vsububm (vector unsigned char,
8456 vector unsigned char vec_vsububm (vector unsigned char,
8457 vector unsigned char);
8459 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
8461 vector unsigned char vec_subs (vector bool char, vector unsigned char);
8462 vector unsigned char vec_subs (vector unsigned char, vector bool char);
8463 vector unsigned char vec_subs (vector unsigned char,
8464 vector unsigned char);
8465 vector signed char vec_subs (vector bool char, vector signed char);
8466 vector signed char vec_subs (vector signed char, vector bool char);
8467 vector signed char vec_subs (vector signed char, vector signed char);
8468 vector unsigned short vec_subs (vector bool short,
8469 vector unsigned short);
8470 vector unsigned short vec_subs (vector unsigned short,
8472 vector unsigned short vec_subs (vector unsigned short,
8473 vector unsigned short);
8474 vector signed short vec_subs (vector bool short, vector signed short);
8475 vector signed short vec_subs (vector signed short, vector bool short);
8476 vector signed short vec_subs (vector signed short, vector signed short);
8477 vector unsigned int vec_subs (vector bool int, vector unsigned int);
8478 vector unsigned int vec_subs (vector unsigned int, vector bool int);
8479 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
8480 vector signed int vec_subs (vector bool int, vector signed int);
8481 vector signed int vec_subs (vector signed int, vector bool int);
8482 vector signed int vec_subs (vector signed int, vector signed int);
8484 vector signed int vec_vsubsws (vector bool int, vector signed int);
8485 vector signed int vec_vsubsws (vector signed int, vector bool int);
8486 vector signed int vec_vsubsws (vector signed int, vector signed int);
8488 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
8489 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
8490 vector unsigned int vec_vsubuws (vector unsigned int,
8491 vector unsigned int);
8493 vector signed short vec_vsubshs (vector bool short,
8494 vector signed short);
8495 vector signed short vec_vsubshs (vector signed short,
8497 vector signed short vec_vsubshs (vector signed short,
8498 vector signed short);
8500 vector unsigned short vec_vsubuhs (vector bool short,
8501 vector unsigned short);
8502 vector unsigned short vec_vsubuhs (vector unsigned short,
8504 vector unsigned short vec_vsubuhs (vector unsigned short,
8505 vector unsigned short);
8507 vector signed char vec_vsubsbs (vector bool char, vector signed char);
8508 vector signed char vec_vsubsbs (vector signed char, vector bool char);
8509 vector signed char vec_vsubsbs (vector signed char, vector signed char);
8511 vector unsigned char vec_vsububs (vector bool char,
8512 vector unsigned char);
8513 vector unsigned char vec_vsububs (vector unsigned char,
8515 vector unsigned char vec_vsububs (vector unsigned char,
8516 vector unsigned char);
8518 vector unsigned int vec_sum4s (vector unsigned char,
8519 vector unsigned int);
8520 vector signed int vec_sum4s (vector signed char, vector signed int);
8521 vector signed int vec_sum4s (vector signed short, vector signed int);
8523 vector signed int vec_vsum4shs (vector signed short, vector signed int);
8525 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
8527 vector unsigned int vec_vsum4ubs (vector unsigned char,
8528 vector unsigned int);
8530 vector signed int vec_sum2s (vector signed int, vector signed int);
8532 vector signed int vec_sums (vector signed int, vector signed int);
8534 vector float vec_trunc (vector float);
8536 vector signed short vec_unpackh (vector signed char);
8537 vector bool short vec_unpackh (vector bool char);
8538 vector signed int vec_unpackh (vector signed short);
8539 vector bool int vec_unpackh (vector bool short);
8540 vector unsigned int vec_unpackh (vector pixel);
8542 vector bool int vec_vupkhsh (vector bool short);
8543 vector signed int vec_vupkhsh (vector signed short);
8545 vector unsigned int vec_vupkhpx (vector pixel);
8547 vector bool short vec_vupkhsb (vector bool char);
8548 vector signed short vec_vupkhsb (vector signed char);
8550 vector signed short vec_unpackl (vector signed char);
8551 vector bool short vec_unpackl (vector bool char);
8552 vector unsigned int vec_unpackl (vector pixel);
8553 vector signed int vec_unpackl (vector signed short);
8554 vector bool int vec_unpackl (vector bool short);
8556 vector unsigned int vec_vupklpx (vector pixel);
8558 vector bool int vec_vupklsh (vector bool short);
8559 vector signed int vec_vupklsh (vector signed short);
8561 vector bool short vec_vupklsb (vector bool char);
8562 vector signed short vec_vupklsb (vector signed char);
8564 vector float vec_xor (vector float, vector float);
8565 vector float vec_xor (vector float, vector bool int);
8566 vector float vec_xor (vector bool int, vector float);
8567 vector bool int vec_xor (vector bool int, vector bool int);
8568 vector signed int vec_xor (vector bool int, vector signed int);
8569 vector signed int vec_xor (vector signed int, vector bool int);
8570 vector signed int vec_xor (vector signed int, vector signed int);
8571 vector unsigned int vec_xor (vector bool int, vector unsigned int);
8572 vector unsigned int vec_xor (vector unsigned int, vector bool int);
8573 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
8574 vector bool short vec_xor (vector bool short, vector bool short);
8575 vector signed short vec_xor (vector bool short, vector signed short);
8576 vector signed short vec_xor (vector signed short, vector bool short);
8577 vector signed short vec_xor (vector signed short, vector signed short);
8578 vector unsigned short vec_xor (vector bool short,
8579 vector unsigned short);
8580 vector unsigned short vec_xor (vector unsigned short,
8582 vector unsigned short vec_xor (vector unsigned short,
8583 vector unsigned short);
8584 vector signed char vec_xor (vector bool char, vector signed char);
8585 vector bool char vec_xor (vector bool char, vector bool char);
8586 vector signed char vec_xor (vector signed char, vector bool char);
8587 vector signed char vec_xor (vector signed char, vector signed char);
8588 vector unsigned char vec_xor (vector bool char, vector unsigned char);
8589 vector unsigned char vec_xor (vector unsigned char, vector bool char);
8590 vector unsigned char vec_xor (vector unsigned char,
8591 vector unsigned char);
8593 int vec_all_eq (vector signed char, vector bool char);
8594 int vec_all_eq (vector signed char, vector signed char);
8595 int vec_all_eq (vector unsigned char, vector bool char);
8596 int vec_all_eq (vector unsigned char, vector unsigned char);
8597 int vec_all_eq (vector bool char, vector bool char);
8598 int vec_all_eq (vector bool char, vector unsigned char);
8599 int vec_all_eq (vector bool char, vector signed char);
8600 int vec_all_eq (vector signed short, vector bool short);
8601 int vec_all_eq (vector signed short, vector signed short);
8602 int vec_all_eq (vector unsigned short, vector bool short);
8603 int vec_all_eq (vector unsigned short, vector unsigned short);
8604 int vec_all_eq (vector bool short, vector bool short);
8605 int vec_all_eq (vector bool short, vector unsigned short);
8606 int vec_all_eq (vector bool short, vector signed short);
8607 int vec_all_eq (vector pixel, vector pixel);
8608 int vec_all_eq (vector signed int, vector bool int);
8609 int vec_all_eq (vector signed int, vector signed int);
8610 int vec_all_eq (vector unsigned int, vector bool int);
8611 int vec_all_eq (vector unsigned int, vector unsigned int);
8612 int vec_all_eq (vector bool int, vector bool int);
8613 int vec_all_eq (vector bool int, vector unsigned int);
8614 int vec_all_eq (vector bool int, vector signed int);
8615 int vec_all_eq (vector float, vector float);
8617 int vec_all_ge (vector bool char, vector unsigned char);
8618 int vec_all_ge (vector unsigned char, vector bool char);
8619 int vec_all_ge (vector unsigned char, vector unsigned char);
8620 int vec_all_ge (vector bool char, vector signed char);
8621 int vec_all_ge (vector signed char, vector bool char);
8622 int vec_all_ge (vector signed char, vector signed char);
8623 int vec_all_ge (vector bool short, vector unsigned short);
8624 int vec_all_ge (vector unsigned short, vector bool short);
8625 int vec_all_ge (vector unsigned short, vector unsigned short);
8626 int vec_all_ge (vector signed short, vector signed short);
8627 int vec_all_ge (vector bool short, vector signed short);
8628 int vec_all_ge (vector signed short, vector bool short);
8629 int vec_all_ge (vector bool int, vector unsigned int);
8630 int vec_all_ge (vector unsigned int, vector bool int);
8631 int vec_all_ge (vector unsigned int, vector unsigned int);
8632 int vec_all_ge (vector bool int, vector signed int);
8633 int vec_all_ge (vector signed int, vector bool int);
8634 int vec_all_ge (vector signed int, vector signed int);
8635 int vec_all_ge (vector float, vector float);
8637 int vec_all_gt (vector bool char, vector unsigned char);
8638 int vec_all_gt (vector unsigned char, vector bool char);
8639 int vec_all_gt (vector unsigned char, vector unsigned char);
8640 int vec_all_gt (vector bool char, vector signed char);
8641 int vec_all_gt (vector signed char, vector bool char);
8642 int vec_all_gt (vector signed char, vector signed char);
8643 int vec_all_gt (vector bool short, vector unsigned short);
8644 int vec_all_gt (vector unsigned short, vector bool short);
8645 int vec_all_gt (vector unsigned short, vector unsigned short);
8646 int vec_all_gt (vector bool short, vector signed short);
8647 int vec_all_gt (vector signed short, vector bool short);
8648 int vec_all_gt (vector signed short, vector signed short);
8649 int vec_all_gt (vector bool int, vector unsigned int);
8650 int vec_all_gt (vector unsigned int, vector bool int);
8651 int vec_all_gt (vector unsigned int, vector unsigned int);
8652 int vec_all_gt (vector bool int, vector signed int);
8653 int vec_all_gt (vector signed int, vector bool int);
8654 int vec_all_gt (vector signed int, vector signed int);
8655 int vec_all_gt (vector float, vector float);
8657 int vec_all_in (vector float, vector float);
8659 int vec_all_le (vector bool char, vector unsigned char);
8660 int vec_all_le (vector unsigned char, vector bool char);
8661 int vec_all_le (vector unsigned char, vector unsigned char);
8662 int vec_all_le (vector bool char, vector signed char);
8663 int vec_all_le (vector signed char, vector bool char);
8664 int vec_all_le (vector signed char, vector signed char);
8665 int vec_all_le (vector bool short, vector unsigned short);
8666 int vec_all_le (vector unsigned short, vector bool short);
8667 int vec_all_le (vector unsigned short, vector unsigned short);
8668 int vec_all_le (vector bool short, vector signed short);
8669 int vec_all_le (vector signed short, vector bool short);
8670 int vec_all_le (vector signed short, vector signed short);
8671 int vec_all_le (vector bool int, vector unsigned int);
8672 int vec_all_le (vector unsigned int, vector bool int);
8673 int vec_all_le (vector unsigned int, vector unsigned int);
8674 int vec_all_le (vector bool int, vector signed int);
8675 int vec_all_le (vector signed int, vector bool int);
8676 int vec_all_le (vector signed int, vector signed int);
8677 int vec_all_le (vector float, vector float);
8679 int vec_all_lt (vector bool char, vector unsigned char);
8680 int vec_all_lt (vector unsigned char, vector bool char);
8681 int vec_all_lt (vector unsigned char, vector unsigned char);
8682 int vec_all_lt (vector bool char, vector signed char);
8683 int vec_all_lt (vector signed char, vector bool char);
8684 int vec_all_lt (vector signed char, vector signed char);
8685 int vec_all_lt (vector bool short, vector unsigned short);
8686 int vec_all_lt (vector unsigned short, vector bool short);
8687 int vec_all_lt (vector unsigned short, vector unsigned short);
8688 int vec_all_lt (vector bool short, vector signed short);
8689 int vec_all_lt (vector signed short, vector bool short);
8690 int vec_all_lt (vector signed short, vector signed short);
8691 int vec_all_lt (vector bool int, vector unsigned int);
8692 int vec_all_lt (vector unsigned int, vector bool int);
8693 int vec_all_lt (vector unsigned int, vector unsigned int);
8694 int vec_all_lt (vector bool int, vector signed int);
8695 int vec_all_lt (vector signed int, vector bool int);
8696 int vec_all_lt (vector signed int, vector signed int);
8697 int vec_all_lt (vector float, vector float);
8699 int vec_all_nan (vector float);
8701 int vec_all_ne (vector signed char, vector bool char);
8702 int vec_all_ne (vector signed char, vector signed char);
8703 int vec_all_ne (vector unsigned char, vector bool char);
8704 int vec_all_ne (vector unsigned char, vector unsigned char);
8705 int vec_all_ne (vector bool char, vector bool char);
8706 int vec_all_ne (vector bool char, vector unsigned char);
8707 int vec_all_ne (vector bool char, vector signed char);
8708 int vec_all_ne (vector signed short, vector bool short);
8709 int vec_all_ne (vector signed short, vector signed short);
8710 int vec_all_ne (vector unsigned short, vector bool short);
8711 int vec_all_ne (vector unsigned short, vector unsigned short);
8712 int vec_all_ne (vector bool short, vector bool short);
8713 int vec_all_ne (vector bool short, vector unsigned short);
8714 int vec_all_ne (vector bool short, vector signed short);
8715 int vec_all_ne (vector pixel, vector pixel);
8716 int vec_all_ne (vector signed int, vector bool int);
8717 int vec_all_ne (vector signed int, vector signed int);
8718 int vec_all_ne (vector unsigned int, vector bool int);
8719 int vec_all_ne (vector unsigned int, vector unsigned int);
8720 int vec_all_ne (vector bool int, vector bool int);
8721 int vec_all_ne (vector bool int, vector unsigned int);
8722 int vec_all_ne (vector bool int, vector signed int);
8723 int vec_all_ne (vector float, vector float);
8725 int vec_all_nge (vector float, vector float);
8727 int vec_all_ngt (vector float, vector float);
8729 int vec_all_nle (vector float, vector float);
8731 int vec_all_nlt (vector float, vector float);
8733 int vec_all_numeric (vector float);
8735 int vec_any_eq (vector signed char, vector bool char);
8736 int vec_any_eq (vector signed char, vector signed char);
8737 int vec_any_eq (vector unsigned char, vector bool char);
8738 int vec_any_eq (vector unsigned char, vector unsigned char);
8739 int vec_any_eq (vector bool char, vector bool char);
8740 int vec_any_eq (vector bool char, vector unsigned char);
8741 int vec_any_eq (vector bool char, vector signed char);
8742 int vec_any_eq (vector signed short, vector bool short);
8743 int vec_any_eq (vector signed short, vector signed short);
8744 int vec_any_eq (vector unsigned short, vector bool short);
8745 int vec_any_eq (vector unsigned short, vector unsigned short);
8746 int vec_any_eq (vector bool short, vector bool short);
8747 int vec_any_eq (vector bool short, vector unsigned short);
8748 int vec_any_eq (vector bool short, vector signed short);
8749 int vec_any_eq (vector pixel, vector pixel);
8750 int vec_any_eq (vector signed int, vector bool int);
8751 int vec_any_eq (vector signed int, vector signed int);
8752 int vec_any_eq (vector unsigned int, vector bool int);
8753 int vec_any_eq (vector unsigned int, vector unsigned int);
8754 int vec_any_eq (vector bool int, vector bool int);
8755 int vec_any_eq (vector bool int, vector unsigned int);
8756 int vec_any_eq (vector bool int, vector signed int);
8757 int vec_any_eq (vector float, vector float);
8759 int vec_any_ge (vector signed char, vector bool char);
8760 int vec_any_ge (vector unsigned char, vector bool char);
8761 int vec_any_ge (vector unsigned char, vector unsigned char);
8762 int vec_any_ge (vector signed char, vector signed char);
8763 int vec_any_ge (vector bool char, vector unsigned char);
8764 int vec_any_ge (vector bool char, vector signed char);
8765 int vec_any_ge (vector unsigned short, vector bool short);
8766 int vec_any_ge (vector unsigned short, vector unsigned short);
8767 int vec_any_ge (vector signed short, vector signed short);
8768 int vec_any_ge (vector signed short, vector bool short);
8769 int vec_any_ge (vector bool short, vector unsigned short);
8770 int vec_any_ge (vector bool short, vector signed short);
8771 int vec_any_ge (vector signed int, vector bool int);
8772 int vec_any_ge (vector unsigned int, vector bool int);
8773 int vec_any_ge (vector unsigned int, vector unsigned int);
8774 int vec_any_ge (vector signed int, vector signed int);
8775 int vec_any_ge (vector bool int, vector unsigned int);
8776 int vec_any_ge (vector bool int, vector signed int);
8777 int vec_any_ge (vector float, vector float);
8779 int vec_any_gt (vector bool char, vector unsigned char);
8780 int vec_any_gt (vector unsigned char, vector bool char);
8781 int vec_any_gt (vector unsigned char, vector unsigned char);
8782 int vec_any_gt (vector bool char, vector signed char);
8783 int vec_any_gt (vector signed char, vector bool char);
8784 int vec_any_gt (vector signed char, vector signed char);
8785 int vec_any_gt (vector bool short, vector unsigned short);
8786 int vec_any_gt (vector unsigned short, vector bool short);
8787 int vec_any_gt (vector unsigned short, vector unsigned short);
8788 int vec_any_gt (vector bool short, vector signed short);
8789 int vec_any_gt (vector signed short, vector bool short);
8790 int vec_any_gt (vector signed short, vector signed short);
8791 int vec_any_gt (vector bool int, vector unsigned int);
8792 int vec_any_gt (vector unsigned int, vector bool int);
8793 int vec_any_gt (vector unsigned int, vector unsigned int);
8794 int vec_any_gt (vector bool int, vector signed int);
8795 int vec_any_gt (vector signed int, vector bool int);
8796 int vec_any_gt (vector signed int, vector signed int);
8797 int vec_any_gt (vector float, vector float);
8799 int vec_any_le (vector bool char, vector unsigned char);
8800 int vec_any_le (vector unsigned char, vector bool char);
8801 int vec_any_le (vector unsigned char, vector unsigned char);
8802 int vec_any_le (vector bool char, vector signed char);
8803 int vec_any_le (vector signed char, vector bool char);
8804 int vec_any_le (vector signed char, vector signed char);
8805 int vec_any_le (vector bool short, vector unsigned short);
8806 int vec_any_le (vector unsigned short, vector bool short);
8807 int vec_any_le (vector unsigned short, vector unsigned short);
8808 int vec_any_le (vector bool short, vector signed short);
8809 int vec_any_le (vector signed short, vector bool short);
8810 int vec_any_le (vector signed short, vector signed short);
8811 int vec_any_le (vector bool int, vector unsigned int);
8812 int vec_any_le (vector unsigned int, vector bool int);
8813 int vec_any_le (vector unsigned int, vector unsigned int);
8814 int vec_any_le (vector bool int, vector signed int);
8815 int vec_any_le (vector signed int, vector bool int);
8816 int vec_any_le (vector signed int, vector signed int);
8817 int vec_any_le (vector float, vector float);
8819 int vec_any_lt (vector bool char, vector unsigned char);
8820 int vec_any_lt (vector unsigned char, vector bool char);
8821 int vec_any_lt (vector unsigned char, vector unsigned char);
8822 int vec_any_lt (vector bool char, vector signed char);
8823 int vec_any_lt (vector signed char, vector bool char);
8824 int vec_any_lt (vector signed char, vector signed char);
8825 int vec_any_lt (vector bool short, vector unsigned short);
8826 int vec_any_lt (vector unsigned short, vector bool short);
8827 int vec_any_lt (vector unsigned short, vector unsigned short);
8828 int vec_any_lt (vector bool short, vector signed short);
8829 int vec_any_lt (vector signed short, vector bool short);
8830 int vec_any_lt (vector signed short, vector signed short);
8831 int vec_any_lt (vector bool int, vector unsigned int);
8832 int vec_any_lt (vector unsigned int, vector bool int);
8833 int vec_any_lt (vector unsigned int, vector unsigned int);
8834 int vec_any_lt (vector bool int, vector signed int);
8835 int vec_any_lt (vector signed int, vector bool int);
8836 int vec_any_lt (vector signed int, vector signed int);
8837 int vec_any_lt (vector float, vector float);
8839 int vec_any_nan (vector float);
8841 int vec_any_ne (vector signed char, vector bool char);
8842 int vec_any_ne (vector signed char, vector signed char);
8843 int vec_any_ne (vector unsigned char, vector bool char);
8844 int vec_any_ne (vector unsigned char, vector unsigned char);
8845 int vec_any_ne (vector bool char, vector bool char);
8846 int vec_any_ne (vector bool char, vector unsigned char);
8847 int vec_any_ne (vector bool char, vector signed char);
8848 int vec_any_ne (vector signed short, vector bool short);
8849 int vec_any_ne (vector signed short, vector signed short);
8850 int vec_any_ne (vector unsigned short, vector bool short);
8851 int vec_any_ne (vector unsigned short, vector unsigned short);
8852 int vec_any_ne (vector bool short, vector bool short);
8853 int vec_any_ne (vector bool short, vector unsigned short);
8854 int vec_any_ne (vector bool short, vector signed short);
8855 int vec_any_ne (vector pixel, vector pixel);
8856 int vec_any_ne (vector signed int, vector bool int);
8857 int vec_any_ne (vector signed int, vector signed int);
8858 int vec_any_ne (vector unsigned int, vector bool int);
8859 int vec_any_ne (vector unsigned int, vector unsigned int);
8860 int vec_any_ne (vector bool int, vector bool int);
8861 int vec_any_ne (vector bool int, vector unsigned int);
8862 int vec_any_ne (vector bool int, vector signed int);
8863 int vec_any_ne (vector float, vector float);
8865 int vec_any_nge (vector float, vector float);
8867 int vec_any_ngt (vector float, vector float);
8869 int vec_any_nle (vector float, vector float);
8871 int vec_any_nlt (vector float, vector float);
8873 int vec_any_numeric (vector float);
8875 int vec_any_out (vector float, vector float);
8878 @node SPARC VIS Built-in Functions
8879 @subsection SPARC VIS Built-in Functions
8881 GCC supports SIMD operations on the SPARC using both the generic vector
8882 extensions (@pxref{Vector Extensions}) as well as built-in functions for
8883 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
8884 switch, the VIS extension is exposed as the following built-in functions:
8887 typedef int v2si __attribute__ ((vector_size (8)));
8888 typedef short v4hi __attribute__ ((vector_size (8)));
8889 typedef short v2hi __attribute__ ((vector_size (4)));
8890 typedef char v8qi __attribute__ ((vector_size (8)));
8891 typedef char v4qi __attribute__ ((vector_size (4)));
8893 void * __builtin_vis_alignaddr (void *, long);
8894 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
8895 v2si __builtin_vis_faligndatav2si (v2si, v2si);
8896 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
8897 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
8899 v4hi __builtin_vis_fexpand (v4qi);
8901 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
8902 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
8903 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
8904 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
8905 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
8906 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
8907 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
8909 v4qi __builtin_vis_fpack16 (v4hi);
8910 v8qi __builtin_vis_fpack32 (v2si, v2si);
8911 v2hi __builtin_vis_fpackfix (v2si);
8912 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
8914 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
8917 @node Target Format Checks
8918 @section Format Checks Specific to Particular Target Machines
8920 For some target machines, GCC supports additional options to the
8922 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
8925 * Solaris Format Checks::
8928 @node Solaris Format Checks
8929 @subsection Solaris Format Checks
8931 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
8932 check. @code{cmn_err} accepts a subset of the standard @code{printf}
8933 conversions, and the two-argument @code{%b} conversion for displaying
8934 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
8937 @section Pragmas Accepted by GCC
8941 GCC supports several types of pragmas, primarily in order to compile
8942 code originally written for other compilers. Note that in general
8943 we do not recommend the use of pragmas; @xref{Function Attributes},
8944 for further explanation.
8948 * RS/6000 and PowerPC Pragmas::
8951 * Symbol-Renaming Pragmas::
8952 * Structure-Packing Pragmas::
8957 @subsection ARM Pragmas
8959 The ARM target defines pragmas for controlling the default addition of
8960 @code{long_call} and @code{short_call} attributes to functions.
8961 @xref{Function Attributes}, for information about the effects of these
8966 @cindex pragma, long_calls
8967 Set all subsequent functions to have the @code{long_call} attribute.
8970 @cindex pragma, no_long_calls
8971 Set all subsequent functions to have the @code{short_call} attribute.
8973 @item long_calls_off
8974 @cindex pragma, long_calls_off
8975 Do not affect the @code{long_call} or @code{short_call} attributes of
8976 subsequent functions.
8979 @node RS/6000 and PowerPC Pragmas
8980 @subsection RS/6000 and PowerPC Pragmas
8982 The RS/6000 and PowerPC targets define one pragma for controlling
8983 whether or not the @code{longcall} attribute is added to function
8984 declarations by default. This pragma overrides the @option{-mlongcall}
8985 option, but not the @code{longcall} and @code{shortcall} attributes.
8986 @xref{RS/6000 and PowerPC Options}, for more information about when long
8987 calls are and are not necessary.
8991 @cindex pragma, longcall
8992 Apply the @code{longcall} attribute to all subsequent function
8996 Do not apply the @code{longcall} attribute to subsequent function
9000 @c Describe c4x pragmas here.
9001 @c Describe h8300 pragmas here.
9002 @c Describe sh pragmas here.
9003 @c Describe v850 pragmas here.
9005 @node Darwin Pragmas
9006 @subsection Darwin Pragmas
9008 The following pragmas are available for all architectures running the
9009 Darwin operating system. These are useful for compatibility with other
9013 @item mark @var{tokens}@dots{}
9014 @cindex pragma, mark
9015 This pragma is accepted, but has no effect.
9017 @item options align=@var{alignment}
9018 @cindex pragma, options align
9019 This pragma sets the alignment of fields in structures. The values of
9020 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
9021 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
9022 properly; to restore the previous setting, use @code{reset} for the
9025 @item segment @var{tokens}@dots{}
9026 @cindex pragma, segment
9027 This pragma is accepted, but has no effect.
9029 @item unused (@var{var} [, @var{var}]@dots{})
9030 @cindex pragma, unused
9031 This pragma declares variables to be possibly unused. GCC will not
9032 produce warnings for the listed variables. The effect is similar to
9033 that of the @code{unused} attribute, except that this pragma may appear
9034 anywhere within the variables' scopes.
9037 @node Solaris Pragmas
9038 @subsection Solaris Pragmas
9040 The Solaris target supports @code{#pragma redefine_extname}
9041 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
9042 @code{#pragma} directives for compatibility with the system compiler.
9045 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
9046 @cindex pragma, align
9048 Increase the minimum alignment of each @var{variable} to @var{alignment}.
9049 This is the same as GCC's @code{aligned} attribute @pxref{Variable
9050 Attributes}). Macro expansion occurs on the arguments to this pragma
9051 when compiling C and Objective-C. It does not currently occur when
9052 compiling C++, but this is a bug which may be fixed in a future
9055 @item fini (@var{function} [, @var{function}]...)
9056 @cindex pragma, fini
9058 This pragma causes each listed @var{function} to be called after
9059 main, or during shared module unloading, by adding a call to the
9060 @code{.fini} section.
9062 @item init (@var{function} [, @var{function}]...)
9063 @cindex pragma, init
9065 This pragma causes each listed @var{function} to be called during
9066 initialization (before @code{main}) or during shared module loading, by
9067 adding a call to the @code{.init} section.
9071 @node Symbol-Renaming Pragmas
9072 @subsection Symbol-Renaming Pragmas
9074 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
9075 supports two @code{#pragma} directives which change the name used in
9076 assembly for a given declaration. These pragmas are only available on
9077 platforms whose system headers need them. To get this effect on all
9078 platforms supported by GCC, use the asm labels extension (@pxref{Asm
9082 @item redefine_extname @var{oldname} @var{newname}
9083 @cindex pragma, redefine_extname
9085 This pragma gives the C function @var{oldname} the assembly symbol
9086 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
9087 will be defined if this pragma is available (currently only on
9090 @item extern_prefix @var{string}
9091 @cindex pragma, extern_prefix
9093 This pragma causes all subsequent external function and variable
9094 declarations to have @var{string} prepended to their assembly symbols.
9095 This effect may be terminated with another @code{extern_prefix} pragma
9096 whose argument is an empty string. The preprocessor macro
9097 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
9098 available (currently only on Tru64 UNIX)@.
9101 These pragmas and the asm labels extension interact in a complicated
9102 manner. Here are some corner cases you may want to be aware of.
9105 @item Both pragmas silently apply only to declarations with external
9106 linkage. Asm labels do not have this restriction.
9108 @item In C++, both pragmas silently apply only to declarations with
9109 ``C'' linkage. Again, asm labels do not have this restriction.
9111 @item If any of the three ways of changing the assembly name of a
9112 declaration is applied to a declaration whose assembly name has
9113 already been determined (either by a previous use of one of these
9114 features, or because the compiler needed the assembly name in order to
9115 generate code), and the new name is different, a warning issues and
9116 the name does not change.
9118 @item The @var{oldname} used by @code{#pragma redefine_extname} is
9119 always the C-language name.
9121 @item If @code{#pragma extern_prefix} is in effect, and a declaration
9122 occurs with an asm label attached, the prefix is silently ignored for
9125 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
9126 apply to the same declaration, whichever triggered first wins, and a
9127 warning issues if they contradict each other. (We would like to have
9128 @code{#pragma redefine_extname} always win, for consistency with asm
9129 labels, but if @code{#pragma extern_prefix} triggers first we have no
9130 way of knowing that that happened.)
9133 @node Structure-Packing Pragmas
9134 @subsection Structure-Packing Pragmas
9136 For compatibility with Win32, GCC supports a set of @code{#pragma}
9137 directives which change the maximum alignment of members of structures,
9138 unions, and classes subsequently defined. The @var{n} value below always
9139 is required to be a small power of two and specifies the new alignment
9143 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
9144 @item @code{#pragma pack()} sets the alignment to the one that was in
9145 effect when compilation started (see also command line option
9146 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
9147 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
9148 setting on an internal stack and then optionally sets the new alignment.
9149 @item @code{#pragma pack(pop)} restores the alignment setting to the one
9150 saved at the top of the internal stack (and removes that stack entry).
9151 Note that @code{#pragma pack([@var{n}])} does not influence this internal
9152 stack; thus it is possible to have @code{#pragma pack(push)} followed by
9153 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
9154 @code{#pragma pack(pop)}.
9158 @subsection Weak Pragmas
9160 For compatibility with SVR4, GCC supports a set of @code{#pragma}
9161 directives for declaring symbols to be weak, and defining weak
9165 @item #pragma weak @var{symbol}
9166 @cindex pragma, weak
9167 This pragma declares @var{symbol} to be weak, as if the declaration
9168 had the attribute of the same name. The pragma may appear before
9169 or after the declaration of @var{symbol}, but must appear before
9170 either its first use or its definition. It is not an error for
9171 @var{symbol} to never be defined at all.
9173 @item #pragma weak @var{symbol1} = @var{symbol2}
9174 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
9175 It is an error if @var{symbol2} is not defined in the current
9179 @node Unnamed Fields
9180 @section Unnamed struct/union fields within structs/unions
9184 For compatibility with other compilers, GCC allows you to define
9185 a structure or union that contains, as fields, structures and unions
9186 without names. For example:
9199 In this example, the user would be able to access members of the unnamed
9200 union with code like @samp{foo.b}. Note that only unnamed structs and
9201 unions are allowed, you may not have, for example, an unnamed
9204 You must never create such structures that cause ambiguous field definitions.
9205 For example, this structure:
9216 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
9217 Such constructs are not supported and must be avoided. In the future,
9218 such constructs may be detected and treated as compilation errors.
9220 @opindex fms-extensions
9221 Unless @option{-fms-extensions} is used, the unnamed field must be a
9222 structure or union definition without a tag (for example, @samp{struct
9223 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
9224 also be a definition with a tag such as @samp{struct foo @{ int a;
9225 @};}, a reference to a previously defined structure or union such as
9226 @samp{struct foo;}, or a reference to a @code{typedef} name for a
9227 previously defined structure or union type.
9230 @section Thread-Local Storage
9231 @cindex Thread-Local Storage
9232 @cindex @acronym{TLS}
9235 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
9236 are allocated such that there is one instance of the variable per extant
9237 thread. The run-time model GCC uses to implement this originates
9238 in the IA-64 processor-specific ABI, but has since been migrated
9239 to other processors as well. It requires significant support from
9240 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
9241 system libraries (@file{libc.so} and @file{libpthread.so}), so it
9242 is not available everywhere.
9244 At the user level, the extension is visible with a new storage
9245 class keyword: @code{__thread}. For example:
9249 extern __thread struct state s;
9250 static __thread char *p;
9253 The @code{__thread} specifier may be used alone, with the @code{extern}
9254 or @code{static} specifiers, but with no other storage class specifier.
9255 When used with @code{extern} or @code{static}, @code{__thread} must appear
9256 immediately after the other storage class specifier.
9258 The @code{__thread} specifier may be applied to any global, file-scoped
9259 static, function-scoped static, or static data member of a class. It may
9260 not be applied to block-scoped automatic or non-static data member.
9262 When the address-of operator is applied to a thread-local variable, it is
9263 evaluated at run-time and returns the address of the current thread's
9264 instance of that variable. An address so obtained may be used by any
9265 thread. When a thread terminates, any pointers to thread-local variables
9266 in that thread become invalid.
9268 No static initialization may refer to the address of a thread-local variable.
9270 In C++, if an initializer is present for a thread-local variable, it must
9271 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
9274 See @uref{http://people.redhat.com/drepper/tls.pdf,
9275 ELF Handling For Thread-Local Storage} for a detailed explanation of
9276 the four thread-local storage addressing models, and how the run-time
9277 is expected to function.
9280 * C99 Thread-Local Edits::
9281 * C++98 Thread-Local Edits::
9284 @node C99 Thread-Local Edits
9285 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
9287 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
9288 that document the exact semantics of the language extension.
9292 @cite{5.1.2 Execution environments}
9294 Add new text after paragraph 1
9297 Within either execution environment, a @dfn{thread} is a flow of
9298 control within a program. It is implementation defined whether
9299 or not there may be more than one thread associated with a program.
9300 It is implementation defined how threads beyond the first are
9301 created, the name and type of the function called at thread
9302 startup, and how threads may be terminated. However, objects
9303 with thread storage duration shall be initialized before thread
9308 @cite{6.2.4 Storage durations of objects}
9310 Add new text before paragraph 3
9313 An object whose identifier is declared with the storage-class
9314 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
9315 Its lifetime is the entire execution of the thread, and its
9316 stored value is initialized only once, prior to thread startup.
9320 @cite{6.4.1 Keywords}
9322 Add @code{__thread}.
9325 @cite{6.7.1 Storage-class specifiers}
9327 Add @code{__thread} to the list of storage class specifiers in
9330 Change paragraph 2 to
9333 With the exception of @code{__thread}, at most one storage-class
9334 specifier may be given [@dots{}]. The @code{__thread} specifier may
9335 be used alone, or immediately following @code{extern} or
9339 Add new text after paragraph 6
9342 The declaration of an identifier for a variable that has
9343 block scope that specifies @code{__thread} shall also
9344 specify either @code{extern} or @code{static}.
9346 The @code{__thread} specifier shall be used only with
9351 @node C++98 Thread-Local Edits
9352 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
9354 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
9355 that document the exact semantics of the language extension.
9359 @b{[intro.execution]}
9361 New text after paragraph 4
9364 A @dfn{thread} is a flow of control within the abstract machine.
9365 It is implementation defined whether or not there may be more than
9369 New text after paragraph 7
9372 It is unspecified whether additional action must be taken to
9373 ensure when and whether side effects are visible to other threads.
9379 Add @code{__thread}.
9382 @b{[basic.start.main]}
9384 Add after paragraph 5
9387 The thread that begins execution at the @code{main} function is called
9388 the @dfn{main thread}. It is implementation defined how functions
9389 beginning threads other than the main thread are designated or typed.
9390 A function so designated, as well as the @code{main} function, is called
9391 a @dfn{thread startup function}. It is implementation defined what
9392 happens if a thread startup function returns. It is implementation
9393 defined what happens to other threads when any thread calls @code{exit}.
9397 @b{[basic.start.init]}
9399 Add after paragraph 4
9402 The storage for an object of thread storage duration shall be
9403 statically initialized before the first statement of the thread startup
9404 function. An object of thread storage duration shall not require
9405 dynamic initialization.
9409 @b{[basic.start.term]}
9411 Add after paragraph 3
9414 The type of an object with thread storage duration shall not have a
9415 non-trivial destructor, nor shall it be an array type whose elements
9416 (directly or indirectly) have non-trivial destructors.
9422 Add ``thread storage duration'' to the list in paragraph 1.
9427 Thread, static, and automatic storage durations are associated with
9428 objects introduced by declarations [@dots{}].
9431 Add @code{__thread} to the list of specifiers in paragraph 3.
9434 @b{[basic.stc.thread]}
9436 New section before @b{[basic.stc.static]}
9439 The keyword @code{__thread} applied to a non-local object gives the
9440 object thread storage duration.
9442 A local variable or class data member declared both @code{static}
9443 and @code{__thread} gives the variable or member thread storage
9448 @b{[basic.stc.static]}
9453 All objects which have neither thread storage duration, dynamic
9454 storage duration nor are local [@dots{}].
9460 Add @code{__thread} to the list in paragraph 1.
9465 With the exception of @code{__thread}, at most one
9466 @var{storage-class-specifier} shall appear in a given
9467 @var{decl-specifier-seq}. The @code{__thread} specifier may
9468 be used alone, or immediately following the @code{extern} or
9469 @code{static} specifiers. [@dots{}]
9472 Add after paragraph 5
9475 The @code{__thread} specifier can be applied only to the names of objects
9476 and to anonymous unions.
9482 Add after paragraph 6
9485 Non-@code{static} members shall not be @code{__thread}.
9489 @node C++ Extensions
9490 @chapter Extensions to the C++ Language
9491 @cindex extensions, C++ language
9492 @cindex C++ language extensions
9494 The GNU compiler provides these extensions to the C++ language (and you
9495 can also use most of the C language extensions in your C++ programs). If you
9496 want to write code that checks whether these features are available, you can
9497 test for the GNU compiler the same way as for C programs: check for a
9498 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
9499 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
9500 Predefined Macros,cpp,The GNU C Preprocessor}).
9503 * Volatiles:: What constitutes an access to a volatile object.
9504 * Restricted Pointers:: C99 restricted pointers and references.
9505 * Vague Linkage:: Where G++ puts inlines, vtables and such.
9506 * C++ Interface:: You can use a single C++ header file for both
9507 declarations and definitions.
9508 * Template Instantiation:: Methods for ensuring that exactly one copy of
9509 each needed template instantiation is emitted.
9510 * Bound member functions:: You can extract a function pointer to the
9511 method denoted by a @samp{->*} or @samp{.*} expression.
9512 * C++ Attributes:: Variable, function, and type attributes for C++ only.
9513 * Strong Using:: Strong using-directives for namespace composition.
9514 * Java Exceptions:: Tweaking exception handling to work with Java.
9515 * Deprecated Features:: Things will disappear from g++.
9516 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
9520 @section When is a Volatile Object Accessed?
9521 @cindex accessing volatiles
9522 @cindex volatile read
9523 @cindex volatile write
9524 @cindex volatile access
9526 Both the C and C++ standard have the concept of volatile objects. These
9527 are normally accessed by pointers and used for accessing hardware. The
9528 standards encourage compilers to refrain from optimizations
9529 concerning accesses to volatile objects that it might perform on
9530 non-volatile objects. The C standard leaves it implementation defined
9531 as to what constitutes a volatile access. The C++ standard omits to
9532 specify this, except to say that C++ should behave in a similar manner
9533 to C with respect to volatiles, where possible. The minimum either
9534 standard specifies is that at a sequence point all previous accesses to
9535 volatile objects have stabilized and no subsequent accesses have
9536 occurred. Thus an implementation is free to reorder and combine
9537 volatile accesses which occur between sequence points, but cannot do so
9538 for accesses across a sequence point. The use of volatiles does not
9539 allow you to violate the restriction on updating objects multiple times
9540 within a sequence point.
9542 In most expressions, it is intuitively obvious what is a read and what is
9543 a write. For instance
9546 volatile int *dst = @var{somevalue};
9547 volatile int *src = @var{someothervalue};
9552 will cause a read of the volatile object pointed to by @var{src} and stores the
9553 value into the volatile object pointed to by @var{dst}. There is no
9554 guarantee that these reads and writes are atomic, especially for objects
9555 larger than @code{int}.
9557 Less obvious expressions are where something which looks like an access
9558 is used in a void context. An example would be,
9561 volatile int *src = @var{somevalue};
9565 With C, such expressions are rvalues, and as rvalues cause a read of
9566 the object, GCC interprets this as a read of the volatile being pointed
9567 to. The C++ standard specifies that such expressions do not undergo
9568 lvalue to rvalue conversion, and that the type of the dereferenced
9569 object may be incomplete. The C++ standard does not specify explicitly
9570 that it is this lvalue to rvalue conversion which is responsible for
9571 causing an access. However, there is reason to believe that it is,
9572 because otherwise certain simple expressions become undefined. However,
9573 because it would surprise most programmers, G++ treats dereferencing a
9574 pointer to volatile object of complete type in a void context as a read
9575 of the object. When the object has incomplete type, G++ issues a
9580 struct T @{int m;@};
9581 volatile S *ptr1 = @var{somevalue};
9582 volatile T *ptr2 = @var{somevalue};
9587 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
9588 causes a read of the object pointed to. If you wish to force an error on
9589 the first case, you must force a conversion to rvalue with, for instance
9590 a static cast, @code{static_cast<S>(*ptr1)}.
9592 When using a reference to volatile, G++ does not treat equivalent
9593 expressions as accesses to volatiles, but instead issues a warning that
9594 no volatile is accessed. The rationale for this is that otherwise it
9595 becomes difficult to determine where volatile access occur, and not
9596 possible to ignore the return value from functions returning volatile
9597 references. Again, if you wish to force a read, cast the reference to
9600 @node Restricted Pointers
9601 @section Restricting Pointer Aliasing
9602 @cindex restricted pointers
9603 @cindex restricted references
9604 @cindex restricted this pointer
9606 As with the C front end, G++ understands the C99 feature of restricted pointers,
9607 specified with the @code{__restrict__}, or @code{__restrict} type
9608 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
9609 language flag, @code{restrict} is not a keyword in C++.
9611 In addition to allowing restricted pointers, you can specify restricted
9612 references, which indicate that the reference is not aliased in the local
9616 void fn (int *__restrict__ rptr, int &__restrict__ rref)
9623 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
9624 @var{rref} refers to a (different) unaliased integer.
9626 You may also specify whether a member function's @var{this} pointer is
9627 unaliased by using @code{__restrict__} as a member function qualifier.
9630 void T::fn () __restrict__
9637 Within the body of @code{T::fn}, @var{this} will have the effective
9638 definition @code{T *__restrict__ const this}. Notice that the
9639 interpretation of a @code{__restrict__} member function qualifier is
9640 different to that of @code{const} or @code{volatile} qualifier, in that it
9641 is applied to the pointer rather than the object. This is consistent with
9642 other compilers which implement restricted pointers.
9644 As with all outermost parameter qualifiers, @code{__restrict__} is
9645 ignored in function definition matching. This means you only need to
9646 specify @code{__restrict__} in a function definition, rather than
9647 in a function prototype as well.
9650 @section Vague Linkage
9651 @cindex vague linkage
9653 There are several constructs in C++ which require space in the object
9654 file but are not clearly tied to a single translation unit. We say that
9655 these constructs have ``vague linkage''. Typically such constructs are
9656 emitted wherever they are needed, though sometimes we can be more
9660 @item Inline Functions
9661 Inline functions are typically defined in a header file which can be
9662 included in many different compilations. Hopefully they can usually be
9663 inlined, but sometimes an out-of-line copy is necessary, if the address
9664 of the function is taken or if inlining fails. In general, we emit an
9665 out-of-line copy in all translation units where one is needed. As an
9666 exception, we only emit inline virtual functions with the vtable, since
9667 it will always require a copy.
9669 Local static variables and string constants used in an inline function
9670 are also considered to have vague linkage, since they must be shared
9671 between all inlined and out-of-line instances of the function.
9675 C++ virtual functions are implemented in most compilers using a lookup
9676 table, known as a vtable. The vtable contains pointers to the virtual
9677 functions provided by a class, and each object of the class contains a
9678 pointer to its vtable (or vtables, in some multiple-inheritance
9679 situations). If the class declares any non-inline, non-pure virtual
9680 functions, the first one is chosen as the ``key method'' for the class,
9681 and the vtable is only emitted in the translation unit where the key
9684 @emph{Note:} If the chosen key method is later defined as inline, the
9685 vtable will still be emitted in every translation unit which defines it.
9686 Make sure that any inline virtuals are declared inline in the class
9687 body, even if they are not defined there.
9689 @item type_info objects
9692 C++ requires information about types to be written out in order to
9693 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
9694 For polymorphic classes (classes with virtual functions), the type_info
9695 object is written out along with the vtable so that @samp{dynamic_cast}
9696 can determine the dynamic type of a class object at runtime. For all
9697 other types, we write out the type_info object when it is used: when
9698 applying @samp{typeid} to an expression, throwing an object, or
9699 referring to a type in a catch clause or exception specification.
9701 @item Template Instantiations
9702 Most everything in this section also applies to template instantiations,
9703 but there are other options as well.
9704 @xref{Template Instantiation,,Where's the Template?}.
9708 When used with GNU ld version 2.8 or later on an ELF system such as
9709 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
9710 these constructs will be discarded at link time. This is known as
9713 On targets that don't support COMDAT, but do support weak symbols, GCC
9714 will use them. This way one copy will override all the others, but
9715 the unused copies will still take up space in the executable.
9717 For targets which do not support either COMDAT or weak symbols,
9718 most entities with vague linkage will be emitted as local symbols to
9719 avoid duplicate definition errors from the linker. This will not happen
9720 for local statics in inlines, however, as having multiple copies will
9721 almost certainly break things.
9723 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
9724 another way to control placement of these constructs.
9727 @section #pragma interface and implementation
9729 @cindex interface and implementation headers, C++
9730 @cindex C++ interface and implementation headers
9731 @cindex pragmas, interface and implementation
9733 @code{#pragma interface} and @code{#pragma implementation} provide the
9734 user with a way of explicitly directing the compiler to emit entities
9735 with vague linkage (and debugging information) in a particular
9738 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
9739 most cases, because of COMDAT support and the ``key method'' heuristic
9740 mentioned in @ref{Vague Linkage}. Using them can actually cause your
9741 program to grow due to unnecessary out-of-line copies of inline
9742 functions. Currently (3.4) the only benefit of these
9743 @code{#pragma}s is reduced duplication of debugging information, and
9744 that should be addressed soon on DWARF 2 targets with the use of
9748 @item #pragma interface
9749 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
9750 @kindex #pragma interface
9751 Use this directive in @emph{header files} that define object classes, to save
9752 space in most of the object files that use those classes. Normally,
9753 local copies of certain information (backup copies of inline member
9754 functions, debugging information, and the internal tables that implement
9755 virtual functions) must be kept in each object file that includes class
9756 definitions. You can use this pragma to avoid such duplication. When a
9757 header file containing @samp{#pragma interface} is included in a
9758 compilation, this auxiliary information will not be generated (unless
9759 the main input source file itself uses @samp{#pragma implementation}).
9760 Instead, the object files will contain references to be resolved at link
9763 The second form of this directive is useful for the case where you have
9764 multiple headers with the same name in different directories. If you
9765 use this form, you must specify the same string to @samp{#pragma
9768 @item #pragma implementation
9769 @itemx #pragma implementation "@var{objects}.h"
9770 @kindex #pragma implementation
9771 Use this pragma in a @emph{main input file}, when you want full output from
9772 included header files to be generated (and made globally visible). The
9773 included header file, in turn, should use @samp{#pragma interface}.
9774 Backup copies of inline member functions, debugging information, and the
9775 internal tables used to implement virtual functions are all generated in
9776 implementation files.
9778 @cindex implied @code{#pragma implementation}
9779 @cindex @code{#pragma implementation}, implied
9780 @cindex naming convention, implementation headers
9781 If you use @samp{#pragma implementation} with no argument, it applies to
9782 an include file with the same basename@footnote{A file's @dfn{basename}
9783 was the name stripped of all leading path information and of trailing
9784 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
9785 file. For example, in @file{allclass.cc}, giving just
9786 @samp{#pragma implementation}
9787 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
9789 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
9790 an implementation file whenever you would include it from
9791 @file{allclass.cc} even if you never specified @samp{#pragma
9792 implementation}. This was deemed to be more trouble than it was worth,
9793 however, and disabled.
9795 Use the string argument if you want a single implementation file to
9796 include code from multiple header files. (You must also use
9797 @samp{#include} to include the header file; @samp{#pragma
9798 implementation} only specifies how to use the file---it doesn't actually
9801 There is no way to split up the contents of a single header file into
9802 multiple implementation files.
9805 @cindex inlining and C++ pragmas
9806 @cindex C++ pragmas, effect on inlining
9807 @cindex pragmas in C++, effect on inlining
9808 @samp{#pragma implementation} and @samp{#pragma interface} also have an
9809 effect on function inlining.
9811 If you define a class in a header file marked with @samp{#pragma
9812 interface}, the effect on an inline function defined in that class is
9813 similar to an explicit @code{extern} declaration---the compiler emits
9814 no code at all to define an independent version of the function. Its
9815 definition is used only for inlining with its callers.
9817 @opindex fno-implement-inlines
9818 Conversely, when you include the same header file in a main source file
9819 that declares it as @samp{#pragma implementation}, the compiler emits
9820 code for the function itself; this defines a version of the function
9821 that can be found via pointers (or by callers compiled without
9822 inlining). If all calls to the function can be inlined, you can avoid
9823 emitting the function by compiling with @option{-fno-implement-inlines}.
9824 If any calls were not inlined, you will get linker errors.
9826 @node Template Instantiation
9827 @section Where's the Template?
9828 @cindex template instantiation
9830 C++ templates are the first language feature to require more
9831 intelligence from the environment than one usually finds on a UNIX
9832 system. Somehow the compiler and linker have to make sure that each
9833 template instance occurs exactly once in the executable if it is needed,
9834 and not at all otherwise. There are two basic approaches to this
9835 problem, which are referred to as the Borland model and the Cfront model.
9839 Borland C++ solved the template instantiation problem by adding the code
9840 equivalent of common blocks to their linker; the compiler emits template
9841 instances in each translation unit that uses them, and the linker
9842 collapses them together. The advantage of this model is that the linker
9843 only has to consider the object files themselves; there is no external
9844 complexity to worry about. This disadvantage is that compilation time
9845 is increased because the template code is being compiled repeatedly.
9846 Code written for this model tends to include definitions of all
9847 templates in the header file, since they must be seen to be
9851 The AT&T C++ translator, Cfront, solved the template instantiation
9852 problem by creating the notion of a template repository, an
9853 automatically maintained place where template instances are stored. A
9854 more modern version of the repository works as follows: As individual
9855 object files are built, the compiler places any template definitions and
9856 instantiations encountered in the repository. At link time, the link
9857 wrapper adds in the objects in the repository and compiles any needed
9858 instances that were not previously emitted. The advantages of this
9859 model are more optimal compilation speed and the ability to use the
9860 system linker; to implement the Borland model a compiler vendor also
9861 needs to replace the linker. The disadvantages are vastly increased
9862 complexity, and thus potential for error; for some code this can be
9863 just as transparent, but in practice it can been very difficult to build
9864 multiple programs in one directory and one program in multiple
9865 directories. Code written for this model tends to separate definitions
9866 of non-inline member templates into a separate file, which should be
9867 compiled separately.
9870 When used with GNU ld version 2.8 or later on an ELF system such as
9871 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
9872 Borland model. On other systems, G++ implements neither automatic
9875 A future version of G++ will support a hybrid model whereby the compiler
9876 will emit any instantiations for which the template definition is
9877 included in the compile, and store template definitions and
9878 instantiation context information into the object file for the rest.
9879 The link wrapper will extract that information as necessary and invoke
9880 the compiler to produce the remaining instantiations. The linker will
9881 then combine duplicate instantiations.
9883 In the mean time, you have the following options for dealing with
9884 template instantiations:
9889 Compile your template-using code with @option{-frepo}. The compiler will
9890 generate files with the extension @samp{.rpo} listing all of the
9891 template instantiations used in the corresponding object files which
9892 could be instantiated there; the link wrapper, @samp{collect2}, will
9893 then update the @samp{.rpo} files to tell the compiler where to place
9894 those instantiations and rebuild any affected object files. The
9895 link-time overhead is negligible after the first pass, as the compiler
9896 will continue to place the instantiations in the same files.
9898 This is your best option for application code written for the Borland
9899 model, as it will just work. Code written for the Cfront model will
9900 need to be modified so that the template definitions are available at
9901 one or more points of instantiation; usually this is as simple as adding
9902 @code{#include <tmethods.cc>} to the end of each template header.
9904 For library code, if you want the library to provide all of the template
9905 instantiations it needs, just try to link all of its object files
9906 together; the link will fail, but cause the instantiations to be
9907 generated as a side effect. Be warned, however, that this may cause
9908 conflicts if multiple libraries try to provide the same instantiations.
9909 For greater control, use explicit instantiation as described in the next
9913 @opindex fno-implicit-templates
9914 Compile your code with @option{-fno-implicit-templates} to disable the
9915 implicit generation of template instances, and explicitly instantiate
9916 all the ones you use. This approach requires more knowledge of exactly
9917 which instances you need than do the others, but it's less
9918 mysterious and allows greater control. You can scatter the explicit
9919 instantiations throughout your program, perhaps putting them in the
9920 translation units where the instances are used or the translation units
9921 that define the templates themselves; you can put all of the explicit
9922 instantiations you need into one big file; or you can create small files
9929 template class Foo<int>;
9930 template ostream& operator <<
9931 (ostream&, const Foo<int>&);
9934 for each of the instances you need, and create a template instantiation
9937 If you are using Cfront-model code, you can probably get away with not
9938 using @option{-fno-implicit-templates} when compiling files that don't
9939 @samp{#include} the member template definitions.
9941 If you use one big file to do the instantiations, you may want to
9942 compile it without @option{-fno-implicit-templates} so you get all of the
9943 instances required by your explicit instantiations (but not by any
9944 other files) without having to specify them as well.
9946 G++ has extended the template instantiation syntax given in the ISO
9947 standard to allow forward declaration of explicit instantiations
9948 (with @code{extern}), instantiation of the compiler support data for a
9949 template class (i.e.@: the vtable) without instantiating any of its
9950 members (with @code{inline}), and instantiation of only the static data
9951 members of a template class, without the support data or member
9952 functions (with (@code{static}):
9955 extern template int max (int, int);
9956 inline template class Foo<int>;
9957 static template class Foo<int>;
9961 Do nothing. Pretend G++ does implement automatic instantiation
9962 management. Code written for the Borland model will work fine, but
9963 each translation unit will contain instances of each of the templates it
9964 uses. In a large program, this can lead to an unacceptable amount of code
9968 @node Bound member functions
9969 @section Extracting the function pointer from a bound pointer to member function
9971 @cindex pointer to member function
9972 @cindex bound pointer to member function
9974 In C++, pointer to member functions (PMFs) are implemented using a wide
9975 pointer of sorts to handle all the possible call mechanisms; the PMF
9976 needs to store information about how to adjust the @samp{this} pointer,
9977 and if the function pointed to is virtual, where to find the vtable, and
9978 where in the vtable to look for the member function. If you are using
9979 PMFs in an inner loop, you should really reconsider that decision. If
9980 that is not an option, you can extract the pointer to the function that
9981 would be called for a given object/PMF pair and call it directly inside
9982 the inner loop, to save a bit of time.
9984 Note that you will still be paying the penalty for the call through a
9985 function pointer; on most modern architectures, such a call defeats the
9986 branch prediction features of the CPU@. This is also true of normal
9987 virtual function calls.
9989 The syntax for this extension is
9993 extern int (A::*fp)();
9994 typedef int (*fptr)(A *);
9996 fptr p = (fptr)(a.*fp);
9999 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
10000 no object is needed to obtain the address of the function. They can be
10001 converted to function pointers directly:
10004 fptr p1 = (fptr)(&A::foo);
10007 @opindex Wno-pmf-conversions
10008 You must specify @option{-Wno-pmf-conversions} to use this extension.
10010 @node C++ Attributes
10011 @section C++-Specific Variable, Function, and Type Attributes
10013 Some attributes only make sense for C++ programs.
10016 @item init_priority (@var{priority})
10017 @cindex init_priority attribute
10020 In Standard C++, objects defined at namespace scope are guaranteed to be
10021 initialized in an order in strict accordance with that of their definitions
10022 @emph{in a given translation unit}. No guarantee is made for initializations
10023 across translation units. However, GNU C++ allows users to control the
10024 order of initialization of objects defined at namespace scope with the
10025 @code{init_priority} attribute by specifying a relative @var{priority},
10026 a constant integral expression currently bounded between 101 and 65535
10027 inclusive. Lower numbers indicate a higher priority.
10029 In the following example, @code{A} would normally be created before
10030 @code{B}, but the @code{init_priority} attribute has reversed that order:
10033 Some_Class A __attribute__ ((init_priority (2000)));
10034 Some_Class B __attribute__ ((init_priority (543)));
10038 Note that the particular values of @var{priority} do not matter; only their
10041 @item java_interface
10042 @cindex java_interface attribute
10044 This type attribute informs C++ that the class is a Java interface. It may
10045 only be applied to classes declared within an @code{extern "Java"} block.
10046 Calls to methods declared in this interface will be dispatched using GCJ's
10047 interface table mechanism, instead of regular virtual table dispatch.
10051 See also @xref{Strong Using}.
10054 @section Strong Using
10056 @strong{Caution:} The semantics of this extension are not fully
10057 defined. Users should refrain from using this extension as its
10058 semantics may change subtly over time. It is possible that this
10059 extension wil be removed in future versions of G++.
10061 A using-directive with @code{__attribute ((strong))} is stronger
10062 than a normal using-directive in two ways:
10066 Templates from the used namespace can be specialized as though they were members of the using namespace.
10069 The using namespace is considered an associated namespace of all
10070 templates in the used namespace for purposes of argument-dependent
10074 This is useful for composing a namespace transparently from
10075 implementation namespaces. For example:
10080 template <class T> struct A @{ @};
10082 using namespace debug __attribute ((__strong__));
10083 template <> struct A<int> @{ @}; // @r{ok to specialize}
10085 template <class T> void f (A<T>);
10090 f (std::A<float>()); // @r{lookup finds} std::f
10095 @node Java Exceptions
10096 @section Java Exceptions
10098 The Java language uses a slightly different exception handling model
10099 from C++. Normally, GNU C++ will automatically detect when you are
10100 writing C++ code that uses Java exceptions, and handle them
10101 appropriately. However, if C++ code only needs to execute destructors
10102 when Java exceptions are thrown through it, GCC will guess incorrectly.
10103 Sample problematic code is:
10106 struct S @{ ~S(); @};
10107 extern void bar(); // @r{is written in Java, and may throw exceptions}
10116 The usual effect of an incorrect guess is a link failure, complaining of
10117 a missing routine called @samp{__gxx_personality_v0}.
10119 You can inform the compiler that Java exceptions are to be used in a
10120 translation unit, irrespective of what it might think, by writing
10121 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
10122 @samp{#pragma} must appear before any functions that throw or catch
10123 exceptions, or run destructors when exceptions are thrown through them.
10125 You cannot mix Java and C++ exceptions in the same translation unit. It
10126 is believed to be safe to throw a C++ exception from one file through
10127 another file compiled for the Java exception model, or vice versa, but
10128 there may be bugs in this area.
10130 @node Deprecated Features
10131 @section Deprecated Features
10133 In the past, the GNU C++ compiler was extended to experiment with new
10134 features, at a time when the C++ language was still evolving. Now that
10135 the C++ standard is complete, some of those features are superseded by
10136 superior alternatives. Using the old features might cause a warning in
10137 some cases that the feature will be dropped in the future. In other
10138 cases, the feature might be gone already.
10140 While the list below is not exhaustive, it documents some of the options
10141 that are now deprecated:
10144 @item -fexternal-templates
10145 @itemx -falt-external-templates
10146 These are two of the many ways for G++ to implement template
10147 instantiation. @xref{Template Instantiation}. The C++ standard clearly
10148 defines how template definitions have to be organized across
10149 implementation units. G++ has an implicit instantiation mechanism that
10150 should work just fine for standard-conforming code.
10152 @item -fstrict-prototype
10153 @itemx -fno-strict-prototype
10154 Previously it was possible to use an empty prototype parameter list to
10155 indicate an unspecified number of parameters (like C), rather than no
10156 parameters, as C++ demands. This feature has been removed, except where
10157 it is required for backwards compatibility @xref{Backwards Compatibility}.
10160 G++ allows a virtual function returning @samp{void *} to be overridden
10161 by one returning a different pointer type. This extension to the
10162 covariant return type rules is now deprecated and will be removed from a
10165 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
10166 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
10167 and will be removed in a future version. Code using these operators
10168 should be modified to use @code{std::min} and @code{std::max} instead.
10170 The named return value extension has been deprecated, and is now
10173 The use of initializer lists with new expressions has been deprecated,
10174 and is now removed from G++.
10176 Floating and complex non-type template parameters have been deprecated,
10177 and are now removed from G++.
10179 The implicit typename extension has been deprecated and is now
10182 The use of default arguments in function pointers, function typedefs and
10183 and other places where they are not permitted by the standard is
10184 deprecated and will be removed from a future version of G++.
10186 G++ allows floating-point literals to appear in integral constant expressions,
10187 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
10188 This extension is deprecated and will be removed from a future version.
10190 G++ allows static data members of const floating-point type to be declared
10191 with an initializer in a class definition. The standard only allows
10192 initializers for static members of const integral types and const
10193 enumeration types so this extension has been deprecated and will be removed
10194 from a future version.
10196 @node Backwards Compatibility
10197 @section Backwards Compatibility
10198 @cindex Backwards Compatibility
10199 @cindex ARM [Annotated C++ Reference Manual]
10201 Now that there is a definitive ISO standard C++, G++ has a specification
10202 to adhere to. The C++ language evolved over time, and features that
10203 used to be acceptable in previous drafts of the standard, such as the ARM
10204 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
10205 compilation of C++ written to such drafts, G++ contains some backwards
10206 compatibilities. @emph{All such backwards compatibility features are
10207 liable to disappear in future versions of G++.} They should be considered
10208 deprecated @xref{Deprecated Features}.
10212 If a variable is declared at for scope, it used to remain in scope until
10213 the end of the scope which contained the for statement (rather than just
10214 within the for scope). G++ retains this, but issues a warning, if such a
10215 variable is accessed outside the for scope.
10217 @item Implicit C language
10218 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
10219 scope to set the language. On such systems, all header files are
10220 implicitly scoped inside a C language scope. Also, an empty prototype
10221 @code{()} will be treated as an unspecified number of arguments, rather
10222 than no arguments, as C++ demands.