1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1996, 1998, 1999, 2000,
2 @c 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
8 @chapter Extensions to the C Language Family
9 @cindex extensions, C language
10 @cindex C language extensions
13 GNU C provides several language features not found in ISO standard C@.
14 (The @option{-pedantic} option directs GCC to print a warning message if
15 any of these features is used.) To test for the availability of these
16 features in conditional compilation, check for a predefined macro
17 @code{__GNUC__}, which is always defined under GCC@.
19 These extensions are available in C. Most of them are also available
20 in C++. @xref{C++ Extensions,,Extensions to the C++ Language}, for
21 extensions that apply @emph{only} to C++.
23 Some features that are in ISO C99 but not C89 or C++ are also, as
24 extensions, accepted by GCC in C89 mode and in C++.
27 * Statement Exprs:: Putting statements and declarations inside expressions.
28 * Local Labels:: Labels local to a block.
29 * Labels as Values:: Getting pointers to labels, and computed gotos.
30 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
31 * Constructing Calls:: Dispatching a call to another function.
32 * Typeof:: @code{typeof}: referring to the type of an expression.
33 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
34 * Long Long:: Double-word integers---@code{long long int}.
35 * Complex:: Data types for complex numbers.
36 * Decimal Float:: Decimal Floating Types.
37 * Hex Floats:: Hexadecimal floating-point constants.
38 * Zero Length:: Zero-length arrays.
39 * Variable Length:: Arrays whose length is computed at run time.
40 * Empty Structures:: Structures with no members.
41 * Variadic Macros:: Macros with a variable number of arguments.
42 * Escaped Newlines:: Slightly looser rules for escaped newlines.
43 * Subscripting:: Any array can be subscripted, even if not an lvalue.
44 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
45 * Initializers:: Non-constant initializers.
46 * Compound Literals:: Compound literals give structures, unions
48 * Designated Inits:: Labeling elements of initializers.
49 * Cast to Union:: Casting to union type from any member of the union.
50 * Case Ranges:: `case 1 ... 9' and such.
51 * Mixed Declarations:: Mixing declarations and code.
52 * Function Attributes:: Declaring that functions have no side effects,
53 or that they can never return.
54 * Attribute Syntax:: Formal syntax for attributes.
55 * Function Prototypes:: Prototype declarations and old-style definitions.
56 * C++ Comments:: C++ comments are recognized.
57 * Dollar Signs:: Dollar sign is allowed in identifiers.
58 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
59 * Variable Attributes:: Specifying attributes of variables.
60 * Type Attributes:: Specifying attributes of types.
61 @c APPLE LOCAL begin for-fsf-4_4 3274130 5295549
62 * Label Attributes:: Specifying attributes of labels and statements.
63 @c APPLE LOCAL end for-fsf-4_4 3274130 5295549
64 * Alignment:: Inquiring about the alignment of a type or variable.
65 * Inline:: Defining inline functions (as fast as macros).
66 * Extended Asm:: Assembler instructions with C expressions as operands.
67 (With them you can define ``built-in'' functions.)
68 * Constraints:: Constraints for asm operands
69 * Asm Labels:: Specifying the assembler name to use for a C symbol.
70 * Explicit Reg Vars:: Defining variables residing in specified registers.
71 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
72 * Incomplete Enums:: @code{enum foo;}, with details to follow.
73 * Function Names:: Printable strings which are the name of the current
75 * Return Address:: Getting the return or frame address of a function.
76 * Vector Extensions:: Using vector instructions through built-in functions.
77 * Offsetof:: Special syntax for implementing @code{offsetof}.
78 * Atomic Builtins:: Built-in functions for atomic memory access.
79 * Object Size Checking:: Built-in functions for limited buffer overflow
81 * Other Builtins:: Other built-in functions.
82 * Target Builtins:: Built-in functions specific to particular targets.
83 * Target Format Checks:: Format checks specific to particular targets.
84 * Pragmas:: Pragmas accepted by GCC.
85 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
86 * Thread-Local:: Per-thread variables.
87 * Binary constants:: Binary constants using the @samp{0b} prefix.
91 @section Statements and Declarations in Expressions
92 @cindex statements inside expressions
93 @cindex declarations inside expressions
94 @cindex expressions containing statements
95 @cindex macros, statements in expressions
97 @c the above section title wrapped and causes an underfull hbox.. i
98 @c changed it from "within" to "in". --mew 4feb93
99 A compound statement enclosed in parentheses may appear as an expression
100 in GNU C@. This allows you to use loops, switches, and local variables
101 within an expression.
103 Recall that a compound statement is a sequence of statements surrounded
104 by braces; in this construct, parentheses go around the braces. For
108 (@{ int y = foo (); int z;
115 is a valid (though slightly more complex than necessary) expression
116 for the absolute value of @code{foo ()}.
118 The last thing in the compound statement should be an expression
119 followed by a semicolon; the value of this subexpression serves as the
120 value of the entire construct. (If you use some other kind of statement
121 last within the braces, the construct has type @code{void}, and thus
122 effectively no value.)
124 This feature is especially useful in making macro definitions ``safe'' (so
125 that they evaluate each operand exactly once). For example, the
126 ``maximum'' function is commonly defined as a macro in standard C as
130 #define max(a,b) ((a) > (b) ? (a) : (b))
134 @cindex side effects, macro argument
135 But this definition computes either @var{a} or @var{b} twice, with bad
136 results if the operand has side effects. In GNU C, if you know the
137 type of the operands (here taken as @code{int}), you can define
138 the macro safely as follows:
141 #define maxint(a,b) \
142 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
145 Embedded statements are not allowed in constant expressions, such as
146 the value of an enumeration constant, the width of a bit-field, or
147 the initial value of a static variable.
149 If you don't know the type of the operand, you can still do this, but you
150 must use @code{typeof} (@pxref{Typeof}).
152 In G++, the result value of a statement expression undergoes array and
153 function pointer decay, and is returned by value to the enclosing
154 expression. For instance, if @code{A} is a class, then
163 will construct a temporary @code{A} object to hold the result of the
164 statement expression, and that will be used to invoke @code{Foo}.
165 Therefore the @code{this} pointer observed by @code{Foo} will not be the
168 Any temporaries created within a statement within a statement expression
169 will be destroyed at the statement's end. This makes statement
170 expressions inside macros slightly different from function calls. In
171 the latter case temporaries introduced during argument evaluation will
172 be destroyed at the end of the statement that includes the function
173 call. In the statement expression case they will be destroyed during
174 the statement expression. For instance,
177 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
178 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
188 will have different places where temporaries are destroyed. For the
189 @code{macro} case, the temporary @code{X} will be destroyed just after
190 the initialization of @code{b}. In the @code{function} case that
191 temporary will be destroyed when the function returns.
193 These considerations mean that it is probably a bad idea to use
194 statement-expressions of this form in header files that are designed to
195 work with C++. (Note that some versions of the GNU C Library contained
196 header files using statement-expression that lead to precisely this
199 Jumping into a statement expression with @code{goto} or using a
200 @code{switch} statement outside the statement expression with a
201 @code{case} or @code{default} label inside the statement expression is
202 not permitted. Jumping into a statement expression with a computed
203 @code{goto} (@pxref{Labels as Values}) yields undefined behavior.
204 Jumping out of a statement expression is permitted, but if the
205 statement expression is part of a larger expression then it is
206 unspecified which other subexpressions of that expression have been
207 evaluated except where the language definition requires certain
208 subexpressions to be evaluated before or after the statement
209 expression. In any case, as with a function call the evaluation of a
210 statement expression is not interleaved with the evaluation of other
211 parts of the containing expression. For example,
214 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
218 will call @code{foo} and @code{bar1} and will not call @code{baz} but
219 may or may not call @code{bar2}. If @code{bar2} is called, it will be
220 called after @code{foo} and before @code{bar1}
223 @section Locally Declared Labels
225 @cindex macros, local labels
227 GCC allows you to declare @dfn{local labels} in any nested block
228 scope. A local label is just like an ordinary label, but you can
229 only reference it (with a @code{goto} statement, or by taking its
230 address) within the block in which it was declared.
232 A local label declaration looks like this:
235 __label__ @var{label};
242 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
245 Local label declarations must come at the beginning of the block,
246 before any ordinary declarations or statements.
248 The label declaration defines the label @emph{name}, but does not define
249 the label itself. You must do this in the usual way, with
250 @code{@var{label}:}, within the statements of the statement expression.
252 The local label feature is useful for complex macros. If a macro
253 contains nested loops, a @code{goto} can be useful for breaking out of
254 them. However, an ordinary label whose scope is the whole function
255 cannot be used: if the macro can be expanded several times in one
256 function, the label will be multiply defined in that function. A
257 local label avoids this problem. For example:
260 #define SEARCH(value, array, target) \
263 typeof (target) _SEARCH_target = (target); \
264 typeof (*(array)) *_SEARCH_array = (array); \
267 for (i = 0; i < max; i++) \
268 for (j = 0; j < max; j++) \
269 if (_SEARCH_array[i][j] == _SEARCH_target) \
270 @{ (value) = i; goto found; @} \
276 This could also be written using a statement-expression:
279 #define SEARCH(array, target) \
282 typeof (target) _SEARCH_target = (target); \
283 typeof (*(array)) *_SEARCH_array = (array); \
286 for (i = 0; i < max; i++) \
287 for (j = 0; j < max; j++) \
288 if (_SEARCH_array[i][j] == _SEARCH_target) \
289 @{ value = i; goto found; @} \
296 Local label declarations also make the labels they declare visible to
297 nested functions, if there are any. @xref{Nested Functions}, for details.
299 @node Labels as Values
300 @section Labels as Values
301 @cindex labels as values
302 @cindex computed gotos
303 @cindex goto with computed label
304 @cindex address of a label
306 You can get the address of a label defined in the current function
307 (or a containing function) with the unary operator @samp{&&}. The
308 value has type @code{void *}. This value is a constant and can be used
309 wherever a constant of that type is valid. For example:
317 To use these values, you need to be able to jump to one. This is done
318 with the computed goto statement@footnote{The analogous feature in
319 Fortran is called an assigned goto, but that name seems inappropriate in
320 C, where one can do more than simply store label addresses in label
321 variables.}, @code{goto *@var{exp};}. For example,
328 Any expression of type @code{void *} is allowed.
330 One way of using these constants is in initializing a static array that
331 will serve as a jump table:
334 static void *array[] = @{ &&foo, &&bar, &&hack @};
337 Then you can select a label with indexing, like this:
344 Note that this does not check whether the subscript is in bounds---array
345 indexing in C never does that.
347 Such an array of label values serves a purpose much like that of the
348 @code{switch} statement. The @code{switch} statement is cleaner, so
349 use that rather than an array unless the problem does not fit a
350 @code{switch} statement very well.
352 Another use of label values is in an interpreter for threaded code.
353 The labels within the interpreter function can be stored in the
354 threaded code for super-fast dispatching.
356 You may not use this mechanism to jump to code in a different function.
357 If you do that, totally unpredictable things will happen. The best way to
358 avoid this is to store the label address only in automatic variables and
359 never pass it as an argument.
361 An alternate way to write the above example is
364 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
366 goto *(&&foo + array[i]);
370 This is more friendly to code living in shared libraries, as it reduces
371 the number of dynamic relocations that are needed, and by consequence,
372 allows the data to be read-only.
374 @node Nested Functions
375 @section Nested Functions
376 @cindex nested functions
377 @cindex downward funargs
380 A @dfn{nested function} is a function defined inside another function.
381 (Nested functions are not supported for GNU C++.) The nested function's
382 name is local to the block where it is defined. For example, here we
383 define a nested function named @code{square}, and call it twice:
387 foo (double a, double b)
389 double square (double z) @{ return z * z; @}
391 return square (a) + square (b);
396 The nested function can access all the variables of the containing
397 function that are visible at the point of its definition. This is
398 called @dfn{lexical scoping}. For example, here we show a nested
399 function which uses an inherited variable named @code{offset}:
403 bar (int *array, int offset, int size)
405 int access (int *array, int index)
406 @{ return array[index + offset]; @}
409 for (i = 0; i < size; i++)
410 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
415 Nested function definitions are permitted within functions in the places
416 where variable definitions are allowed; that is, in any block, mixed
417 with the other declarations and statements in the block.
419 It is possible to call the nested function from outside the scope of its
420 name by storing its address or passing the address to another function:
423 hack (int *array, int size)
425 void store (int index, int value)
426 @{ array[index] = value; @}
428 intermediate (store, size);
432 Here, the function @code{intermediate} receives the address of
433 @code{store} as an argument. If @code{intermediate} calls @code{store},
434 the arguments given to @code{store} are used to store into @code{array}.
435 But this technique works only so long as the containing function
436 (@code{hack}, in this example) does not exit.
438 If you try to call the nested function through its address after the
439 containing function has exited, all hell will break loose. If you try
440 to call it after a containing scope level has exited, and if it refers
441 to some of the variables that are no longer in scope, you may be lucky,
442 but it's not wise to take the risk. If, however, the nested function
443 does not refer to anything that has gone out of scope, you should be
446 GCC implements taking the address of a nested function using a technique
447 called @dfn{trampolines}. A paper describing them is available as
450 @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}.
452 A nested function can jump to a label inherited from a containing
453 function, provided the label was explicitly declared in the containing
454 function (@pxref{Local Labels}). Such a jump returns instantly to the
455 containing function, exiting the nested function which did the
456 @code{goto} and any intermediate functions as well. Here is an example:
460 bar (int *array, int offset, int size)
463 int access (int *array, int index)
467 return array[index + offset];
471 for (i = 0; i < size; i++)
472 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
476 /* @r{Control comes here from @code{access}
477 if it detects an error.} */
484 A nested function always has no linkage. Declaring one with
485 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
486 before its definition, use @code{auto} (which is otherwise meaningless
487 for function declarations).
490 bar (int *array, int offset, int size)
493 auto int access (int *, int);
495 int access (int *array, int index)
499 return array[index + offset];
505 @node Constructing Calls
506 @section Constructing Function Calls
507 @cindex constructing calls
508 @cindex forwarding calls
510 Using the built-in functions described below, you can record
511 the arguments a function received, and call another function
512 with the same arguments, without knowing the number or types
515 You can also record the return value of that function call,
516 and later return that value, without knowing what data type
517 the function tried to return (as long as your caller expects
520 However, these built-in functions may interact badly with some
521 sophisticated features or other extensions of the language. It
522 is, therefore, not recommended to use them outside very simple
523 functions acting as mere forwarders for their arguments.
525 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
526 This built-in function returns a pointer to data
527 describing how to perform a call with the same arguments as were passed
528 to the current function.
530 The function saves the arg pointer register, structure value address,
531 and all registers that might be used to pass arguments to a function
532 into a block of memory allocated on the stack. Then it returns the
533 address of that block.
536 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
537 This built-in function invokes @var{function}
538 with a copy of the parameters described by @var{arguments}
541 The value of @var{arguments} should be the value returned by
542 @code{__builtin_apply_args}. The argument @var{size} specifies the size
543 of the stack argument data, in bytes.
545 This function returns a pointer to data describing
546 how to return whatever value was returned by @var{function}. The data
547 is saved in a block of memory allocated on the stack.
549 It is not always simple to compute the proper value for @var{size}. The
550 value is used by @code{__builtin_apply} to compute the amount of data
551 that should be pushed on the stack and copied from the incoming argument
555 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
556 This built-in function returns the value described by @var{result} from
557 the containing function. You should specify, for @var{result}, a value
558 returned by @code{__builtin_apply}.
562 @section Referring to a Type with @code{typeof}
565 @cindex macros, types of arguments
567 Another way to refer to the type of an expression is with @code{typeof}.
568 The syntax of using of this keyword looks like @code{sizeof}, but the
569 construct acts semantically like a type name defined with @code{typedef}.
571 There are two ways of writing the argument to @code{typeof}: with an
572 expression or with a type. Here is an example with an expression:
579 This assumes that @code{x} is an array of pointers to functions;
580 the type described is that of the values of the functions.
582 Here is an example with a typename as the argument:
589 Here the type described is that of pointers to @code{int}.
591 If you are writing a header file that must work when included in ISO C
592 programs, write @code{__typeof__} instead of @code{typeof}.
593 @xref{Alternate Keywords}.
595 A @code{typeof}-construct can be used anywhere a typedef name could be
596 used. For example, you can use it in a declaration, in a cast, or inside
597 of @code{sizeof} or @code{typeof}.
599 @code{typeof} is often useful in conjunction with the
600 statements-within-expressions feature. Here is how the two together can
601 be used to define a safe ``maximum'' macro that operates on any
602 arithmetic type and evaluates each of its arguments exactly once:
606 (@{ typeof (a) _a = (a); \
607 typeof (b) _b = (b); \
608 _a > _b ? _a : _b; @})
611 @cindex underscores in variables in macros
612 @cindex @samp{_} in variables in macros
613 @cindex local variables in macros
614 @cindex variables, local, in macros
615 @cindex macros, local variables in
617 The reason for using names that start with underscores for the local
618 variables is to avoid conflicts with variable names that occur within the
619 expressions that are substituted for @code{a} and @code{b}. Eventually we
620 hope to design a new form of declaration syntax that allows you to declare
621 variables whose scopes start only after their initializers; this will be a
622 more reliable way to prevent such conflicts.
625 Some more examples of the use of @code{typeof}:
629 This declares @code{y} with the type of what @code{x} points to.
636 This declares @code{y} as an array of such values.
643 This declares @code{y} as an array of pointers to characters:
646 typeof (typeof (char *)[4]) y;
650 It is equivalent to the following traditional C declaration:
656 To see the meaning of the declaration using @code{typeof}, and why it
657 might be a useful way to write, rewrite it with these macros:
660 #define pointer(T) typeof(T *)
661 #define array(T, N) typeof(T [N])
665 Now the declaration can be rewritten this way:
668 array (pointer (char), 4) y;
672 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
673 pointers to @code{char}.
676 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
677 a more limited extension which permitted one to write
680 typedef @var{T} = @var{expr};
684 with the effect of declaring @var{T} to have the type of the expression
685 @var{expr}. This extension does not work with GCC 3 (versions between
686 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
687 relies on it should be rewritten to use @code{typeof}:
690 typedef typeof(@var{expr}) @var{T};
694 This will work with all versions of GCC@.
697 @section Conditionals with Omitted Operands
698 @cindex conditional expressions, extensions
699 @cindex omitted middle-operands
700 @cindex middle-operands, omitted
701 @cindex extensions, @code{?:}
702 @cindex @code{?:} extensions
704 The middle operand in a conditional expression may be omitted. Then
705 if the first operand is nonzero, its value is the value of the conditional
708 Therefore, the expression
715 has the value of @code{x} if that is nonzero; otherwise, the value of
718 This example is perfectly equivalent to
724 @cindex side effect in ?:
725 @cindex ?: side effect
727 In this simple case, the ability to omit the middle operand is not
728 especially useful. When it becomes useful is when the first operand does,
729 or may (if it is a macro argument), contain a side effect. Then repeating
730 the operand in the middle would perform the side effect twice. Omitting
731 the middle operand uses the value already computed without the undesirable
732 effects of recomputing it.
735 @section Double-Word Integers
736 @cindex @code{long long} data types
737 @cindex double-word arithmetic
738 @cindex multiprecision arithmetic
739 @cindex @code{LL} integer suffix
740 @cindex @code{ULL} integer suffix
742 ISO C99 supports data types for integers that are at least 64 bits wide,
743 and as an extension GCC supports them in C89 mode and in C++.
744 Simply write @code{long long int} for a signed integer, or
745 @code{unsigned long long int} for an unsigned integer. To make an
746 integer constant of type @code{long long int}, add the suffix @samp{LL}
747 to the integer. To make an integer constant of type @code{unsigned long
748 long int}, add the suffix @samp{ULL} to the integer.
750 You can use these types in arithmetic like any other integer types.
751 Addition, subtraction, and bitwise boolean operations on these types
752 are open-coded on all types of machines. Multiplication is open-coded
753 if the machine supports fullword-to-doubleword a widening multiply
754 instruction. Division and shifts are open-coded only on machines that
755 provide special support. The operations that are not open-coded use
756 special library routines that come with GCC@.
758 There may be pitfalls when you use @code{long long} types for function
759 arguments, unless you declare function prototypes. If a function
760 expects type @code{int} for its argument, and you pass a value of type
761 @code{long long int}, confusion will result because the caller and the
762 subroutine will disagree about the number of bytes for the argument.
763 Likewise, if the function expects @code{long long int} and you pass
764 @code{int}. The best way to avoid such problems is to use prototypes.
767 @section Complex Numbers
768 @cindex complex numbers
769 @cindex @code{_Complex} keyword
770 @cindex @code{__complex__} keyword
772 ISO C99 supports complex floating data types, and as an extension GCC
773 supports them in C89 mode and in C++, and supports complex integer data
774 types which are not part of ISO C99. You can declare complex types
775 using the keyword @code{_Complex}. As an extension, the older GNU
776 keyword @code{__complex__} is also supported.
778 For example, @samp{_Complex double x;} declares @code{x} as a
779 variable whose real part and imaginary part are both of type
780 @code{double}. @samp{_Complex short int y;} declares @code{y} to
781 have real and imaginary parts of type @code{short int}; this is not
782 likely to be useful, but it shows that the set of complex types is
785 To write a constant with a complex data type, use the suffix @samp{i} or
786 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
787 has type @code{_Complex float} and @code{3i} has type
788 @code{_Complex int}. Such a constant always has a pure imaginary
789 value, but you can form any complex value you like by adding one to a
790 real constant. This is a GNU extension; if you have an ISO C99
791 conforming C library (such as GNU libc), and want to construct complex
792 constants of floating type, you should include @code{<complex.h>} and
793 use the macros @code{I} or @code{_Complex_I} instead.
795 @cindex @code{__real__} keyword
796 @cindex @code{__imag__} keyword
797 To extract the real part of a complex-valued expression @var{exp}, write
798 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
799 extract the imaginary part. This is a GNU extension; for values of
800 floating type, you should use the ISO C99 functions @code{crealf},
801 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
802 @code{cimagl}, declared in @code{<complex.h>} and also provided as
803 built-in functions by GCC@.
805 @cindex complex conjugation
806 The operator @samp{~} performs complex conjugation when used on a value
807 with a complex type. This is a GNU extension; for values of
808 floating type, you should use the ISO C99 functions @code{conjf},
809 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
810 provided as built-in functions by GCC@.
812 GCC can allocate complex automatic variables in a noncontiguous
813 fashion; it's even possible for the real part to be in a register while
814 the imaginary part is on the stack (or vice-versa). Only the DWARF2
815 debug info format can represent this, so use of DWARF2 is recommended.
816 If you are using the stabs debug info format, GCC describes a noncontiguous
817 complex variable as if it were two separate variables of noncomplex type.
818 If the variable's actual name is @code{foo}, the two fictitious
819 variables are named @code{foo$real} and @code{foo$imag}. You can
820 examine and set these two fictitious variables with your debugger.
823 @section Decimal Floating Types
824 @cindex decimal floating types
825 @cindex @code{_Decimal32} data type
826 @cindex @code{_Decimal64} data type
827 @cindex @code{_Decimal128} data type
828 @cindex @code{df} integer suffix
829 @cindex @code{dd} integer suffix
830 @cindex @code{dl} integer suffix
831 @cindex @code{DF} integer suffix
832 @cindex @code{DD} integer suffix
833 @cindex @code{DL} integer suffix
835 As an extension, the GNU C compiler supports decimal floating types as
836 defined in the N1176 draft of ISO/IEC WDTR24732. Support for decimal
837 floating types in GCC will evolve as the draft technical report changes.
838 Calling conventions for any target might also change. Not all targets
839 support decimal floating types.
841 The decimal floating types are @code{_Decimal32}, @code{_Decimal64}, and
842 @code{_Decimal128}. They use a radix of ten, unlike the floating types
843 @code{float}, @code{double}, and @code{long double} whose radix is not
844 specified by the C standard but is usually two.
846 Support for decimal floating types includes the arithmetic operators
847 add, subtract, multiply, divide; unary arithmetic operators;
848 relational operators; equality operators; and conversions to and from
849 integer and other floating types. Use a suffix @samp{df} or
850 @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd}
851 or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for
854 GCC support of decimal float as specified by the draft technical report
859 Translation time data type (TTDT) is not supported.
862 Characteristics of decimal floating types are defined in header file
863 @file{decfloat.h} rather than @file{float.h}.
866 When the value of a decimal floating type cannot be represented in the
867 integer type to which it is being converted, the result is undefined
868 rather than the result value specified by the draft technical report.
871 Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128}
872 are supported by the DWARF2 debug information format.
878 ISO C99 supports floating-point numbers written not only in the usual
879 decimal notation, such as @code{1.55e1}, but also numbers such as
880 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
881 supports this in C89 mode (except in some cases when strictly
882 conforming) and in C++. In that format the
883 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
884 mandatory. The exponent is a decimal number that indicates the power of
885 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
892 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
893 is the same as @code{1.55e1}.
895 Unlike for floating-point numbers in the decimal notation the exponent
896 is always required in the hexadecimal notation. Otherwise the compiler
897 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
898 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
899 extension for floating-point constants of type @code{float}.
902 @section Arrays of Length Zero
903 @cindex arrays of length zero
904 @cindex zero-length arrays
905 @cindex length-zero arrays
906 @cindex flexible array members
908 Zero-length arrays are allowed in GNU C@. They are very useful as the
909 last element of a structure which is really a header for a variable-length
918 struct line *thisline = (struct line *)
919 malloc (sizeof (struct line) + this_length);
920 thisline->length = this_length;
923 In ISO C90, you would have to give @code{contents} a length of 1, which
924 means either you waste space or complicate the argument to @code{malloc}.
926 In ISO C99, you would use a @dfn{flexible array member}, which is
927 slightly different in syntax and semantics:
931 Flexible array members are written as @code{contents[]} without
935 Flexible array members have incomplete type, and so the @code{sizeof}
936 operator may not be applied. As a quirk of the original implementation
937 of zero-length arrays, @code{sizeof} evaluates to zero.
940 Flexible array members may only appear as the last member of a
941 @code{struct} that is otherwise non-empty.
944 A structure containing a flexible array member, or a union containing
945 such a structure (possibly recursively), may not be a member of a
946 structure or an element of an array. (However, these uses are
947 permitted by GCC as extensions.)
950 GCC versions before 3.0 allowed zero-length arrays to be statically
951 initialized, as if they were flexible arrays. In addition to those
952 cases that were useful, it also allowed initializations in situations
953 that would corrupt later data. Non-empty initialization of zero-length
954 arrays is now treated like any case where there are more initializer
955 elements than the array holds, in that a suitable warning about "excess
956 elements in array" is given, and the excess elements (all of them, in
957 this case) are ignored.
959 Instead GCC allows static initialization of flexible array members.
960 This is equivalent to defining a new structure containing the original
961 structure followed by an array of sufficient size to contain the data.
962 I.e.@: in the following, @code{f1} is constructed as if it were declared
968 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
971 struct f1 f1; int data[3];
972 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
976 The convenience of this extension is that @code{f1} has the desired
977 type, eliminating the need to consistently refer to @code{f2.f1}.
979 This has symmetry with normal static arrays, in that an array of
980 unknown size is also written with @code{[]}.
982 Of course, this extension only makes sense if the extra data comes at
983 the end of a top-level object, as otherwise we would be overwriting
984 data at subsequent offsets. To avoid undue complication and confusion
985 with initialization of deeply nested arrays, we simply disallow any
986 non-empty initialization except when the structure is the top-level
990 struct foo @{ int x; int y[]; @};
991 struct bar @{ struct foo z; @};
993 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
994 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
995 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
996 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
999 @node Empty Structures
1000 @section Structures With No Members
1001 @cindex empty structures
1002 @cindex zero-size structures
1004 GCC permits a C structure to have no members:
1011 The structure will have size zero. In C++, empty structures are part
1012 of the language. G++ treats empty structures as if they had a single
1013 member of type @code{char}.
1015 @node Variable Length
1016 @section Arrays of Variable Length
1017 @cindex variable-length arrays
1018 @cindex arrays of variable length
1021 Variable-length automatic arrays are allowed in ISO C99, and as an
1022 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1023 implementation of variable-length arrays does not yet conform in detail
1024 to the ISO C99 standard.) These arrays are
1025 declared like any other automatic arrays, but with a length that is not
1026 a constant expression. The storage is allocated at the point of
1027 declaration and deallocated when the brace-level is exited. For
1032 concat_fopen (char *s1, char *s2, char *mode)
1034 char str[strlen (s1) + strlen (s2) + 1];
1037 return fopen (str, mode);
1041 @cindex scope of a variable length array
1042 @cindex variable-length array scope
1043 @cindex deallocating variable length arrays
1044 Jumping or breaking out of the scope of the array name deallocates the
1045 storage. Jumping into the scope is not allowed; you get an error
1048 @cindex @code{alloca} vs variable-length arrays
1049 You can use the function @code{alloca} to get an effect much like
1050 variable-length arrays. The function @code{alloca} is available in
1051 many other C implementations (but not in all). On the other hand,
1052 variable-length arrays are more elegant.
1054 There are other differences between these two methods. Space allocated
1055 with @code{alloca} exists until the containing @emph{function} returns.
1056 The space for a variable-length array is deallocated as soon as the array
1057 name's scope ends. (If you use both variable-length arrays and
1058 @code{alloca} in the same function, deallocation of a variable-length array
1059 will also deallocate anything more recently allocated with @code{alloca}.)
1061 You can also use variable-length arrays as arguments to functions:
1065 tester (int len, char data[len][len])
1071 The length of an array is computed once when the storage is allocated
1072 and is remembered for the scope of the array in case you access it with
1075 If you want to pass the array first and the length afterward, you can
1076 use a forward declaration in the parameter list---another GNU extension.
1080 tester (int len; char data[len][len], int len)
1086 @cindex parameter forward declaration
1087 The @samp{int len} before the semicolon is a @dfn{parameter forward
1088 declaration}, and it serves the purpose of making the name @code{len}
1089 known when the declaration of @code{data} is parsed.
1091 You can write any number of such parameter forward declarations in the
1092 parameter list. They can be separated by commas or semicolons, but the
1093 last one must end with a semicolon, which is followed by the ``real''
1094 parameter declarations. Each forward declaration must match a ``real''
1095 declaration in parameter name and data type. ISO C99 does not support
1096 parameter forward declarations.
1098 @node Variadic Macros
1099 @section Macros with a Variable Number of Arguments.
1100 @cindex variable number of arguments
1101 @cindex macro with variable arguments
1102 @cindex rest argument (in macro)
1103 @cindex variadic macros
1105 In the ISO C standard of 1999, a macro can be declared to accept a
1106 variable number of arguments much as a function can. The syntax for
1107 defining the macro is similar to that of a function. Here is an
1111 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1114 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1115 such a macro, it represents the zero or more tokens until the closing
1116 parenthesis that ends the invocation, including any commas. This set of
1117 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1118 wherever it appears. See the CPP manual for more information.
1120 GCC has long supported variadic macros, and used a different syntax that
1121 allowed you to give a name to the variable arguments just like any other
1122 argument. Here is an example:
1125 #define debug(format, args...) fprintf (stderr, format, args)
1128 This is in all ways equivalent to the ISO C example above, but arguably
1129 more readable and descriptive.
1131 GNU CPP has two further variadic macro extensions, and permits them to
1132 be used with either of the above forms of macro definition.
1134 In standard C, you are not allowed to leave the variable argument out
1135 entirely; but you are allowed to pass an empty argument. For example,
1136 this invocation is invalid in ISO C, because there is no comma after
1143 GNU CPP permits you to completely omit the variable arguments in this
1144 way. In the above examples, the compiler would complain, though since
1145 the expansion of the macro still has the extra comma after the format
1148 To help solve this problem, CPP behaves specially for variable arguments
1149 used with the token paste operator, @samp{##}. If instead you write
1152 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1155 and if the variable arguments are omitted or empty, the @samp{##}
1156 operator causes the preprocessor to remove the comma before it. If you
1157 do provide some variable arguments in your macro invocation, GNU CPP
1158 does not complain about the paste operation and instead places the
1159 variable arguments after the comma. Just like any other pasted macro
1160 argument, these arguments are not macro expanded.
1162 @node Escaped Newlines
1163 @section Slightly Looser Rules for Escaped Newlines
1164 @cindex escaped newlines
1165 @cindex newlines (escaped)
1167 Recently, the preprocessor has relaxed its treatment of escaped
1168 newlines. Previously, the newline had to immediately follow a
1169 backslash. The current implementation allows whitespace in the form
1170 of spaces, horizontal and vertical tabs, and form feeds between the
1171 backslash and the subsequent newline. The preprocessor issues a
1172 warning, but treats it as a valid escaped newline and combines the two
1173 lines to form a single logical line. This works within comments and
1174 tokens, as well as between tokens. Comments are @emph{not} treated as
1175 whitespace for the purposes of this relaxation, since they have not
1176 yet been replaced with spaces.
1179 @section Non-Lvalue Arrays May Have Subscripts
1180 @cindex subscripting
1181 @cindex arrays, non-lvalue
1183 @cindex subscripting and function values
1184 In ISO C99, arrays that are not lvalues still decay to pointers, and
1185 may be subscripted, although they may not be modified or used after
1186 the next sequence point and the unary @samp{&} operator may not be
1187 applied to them. As an extension, GCC allows such arrays to be
1188 subscripted in C89 mode, though otherwise they do not decay to
1189 pointers outside C99 mode. For example,
1190 this is valid in GNU C though not valid in C89:
1194 struct foo @{int a[4];@};
1200 return f().a[index];
1206 @section Arithmetic on @code{void}- and Function-Pointers
1207 @cindex void pointers, arithmetic
1208 @cindex void, size of pointer to
1209 @cindex function pointers, arithmetic
1210 @cindex function, size of pointer to
1212 In GNU C, addition and subtraction operations are supported on pointers to
1213 @code{void} and on pointers to functions. This is done by treating the
1214 size of a @code{void} or of a function as 1.
1216 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1217 and on function types, and returns 1.
1219 @opindex Wpointer-arith
1220 The option @option{-Wpointer-arith} requests a warning if these extensions
1224 @section Non-Constant Initializers
1225 @cindex initializers, non-constant
1226 @cindex non-constant initializers
1228 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1229 automatic variable are not required to be constant expressions in GNU C@.
1230 Here is an example of an initializer with run-time varying elements:
1233 foo (float f, float g)
1235 float beat_freqs[2] = @{ f-g, f+g @};
1240 @node Compound Literals
1241 @section Compound Literals
1242 @cindex constructor expressions
1243 @cindex initializations in expressions
1244 @cindex structures, constructor expression
1245 @cindex expressions, constructor
1246 @cindex compound literals
1247 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1249 ISO C99 supports compound literals. A compound literal looks like
1250 a cast containing an initializer. Its value is an object of the
1251 type specified in the cast, containing the elements specified in
1252 the initializer; it is an lvalue. As an extension, GCC supports
1253 compound literals in C89 mode and in C++.
1255 Usually, the specified type is a structure. Assume that
1256 @code{struct foo} and @code{structure} are declared as shown:
1259 struct foo @{int a; char b[2];@} structure;
1263 Here is an example of constructing a @code{struct foo} with a compound literal:
1266 structure = ((struct foo) @{x + y, 'a', 0@});
1270 This is equivalent to writing the following:
1274 struct foo temp = @{x + y, 'a', 0@};
1279 You can also construct an array. If all the elements of the compound literal
1280 are (made up of) simple constant expressions, suitable for use in
1281 initializers of objects of static storage duration, then the compound
1282 literal can be coerced to a pointer to its first element and used in
1283 such an initializer, as shown here:
1286 char **foo = (char *[]) @{ "x", "y", "z" @};
1289 Compound literals for scalar types and union types are is
1290 also allowed, but then the compound literal is equivalent
1293 As a GNU extension, GCC allows initialization of objects with static storage
1294 duration by compound literals (which is not possible in ISO C99, because
1295 the initializer is not a constant).
1296 It is handled as if the object was initialized only with the bracket
1297 enclosed list if the types of the compound literal and the object match.
1298 The initializer list of the compound literal must be constant.
1299 If the object being initialized has array type of unknown size, the size is
1300 determined by compound literal size.
1303 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1304 static int y[] = (int []) @{1, 2, 3@};
1305 static int z[] = (int [3]) @{1@};
1309 The above lines are equivalent to the following:
1311 static struct foo x = @{1, 'a', 'b'@};
1312 static int y[] = @{1, 2, 3@};
1313 static int z[] = @{1, 0, 0@};
1316 @node Designated Inits
1317 @section Designated Initializers
1318 @cindex initializers with labeled elements
1319 @cindex labeled elements in initializers
1320 @cindex case labels in initializers
1321 @cindex designated initializers
1323 Standard C89 requires the elements of an initializer to appear in a fixed
1324 order, the same as the order of the elements in the array or structure
1327 In ISO C99 you can give the elements in any order, specifying the array
1328 indices or structure field names they apply to, and GNU C allows this as
1329 an extension in C89 mode as well. This extension is not
1330 implemented in GNU C++.
1332 To specify an array index, write
1333 @samp{[@var{index}] =} before the element value. For example,
1336 int a[6] = @{ [4] = 29, [2] = 15 @};
1343 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1347 The index values must be constant expressions, even if the array being
1348 initialized is automatic.
1350 An alternative syntax for this which has been obsolete since GCC 2.5 but
1351 GCC still accepts is to write @samp{[@var{index}]} before the element
1352 value, with no @samp{=}.
1354 To initialize a range of elements to the same value, write
1355 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1356 extension. For example,
1359 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1363 If the value in it has side-effects, the side-effects will happen only once,
1364 not for each initialized field by the range initializer.
1367 Note that the length of the array is the highest value specified
1370 In a structure initializer, specify the name of a field to initialize
1371 with @samp{.@var{fieldname} =} before the element value. For example,
1372 given the following structure,
1375 struct point @{ int x, y; @};
1379 the following initialization
1382 struct point p = @{ .y = yvalue, .x = xvalue @};
1389 struct point p = @{ xvalue, yvalue @};
1392 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1393 @samp{@var{fieldname}:}, as shown here:
1396 struct point p = @{ y: yvalue, x: xvalue @};
1400 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1401 @dfn{designator}. You can also use a designator (or the obsolete colon
1402 syntax) when initializing a union, to specify which element of the union
1403 should be used. For example,
1406 union foo @{ int i; double d; @};
1408 union foo f = @{ .d = 4 @};
1412 will convert 4 to a @code{double} to store it in the union using
1413 the second element. By contrast, casting 4 to type @code{union foo}
1414 would store it into the union as the integer @code{i}, since it is
1415 an integer. (@xref{Cast to Union}.)
1417 You can combine this technique of naming elements with ordinary C
1418 initialization of successive elements. Each initializer element that
1419 does not have a designator applies to the next consecutive element of the
1420 array or structure. For example,
1423 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1430 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1433 Labeling the elements of an array initializer is especially useful
1434 when the indices are characters or belong to an @code{enum} type.
1439 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1440 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1443 @cindex designator lists
1444 You can also write a series of @samp{.@var{fieldname}} and
1445 @samp{[@var{index}]} designators before an @samp{=} to specify a
1446 nested subobject to initialize; the list is taken relative to the
1447 subobject corresponding to the closest surrounding brace pair. For
1448 example, with the @samp{struct point} declaration above:
1451 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1455 If the same field is initialized multiple times, it will have value from
1456 the last initialization. If any such overridden initialization has
1457 side-effect, it is unspecified whether the side-effect happens or not.
1458 Currently, GCC will discard them and issue a warning.
1461 @section Case Ranges
1463 @cindex ranges in case statements
1465 You can specify a range of consecutive values in a single @code{case} label,
1469 case @var{low} ... @var{high}:
1473 This has the same effect as the proper number of individual @code{case}
1474 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1476 This feature is especially useful for ranges of ASCII character codes:
1482 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1483 it may be parsed wrong when you use it with integer values. For example,
1498 @section Cast to a Union Type
1499 @cindex cast to a union
1500 @cindex union, casting to a
1502 A cast to union type is similar to other casts, except that the type
1503 specified is a union type. You can specify the type either with
1504 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1505 a constructor though, not a cast, and hence does not yield an lvalue like
1506 normal casts. (@xref{Compound Literals}.)
1508 The types that may be cast to the union type are those of the members
1509 of the union. Thus, given the following union and variables:
1512 union foo @{ int i; double d; @};
1518 both @code{x} and @code{y} can be cast to type @code{union foo}.
1520 Using the cast as the right-hand side of an assignment to a variable of
1521 union type is equivalent to storing in a member of the union:
1526 u = (union foo) x @equiv{} u.i = x
1527 u = (union foo) y @equiv{} u.d = y
1530 You can also use the union cast as a function argument:
1533 void hack (union foo);
1535 hack ((union foo) x);
1538 @node Mixed Declarations
1539 @section Mixed Declarations and Code
1540 @cindex mixed declarations and code
1541 @cindex declarations, mixed with code
1542 @cindex code, mixed with declarations
1544 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1545 within compound statements. As an extension, GCC also allows this in
1546 C89 mode. For example, you could do:
1555 Each identifier is visible from where it is declared until the end of
1556 the enclosing block.
1558 @node Function Attributes
1559 @section Declaring Attributes of Functions
1560 @cindex function attributes
1561 @cindex declaring attributes of functions
1562 @cindex functions that never return
1563 @cindex functions that return more than once
1564 @cindex functions that have no side effects
1565 @cindex functions in arbitrary sections
1566 @cindex functions that behave like malloc
1567 @cindex @code{volatile} applied to function
1568 @cindex @code{const} applied to function
1569 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1570 @cindex functions with non-null pointer arguments
1571 @cindex functions that are passed arguments in registers on the 386
1572 @cindex functions that pop the argument stack on the 386
1573 @cindex functions that do not pop the argument stack on the 386
1575 In GNU C, you declare certain things about functions called in your program
1576 which help the compiler optimize function calls and check your code more
1579 The keyword @code{__attribute__} allows you to specify special
1580 attributes when making a declaration. This keyword is followed by an
1581 attribute specification inside double parentheses. The following
1582 attributes are currently defined for functions on all targets:
1584 @code{noreturn}, @code{returns_twice}, @code{noinline}, @code{always_inline},
1585 @code{flatten}, @code{pure}, @code{const}, @code{nothrow}, @code{sentinel},
1586 @code{format}, @code{format_arg}, @code{no_instrument_function},
1587 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1588 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
1589 @code{alias}, @code{warn_unused_result}, @code{nonnull},
1590 @code{gnu_inline} and @code{externally_visible}. Several other
1591 attributes are defined for functions on particular target systems. Other
1592 attributes, including @code{section} are supported for variables declarations
1593 @c APPLE LOCAL begin for-fsf-4_4 3274130 5295549
1594 (@pxref{Variable Attributes}), for types (@pxref{Type Attributes}),
1595 and labels (@pxref{Label Attributes}).
1597 @c APPLE LOCAL end for-fsf-4_4 3274130 5295549
1598 You may also specify attributes with @samp{__} preceding and following
1599 each keyword. This allows you to use them in header files without
1600 being concerned about a possible macro of the same name. For example,
1601 you may use @code{__noreturn__} instead of @code{noreturn}.
1603 @xref{Attribute Syntax}, for details of the exact syntax for using
1607 @c Keep this table alphabetized by attribute name. Treat _ as space.
1609 @item alias ("@var{target}")
1610 @cindex @code{alias} attribute
1611 The @code{alias} attribute causes the declaration to be emitted as an
1612 alias for another symbol, which must be specified. For instance,
1615 void __f () @{ /* @r{Do something.} */; @}
1616 void f () __attribute__ ((weak, alias ("__f")));
1619 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
1620 mangled name for the target must be used. It is an error if @samp{__f}
1621 is not defined in the same translation unit.
1623 Not all target machines support this attribute.
1625 @item aligned (@var{alignment})
1626 @cindex @code{aligned} attribute
1627 This attribute specifies a minimum alignment for the function,
1630 You cannot use this attribute to decrease the alignment of a function,
1631 only to increase it. However, when you explicitly specify a function
1632 alignment this will override the effect of the
1633 @option{-falign-functions} (@pxref{Optimize Options}) option for this
1636 Note that the effectiveness of @code{aligned} attributes may be
1637 limited by inherent limitations in your linker. On many systems, the
1638 linker is only able to arrange for functions to be aligned up to a
1639 certain maximum alignment. (For some linkers, the maximum supported
1640 alignment may be very very small.) See your linker documentation for
1641 further information.
1643 The @code{aligned} attribute can also be used for variables and fields
1644 (@pxref{Variable Attributes}.)
1647 @cindex @code{always_inline} function attribute
1648 Generally, functions are not inlined unless optimization is specified.
1649 For functions declared inline, this attribute inlines the function even
1650 if no optimization level was specified.
1653 @cindex @code{gnu_inline} function attribute
1654 This attribute should be used with a function which is also declared
1655 with the @code{inline} keyword. It directs GCC to treat the function
1656 as if it were defined in gnu89 mode even when compiling in C99 or
1659 If the function is declared @code{extern}, then this definition of the
1660 function is used only for inlining. In no case is the function
1661 compiled as a standalone function, not even if you take its address
1662 explicitly. Such an address becomes an external reference, as if you
1663 had only declared the function, and had not defined it. This has
1664 almost the effect of a macro. The way to use this is to put a
1665 function definition in a header file with this attribute, and put
1666 another copy of the function, without @code{extern}, in a library
1667 file. The definition in the header file will cause most calls to the
1668 function to be inlined. If any uses of the function remain, they will
1669 refer to the single copy in the library. Note that the two
1670 definitions of the functions need not be precisely the same, although
1671 if they do not have the same effect your program may behave oddly.
1673 If the function is neither @code{extern} nor @code{static}, then the
1674 function is compiled as a standalone function, as well as being
1675 inlined where possible.
1677 This is how GCC traditionally handled functions declared
1678 @code{inline}. Since ISO C99 specifies a different semantics for
1679 @code{inline}, this function attribute is provided as a transition
1680 measure and as a useful feature in its own right. This attribute is
1681 available in GCC 4.1.3 and later. It is available if either of the
1682 preprocessor macros @code{__GNUC_GNU_INLINE__} or
1683 @code{__GNUC_STDC_INLINE__} are defined. @xref{Inline,,An Inline
1684 Function is As Fast As a Macro}.
1686 Note that since the first version of GCC to support C99 inline semantics
1687 is 4.3, earlier versions of GCC which accept this attribute effectively
1688 assume that it is always present, whether or not it is given explicitly.
1689 In versions prior to 4.3, the only effect of explicitly including it is
1690 to disable warnings about using inline functions in C99 mode.
1692 @cindex @code{flatten} function attribute
1694 Generally, inlining into a function is limited. For a function marked with
1695 this attribute, every call inside this function will be inlined, if possible.
1696 Whether the function itself is considered for inlining depends on its size and
1697 the current inlining parameters. The @code{flatten} attribute only works
1698 reliably in unit-at-a-time mode.
1701 @cindex functions that do pop the argument stack on the 386
1703 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1704 assume that the calling function will pop off the stack space used to
1705 pass arguments. This is
1706 useful to override the effects of the @option{-mrtd} switch.
1709 @cindex @code{const} function attribute
1710 Many functions do not examine any values except their arguments, and
1711 have no effects except the return value. Basically this is just slightly
1712 more strict class than the @code{pure} attribute below, since function is not
1713 allowed to read global memory.
1715 @cindex pointer arguments
1716 Note that a function that has pointer arguments and examines the data
1717 pointed to must @emph{not} be declared @code{const}. Likewise, a
1718 function that calls a non-@code{const} function usually must not be
1719 @code{const}. It does not make sense for a @code{const} function to
1722 The attribute @code{const} is not implemented in GCC versions earlier
1723 than 2.5. An alternative way to declare that a function has no side
1724 effects, which works in the current version and in some older versions,
1728 typedef int intfn ();
1730 extern const intfn square;
1733 This approach does not work in GNU C++ from 2.6.0 on, since the language
1734 specifies that the @samp{const} must be attached to the return value.
1738 @cindex @code{constructor} function attribute
1739 @cindex @code{destructor} function attribute
1740 The @code{constructor} attribute causes the function to be called
1741 automatically before execution enters @code{main ()}. Similarly, the
1742 @code{destructor} attribute causes the function to be called
1743 automatically after @code{main ()} has completed or @code{exit ()} has
1744 been called. Functions with these attributes are useful for
1745 initializing data that will be used implicitly during the execution of
1749 @cindex @code{deprecated} attribute.
1750 The @code{deprecated} attribute results in a warning if the function
1751 is used anywhere in the source file. This is useful when identifying
1752 functions that are expected to be removed in a future version of a
1753 program. The warning also includes the location of the declaration
1754 of the deprecated function, to enable users to easily find further
1755 information about why the function is deprecated, or what they should
1756 do instead. Note that the warnings only occurs for uses:
1759 int old_fn () __attribute__ ((deprecated));
1761 int (*fn_ptr)() = old_fn;
1764 results in a warning on line 3 but not line 2.
1766 The @code{deprecated} attribute can also be used for variables and
1767 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
1770 @cindex @code{__declspec(dllexport)}
1771 On Microsoft Windows targets and Symbian OS targets the
1772 @code{dllexport} attribute causes the compiler to provide a global
1773 pointer to a pointer in a DLL, so that it can be referenced with the
1774 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
1775 name is formed by combining @code{_imp__} and the function or variable
1778 You can use @code{__declspec(dllexport)} as a synonym for
1779 @code{__attribute__ ((dllexport))} for compatibility with other
1782 On systems that support the @code{visibility} attribute, this
1783 attribute also implies ``default'' visibility, unless a
1784 @code{visibility} attribute is explicitly specified. You should avoid
1785 the use of @code{dllexport} with ``hidden'' or ``internal''
1786 visibility; in the future GCC may issue an error for those cases.
1788 Currently, the @code{dllexport} attribute is ignored for inlined
1789 functions, unless the @option{-fkeep-inline-functions} flag has been
1790 used. The attribute is also ignored for undefined symbols.
1792 When applied to C++ classes, the attribute marks defined non-inlined
1793 member functions and static data members as exports. Static consts
1794 initialized in-class are not marked unless they are also defined
1797 For Microsoft Windows targets there are alternative methods for
1798 including the symbol in the DLL's export table such as using a
1799 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
1800 the @option{--export-all} linker flag.
1803 @cindex @code{__declspec(dllimport)}
1804 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
1805 attribute causes the compiler to reference a function or variable via
1806 a global pointer to a pointer that is set up by the DLL exporting the
1807 symbol. The attribute implies @code{extern} storage. On Microsoft
1808 Windows targets, the pointer name is formed by combining @code{_imp__}
1809 and the function or variable name.
1811 You can use @code{__declspec(dllimport)} as a synonym for
1812 @code{__attribute__ ((dllimport))} for compatibility with other
1815 Currently, the attribute is ignored for inlined functions. If the
1816 attribute is applied to a symbol @emph{definition}, an error is reported.
1817 If a symbol previously declared @code{dllimport} is later defined, the
1818 attribute is ignored in subsequent references, and a warning is emitted.
1819 The attribute is also overridden by a subsequent declaration as
1822 When applied to C++ classes, the attribute marks non-inlined
1823 member functions and static data members as imports. However, the
1824 attribute is ignored for virtual methods to allow creation of vtables
1827 On the SH Symbian OS target the @code{dllimport} attribute also has
1828 another affect---it can cause the vtable and run-time type information
1829 for a class to be exported. This happens when the class has a
1830 dllimport'ed constructor or a non-inline, non-pure virtual function
1831 and, for either of those two conditions, the class also has a inline
1832 constructor or destructor and has a key function that is defined in
1833 the current translation unit.
1835 For Microsoft Windows based targets the use of the @code{dllimport}
1836 attribute on functions is not necessary, but provides a small
1837 performance benefit by eliminating a thunk in the DLL@. The use of the
1838 @code{dllimport} attribute on imported variables was required on older
1839 versions of the GNU linker, but can now be avoided by passing the
1840 @option{--enable-auto-import} switch to the GNU linker. As with
1841 functions, using the attribute for a variable eliminates a thunk in
1844 One drawback to using this attribute is that a pointer to a function
1845 or variable marked as @code{dllimport} cannot be used as a constant
1846 address. On Microsoft Windows targets, the attribute can be disabled
1847 for functions by setting the @option{-mnop-fun-dllimport} flag.
1850 @cindex eight bit data on the H8/300, H8/300H, and H8S
1851 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1852 variable should be placed into the eight bit data section.
1853 The compiler will generate more efficient code for certain operations
1854 on data in the eight bit data area. Note the eight bit data area is limited to
1857 You must use GAS and GLD from GNU binutils version 2.7 or later for
1858 this attribute to work correctly.
1860 @item exception_handler
1861 @cindex exception handler functions on the Blackfin processor
1862 Use this attribute on the Blackfin to indicate that the specified function
1863 is an exception handler. The compiler will generate function entry and
1864 exit sequences suitable for use in an exception handler when this
1865 attribute is present.
1868 @cindex functions which handle memory bank switching
1869 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
1870 use a calling convention that takes care of switching memory banks when
1871 entering and leaving a function. This calling convention is also the
1872 default when using the @option{-mlong-calls} option.
1874 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
1875 to call and return from a function.
1877 On 68HC11 the compiler will generate a sequence of instructions
1878 to invoke a board-specific routine to switch the memory bank and call the
1879 real function. The board-specific routine simulates a @code{call}.
1880 At the end of a function, it will jump to a board-specific routine
1881 instead of using @code{rts}. The board-specific return routine simulates
1885 @cindex functions that pop the argument stack on the 386
1886 On the Intel 386, the @code{fastcall} attribute causes the compiler to
1887 pass the first argument (if of integral type) in the register ECX and
1888 the second argument (if of integral type) in the register EDX@. Subsequent
1889 and other typed arguments are passed on the stack. The called function will
1890 pop the arguments off the stack. If the number of arguments is variable all
1891 arguments are pushed on the stack.
1893 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
1894 @cindex @code{format} function attribute
1896 The @code{format} attribute specifies that a function takes @code{printf},
1897 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
1898 should be type-checked against a format string. For example, the
1903 my_printf (void *my_object, const char *my_format, ...)
1904 __attribute__ ((format (printf, 2, 3)));
1908 causes the compiler to check the arguments in calls to @code{my_printf}
1909 for consistency with the @code{printf} style format string argument
1912 The parameter @var{archetype} determines how the format string is
1913 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
1914 or @code{strfmon}. (You can also use @code{__printf__},
1915 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
1916 parameter @var{string-index} specifies which argument is the format
1917 string argument (starting from 1), while @var{first-to-check} is the
1918 number of the first argument to check against the format string. For
1919 functions where the arguments are not available to be checked (such as
1920 @code{vprintf}), specify the third parameter as zero. In this case the
1921 compiler only checks the format string for consistency. For
1922 @code{strftime} formats, the third parameter is required to be zero.
1923 Since non-static C++ methods have an implicit @code{this} argument, the
1924 arguments of such methods should be counted from two, not one, when
1925 giving values for @var{string-index} and @var{first-to-check}.
1927 In the example above, the format string (@code{my_format}) is the second
1928 argument of the function @code{my_print}, and the arguments to check
1929 start with the third argument, so the correct parameters for the format
1930 attribute are 2 and 3.
1932 @opindex ffreestanding
1933 @opindex fno-builtin
1934 The @code{format} attribute allows you to identify your own functions
1935 which take format strings as arguments, so that GCC can check the
1936 calls to these functions for errors. The compiler always (unless
1937 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
1938 for the standard library functions @code{printf}, @code{fprintf},
1939 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
1940 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
1941 warnings are requested (using @option{-Wformat}), so there is no need to
1942 modify the header file @file{stdio.h}. In C99 mode, the functions
1943 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
1944 @code{vsscanf} are also checked. Except in strictly conforming C
1945 standard modes, the X/Open function @code{strfmon} is also checked as
1946 are @code{printf_unlocked} and @code{fprintf_unlocked}.
1947 @xref{C Dialect Options,,Options Controlling C Dialect}.
1949 The target may provide additional types of format checks.
1950 @xref{Target Format Checks,,Format Checks Specific to Particular
1953 @item format_arg (@var{string-index})
1954 @cindex @code{format_arg} function attribute
1955 @opindex Wformat-nonliteral
1956 The @code{format_arg} attribute specifies that a function takes a format
1957 string for a @code{printf}, @code{scanf}, @code{strftime} or
1958 @code{strfmon} style function and modifies it (for example, to translate
1959 it into another language), so the result can be passed to a
1960 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
1961 function (with the remaining arguments to the format function the same
1962 as they would have been for the unmodified string). For example, the
1967 my_dgettext (char *my_domain, const char *my_format)
1968 __attribute__ ((format_arg (2)));
1972 causes the compiler to check the arguments in calls to a @code{printf},
1973 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
1974 format string argument is a call to the @code{my_dgettext} function, for
1975 consistency with the format string argument @code{my_format}. If the
1976 @code{format_arg} attribute had not been specified, all the compiler
1977 could tell in such calls to format functions would be that the format
1978 string argument is not constant; this would generate a warning when
1979 @option{-Wformat-nonliteral} is used, but the calls could not be checked
1980 without the attribute.
1982 The parameter @var{string-index} specifies which argument is the format
1983 string argument (starting from one). Since non-static C++ methods have
1984 an implicit @code{this} argument, the arguments of such methods should
1985 be counted from two.
1987 The @code{format-arg} attribute allows you to identify your own
1988 functions which modify format strings, so that GCC can check the
1989 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
1990 type function whose operands are a call to one of your own function.
1991 The compiler always treats @code{gettext}, @code{dgettext}, and
1992 @code{dcgettext} in this manner except when strict ISO C support is
1993 requested by @option{-ansi} or an appropriate @option{-std} option, or
1994 @option{-ffreestanding} or @option{-fno-builtin}
1995 is used. @xref{C Dialect Options,,Options
1996 Controlling C Dialect}.
1998 @item function_vector
1999 @cindex calling functions through the function vector on the H8/300 processors
2000 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2001 function should be called through the function vector. Calling a
2002 function through the function vector will reduce code size, however;
2003 the function vector has a limited size (maximum 128 entries on the H8/300
2004 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2006 You must use GAS and GLD from GNU binutils version 2.7 or later for
2007 this attribute to work correctly.
2010 @cindex interrupt handler functions
2011 Use this attribute on the ARM, AVR, C4x, CRX, M32C, M32R/D, MS1, and Xstormy16
2012 ports to indicate that the specified function is an interrupt handler.
2013 The compiler will generate function entry and exit sequences suitable
2014 for use in an interrupt handler when this attribute is present.
2016 Note, interrupt handlers for the Blackfin, m68k, H8/300, H8/300H, H8S, and
2017 SH processors can be specified via the @code{interrupt_handler} attribute.
2019 Note, on the AVR, interrupts will be enabled inside the function.
2021 Note, for the ARM, you can specify the kind of interrupt to be handled by
2022 adding an optional parameter to the interrupt attribute like this:
2025 void f () __attribute__ ((interrupt ("IRQ")));
2028 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2030 @item interrupt_handler
2031 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
2032 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
2033 indicate that the specified function is an interrupt handler. The compiler
2034 will generate function entry and exit sequences suitable for use in an
2035 interrupt handler when this attribute is present.
2038 @cindex User stack pointer in interrupts on the Blackfin
2039 When used together with @code{interrupt_handler}, @code{exception_handler}
2040 or @code{nmi_handler}, code will be generated to load the stack pointer
2041 from the USP register in the function prologue.
2043 @item long_call/short_call
2044 @cindex indirect calls on ARM
2045 This attribute specifies how a particular function is called on
2046 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2047 command line switch and @code{#pragma long_calls} settings. The
2048 @code{long_call} attribute indicates that the function might be far
2049 away from the call site and require a different (more expensive)
2050 calling sequence. The @code{short_call} attribute always places
2051 the offset to the function from the call site into the @samp{BL}
2052 instruction directly.
2054 @item longcall/shortcall
2055 @cindex functions called via pointer on the RS/6000 and PowerPC
2056 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
2057 indicates that the function might be far away from the call site and
2058 require a different (more expensive) calling sequence. The
2059 @code{shortcall} attribute indicates that the function is always close
2060 enough for the shorter calling sequence to be used. These attributes
2061 override both the @option{-mlongcall} switch and, on the RS/6000 and
2062 PowerPC, the @code{#pragma longcall} setting.
2064 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2065 calls are necessary.
2068 @cindex indirect calls on MIPS
2069 This attribute specifies how a particular function is called on MIPS@.
2070 The attribute overrides the @option{-mlong-calls} (@pxref{MIPS Options})
2071 command line switch. This attribute causes the compiler to always call
2072 the function by first loading its address into a register, and then using
2073 the contents of that register.
2076 @cindex @code{malloc} attribute
2077 The @code{malloc} attribute is used to tell the compiler that a function
2078 may be treated as if any non-@code{NULL} pointer it returns cannot
2079 alias any other pointer valid when the function returns.
2080 This will often improve optimization.
2081 Standard functions with this property include @code{malloc} and
2082 @code{calloc}. @code{realloc}-like functions have this property as
2083 long as the old pointer is never referred to (including comparing it
2084 to the new pointer) after the function returns a non-@code{NULL}
2087 @item model (@var{model-name})
2088 @cindex function addressability on the M32R/D
2089 @cindex variable addressability on the IA-64
2091 On the M32R/D, use this attribute to set the addressability of an
2092 object, and of the code generated for a function. The identifier
2093 @var{model-name} is one of @code{small}, @code{medium}, or
2094 @code{large}, representing each of the code models.
2096 Small model objects live in the lower 16MB of memory (so that their
2097 addresses can be loaded with the @code{ld24} instruction), and are
2098 callable with the @code{bl} instruction.
2100 Medium model objects may live anywhere in the 32-bit address space (the
2101 compiler will generate @code{seth/add3} instructions to load their addresses),
2102 and are callable with the @code{bl} instruction.
2104 Large model objects may live anywhere in the 32-bit address space (the
2105 compiler will generate @code{seth/add3} instructions to load their addresses),
2106 and may not be reachable with the @code{bl} instruction (the compiler will
2107 generate the much slower @code{seth/add3/jl} instruction sequence).
2109 On IA-64, use this attribute to set the addressability of an object.
2110 At present, the only supported identifier for @var{model-name} is
2111 @code{small}, indicating addressability via ``small'' (22-bit)
2112 addresses (so that their addresses can be loaded with the @code{addl}
2113 instruction). Caveat: such addressing is by definition not position
2114 independent and hence this attribute must not be used for objects
2115 defined by shared libraries.
2118 @cindex function without a prologue/epilogue code
2119 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
2120 specified function does not need prologue/epilogue sequences generated by
2121 the compiler. It is up to the programmer to provide these sequences.
2124 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2125 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2126 use the normal calling convention based on @code{jsr} and @code{rts}.
2127 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2131 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2132 Use this attribute together with @code{interrupt_handler},
2133 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2134 entry code should enable nested interrupts or exceptions.
2137 @cindex NMI handler functions on the Blackfin processor
2138 Use this attribute on the Blackfin to indicate that the specified function
2139 is an NMI handler. The compiler will generate function entry and
2140 exit sequences suitable for use in an NMI handler when this
2141 attribute is present.
2143 @item no_instrument_function
2144 @cindex @code{no_instrument_function} function attribute
2145 @opindex finstrument-functions
2146 If @option{-finstrument-functions} is given, profiling function calls will
2147 be generated at entry and exit of most user-compiled functions.
2148 Functions with this attribute will not be so instrumented.
2151 @cindex @code{noinline} function attribute
2152 This function attribute prevents a function from being considered for
2155 @item nonnull (@var{arg-index}, @dots{})
2156 @cindex @code{nonnull} function attribute
2157 The @code{nonnull} attribute specifies that some function parameters should
2158 be non-null pointers. For instance, the declaration:
2162 my_memcpy (void *dest, const void *src, size_t len)
2163 __attribute__((nonnull (1, 2)));
2167 causes the compiler to check that, in calls to @code{my_memcpy},
2168 arguments @var{dest} and @var{src} are non-null. If the compiler
2169 determines that a null pointer is passed in an argument slot marked
2170 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2171 is issued. The compiler may also choose to make optimizations based
2172 on the knowledge that certain function arguments will not be null.
2174 If no argument index list is given to the @code{nonnull} attribute,
2175 all pointer arguments are marked as non-null. To illustrate, the
2176 following declaration is equivalent to the previous example:
2180 my_memcpy (void *dest, const void *src, size_t len)
2181 __attribute__((nonnull));
2185 @cindex @code{noreturn} function attribute
2186 A few standard library functions, such as @code{abort} and @code{exit},
2187 cannot return. GCC knows this automatically. Some programs define
2188 their own functions that never return. You can declare them
2189 @code{noreturn} to tell the compiler this fact. For example,
2193 void fatal () __attribute__ ((noreturn));
2196 fatal (/* @r{@dots{}} */)
2198 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2204 The @code{noreturn} keyword tells the compiler to assume that
2205 @code{fatal} cannot return. It can then optimize without regard to what
2206 would happen if @code{fatal} ever did return. This makes slightly
2207 better code. More importantly, it helps avoid spurious warnings of
2208 uninitialized variables.
2210 The @code{noreturn} keyword does not affect the exceptional path when that
2211 applies: a @code{noreturn}-marked function may still return to the caller
2212 by throwing an exception or calling @code{longjmp}.
2214 Do not assume that registers saved by the calling function are
2215 restored before calling the @code{noreturn} function.
2217 It does not make sense for a @code{noreturn} function to have a return
2218 type other than @code{void}.
2220 The attribute @code{noreturn} is not implemented in GCC versions
2221 earlier than 2.5. An alternative way to declare that a function does
2222 not return, which works in the current version and in some older
2223 versions, is as follows:
2226 typedef void voidfn ();
2228 volatile voidfn fatal;
2231 This approach does not work in GNU C++.
2234 @cindex @code{nothrow} function attribute
2235 The @code{nothrow} attribute is used to inform the compiler that a
2236 function cannot throw an exception. For example, most functions in
2237 the standard C library can be guaranteed not to throw an exception
2238 with the notable exceptions of @code{qsort} and @code{bsearch} that
2239 take function pointer arguments. The @code{nothrow} attribute is not
2240 implemented in GCC versions earlier than 3.3.
2243 @cindex @code{pure} function attribute
2244 Many functions have no effects except the return value and their
2245 return value depends only on the parameters and/or global variables.
2246 Such a function can be subject
2247 to common subexpression elimination and loop optimization just as an
2248 arithmetic operator would be. These functions should be declared
2249 with the attribute @code{pure}. For example,
2252 int square (int) __attribute__ ((pure));
2256 says that the hypothetical function @code{square} is safe to call
2257 fewer times than the program says.
2259 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2260 Interesting non-pure functions are functions with infinite loops or those
2261 depending on volatile memory or other system resource, that may change between
2262 two consecutive calls (such as @code{feof} in a multithreading environment).
2264 The attribute @code{pure} is not implemented in GCC versions earlier
2267 @item regparm (@var{number})
2268 @cindex @code{regparm} attribute
2269 @cindex functions that are passed arguments in registers on the 386
2270 On the Intel 386, the @code{regparm} attribute causes the compiler to
2271 pass arguments number one to @var{number} if they are of integral type
2272 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2273 take a variable number of arguments will continue to be passed all of their
2274 arguments on the stack.
2276 Beware that on some ELF systems this attribute is unsuitable for
2277 global functions in shared libraries with lazy binding (which is the
2278 default). Lazy binding will send the first call via resolving code in
2279 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2280 per the standard calling conventions. Solaris 8 is affected by this.
2281 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2282 safe since the loaders there save all registers. (Lazy binding can be
2283 disabled with the linker or the loader if desired, to avoid the
2287 @cindex @code{sseregparm} attribute
2288 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2289 causes the compiler to pass up to 3 floating point arguments in
2290 SSE registers instead of on the stack. Functions that take a
2291 variable number of arguments will continue to pass all of their
2292 floating point arguments on the stack.
2294 @item force_align_arg_pointer
2295 @cindex @code{force_align_arg_pointer} attribute
2296 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
2297 applied to individual function definitions, generating an alternate
2298 prologue and epilogue that realigns the runtime stack. This supports
2299 mixing legacy codes that run with a 4-byte aligned stack with modern
2300 codes that keep a 16-byte stack for SSE compatibility. The alternate
2301 prologue and epilogue are slower and bigger than the regular ones, and
2302 the alternate prologue requires a scratch register; this lowers the
2303 number of registers available if used in conjunction with the
2304 @code{regparm} attribute. The @code{force_align_arg_pointer}
2305 attribute is incompatible with nested functions; this is considered a
2309 @cindex @code{returns_twice} attribute
2310 The @code{returns_twice} attribute tells the compiler that a function may
2311 return more than one time. The compiler will ensure that all registers
2312 are dead before calling such a function and will emit a warning about
2313 the variables that may be clobbered after the second return from the
2314 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2315 The @code{longjmp}-like counterpart of such function, if any, might need
2316 to be marked with the @code{noreturn} attribute.
2319 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2320 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2321 all registers except the stack pointer should be saved in the prologue
2322 regardless of whether they are used or not.
2324 @item section ("@var{section-name}")
2325 @cindex @code{section} function attribute
2326 Normally, the compiler places the code it generates in the @code{text} section.
2327 Sometimes, however, you need additional sections, or you need certain
2328 particular functions to appear in special sections. The @code{section}
2329 attribute specifies that a function lives in a particular section.
2330 For example, the declaration:
2333 extern void foobar (void) __attribute__ ((section ("bar")));
2337 puts the function @code{foobar} in the @code{bar} section.
2339 Some file formats do not support arbitrary sections so the @code{section}
2340 attribute is not available on all platforms.
2341 If you need to map the entire contents of a module to a particular
2342 section, consider using the facilities of the linker instead.
2345 @cindex @code{sentinel} function attribute
2346 This function attribute ensures that a parameter in a function call is
2347 an explicit @code{NULL}. The attribute is only valid on variadic
2348 functions. By default, the sentinel is located at position zero, the
2349 last parameter of the function call. If an optional integer position
2350 argument P is supplied to the attribute, the sentinel must be located at
2351 position P counting backwards from the end of the argument list.
2354 __attribute__ ((sentinel))
2356 __attribute__ ((sentinel(0)))
2359 The attribute is automatically set with a position of 0 for the built-in
2360 functions @code{execl} and @code{execlp}. The built-in function
2361 @code{execle} has the attribute set with a position of 1.
2363 A valid @code{NULL} in this context is defined as zero with any pointer
2364 type. If your system defines the @code{NULL} macro with an integer type
2365 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2366 with a copy that redefines NULL appropriately.
2368 The warnings for missing or incorrect sentinels are enabled with
2372 See long_call/short_call.
2375 See longcall/shortcall.
2378 @cindex signal handler functions on the AVR processors
2379 Use this attribute on the AVR to indicate that the specified
2380 function is a signal handler. The compiler will generate function
2381 entry and exit sequences suitable for use in a signal handler when this
2382 attribute is present. Interrupts will be disabled inside the function.
2385 Use this attribute on the SH to indicate an @code{interrupt_handler}
2386 function should switch to an alternate stack. It expects a string
2387 argument that names a global variable holding the address of the
2392 void f () __attribute__ ((interrupt_handler,
2393 sp_switch ("alt_stack")));
2397 @cindex functions that pop the argument stack on the 386
2398 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2399 assume that the called function will pop off the stack space used to
2400 pass arguments, unless it takes a variable number of arguments.
2403 @cindex tiny data section on the H8/300H and H8S
2404 Use this attribute on the H8/300H and H8S to indicate that the specified
2405 variable should be placed into the tiny data section.
2406 The compiler will generate more efficient code for loads and stores
2407 on data in the tiny data section. Note the tiny data area is limited to
2408 slightly under 32kbytes of data.
2411 Use this attribute on the SH for an @code{interrupt_handler} to return using
2412 @code{trapa} instead of @code{rte}. This attribute expects an integer
2413 argument specifying the trap number to be used.
2416 @cindex @code{unused} attribute.
2417 This attribute, attached to a function, means that the function is meant
2418 to be possibly unused. GCC will not produce a warning for this
2422 @cindex @code{used} attribute.
2423 This attribute, attached to a function, means that code must be emitted
2424 for the function even if it appears that the function is not referenced.
2425 This is useful, for example, when the function is referenced only in
2428 @item visibility ("@var{visibility_type}")
2429 @cindex @code{visibility} attribute
2430 This attribute affects the linkage of the declaration to which it is attached.
2431 There are four supported @var{visibility_type} values: default,
2432 hidden, protected or internal visibility.
2435 void __attribute__ ((visibility ("protected")))
2436 f () @{ /* @r{Do something.} */; @}
2437 int i __attribute__ ((visibility ("hidden")));
2440 The possible values of @var{visibility_type} correspond to the
2441 visibility settings in the ELF gABI.
2444 @c keep this list of visibilities in alphabetical order.
2447 Default visibility is the normal case for the object file format.
2448 This value is available for the visibility attribute to override other
2449 options that may change the assumed visibility of entities.
2451 On ELF, default visibility means that the declaration is visible to other
2452 modules and, in shared libraries, means that the declared entity may be
2455 On Darwin, default visibility means that the declaration is visible to
2458 Default visibility corresponds to ``external linkage'' in the language.
2461 Hidden visibility indicates that the entity declared will have a new
2462 form of linkage, which we'll call ``hidden linkage''. Two
2463 declarations of an object with hidden linkage refer to the same object
2464 if they are in the same shared object.
2467 Internal visibility is like hidden visibility, but with additional
2468 processor specific semantics. Unless otherwise specified by the
2469 psABI, GCC defines internal visibility to mean that a function is
2470 @emph{never} called from another module. Compare this with hidden
2471 functions which, while they cannot be referenced directly by other
2472 modules, can be referenced indirectly via function pointers. By
2473 indicating that a function cannot be called from outside the module,
2474 GCC may for instance omit the load of a PIC register since it is known
2475 that the calling function loaded the correct value.
2478 Protected visibility is like default visibility except that it
2479 indicates that references within the defining module will bind to the
2480 definition in that module. That is, the declared entity cannot be
2481 overridden by another module.
2485 All visibilities are supported on many, but not all, ELF targets
2486 (supported when the assembler supports the @samp{.visibility}
2487 pseudo-op). Default visibility is supported everywhere. Hidden
2488 visibility is supported on Darwin targets.
2490 The visibility attribute should be applied only to declarations which
2491 would otherwise have external linkage. The attribute should be applied
2492 consistently, so that the same entity should not be declared with
2493 different settings of the attribute.
2495 In C++, the visibility attribute applies to types as well as functions
2496 and objects, because in C++ types have linkage. A class must not have
2497 greater visibility than its non-static data member types and bases,
2498 and class members default to the visibility of their class. Also, a
2499 declaration without explicit visibility is limited to the visibility
2502 In C++, you can mark member functions and static member variables of a
2503 class with the visibility attribute. This is useful if if you know a
2504 particular method or static member variable should only be used from
2505 one shared object; then you can mark it hidden while the rest of the
2506 class has default visibility. Care must be taken to avoid breaking
2507 the One Definition Rule; for example, it is usually not useful to mark
2508 an inline method as hidden without marking the whole class as hidden.
2510 A C++ namespace declaration can also have the visibility attribute.
2511 This attribute applies only to the particular namespace body, not to
2512 other definitions of the same namespace; it is equivalent to using
2513 @samp{#pragma GCC visibility} before and after the namespace
2514 definition (@pxref{Visibility Pragmas}).
2516 In C++, if a template argument has limited visibility, this
2517 restriction is implicitly propagated to the template instantiation.
2518 Otherwise, template instantiations and specializations default to the
2519 visibility of their template.
2521 If both the template and enclosing class have explicit visibility, the
2522 visibility from the template is used.
2524 @item warn_unused_result
2525 @cindex @code{warn_unused_result} attribute
2526 The @code{warn_unused_result} attribute causes a warning to be emitted
2527 if a caller of the function with this attribute does not use its
2528 return value. This is useful for functions where not checking
2529 the result is either a security problem or always a bug, such as
2533 int fn () __attribute__ ((warn_unused_result));
2536 if (fn () < 0) return -1;
2542 results in warning on line 5.
2545 @cindex @code{weak} attribute
2546 The @code{weak} attribute causes the declaration to be emitted as a weak
2547 symbol rather than a global. This is primarily useful in defining
2548 library functions which can be overridden in user code, though it can
2549 also be used with non-function declarations. Weak symbols are supported
2550 for ELF targets, and also for a.out targets when using the GNU assembler
2554 @itemx weakref ("@var{target}")
2555 @cindex @code{weakref} attribute
2556 The @code{weakref} attribute marks a declaration as a weak reference.
2557 Without arguments, it should be accompanied by an @code{alias} attribute
2558 naming the target symbol. Optionally, the @var{target} may be given as
2559 an argument to @code{weakref} itself. In either case, @code{weakref}
2560 implicitly marks the declaration as @code{weak}. Without a
2561 @var{target}, given as an argument to @code{weakref} or to @code{alias},
2562 @code{weakref} is equivalent to @code{weak}.
2565 static int x() __attribute__ ((weakref ("y")));
2566 /* is equivalent to... */
2567 static int x() __attribute__ ((weak, weakref, alias ("y")));
2569 static int x() __attribute__ ((weakref));
2570 static int x() __attribute__ ((alias ("y")));
2573 A weak reference is an alias that does not by itself require a
2574 definition to be given for the target symbol. If the target symbol is
2575 only referenced through weak references, then the becomes a @code{weak}
2576 undefined symbol. If it is directly referenced, however, then such
2577 strong references prevail, and a definition will be required for the
2578 symbol, not necessarily in the same translation unit.
2580 The effect is equivalent to moving all references to the alias to a
2581 separate translation unit, renaming the alias to the aliased symbol,
2582 declaring it as weak, compiling the two separate translation units and
2583 performing a reloadable link on them.
2585 At present, a declaration to which @code{weakref} is attached can
2586 only be @code{static}.
2588 @item externally_visible
2589 @cindex @code{externally_visible} attribute.
2590 This attribute, attached to a global variable or function nullify
2591 effect of @option{-fwhole-program} command line option, so the object
2592 remain visible outside the current compilation unit
2596 You can specify multiple attributes in a declaration by separating them
2597 by commas within the double parentheses or by immediately following an
2598 attribute declaration with another attribute declaration.
2600 @cindex @code{#pragma}, reason for not using
2601 @cindex pragma, reason for not using
2602 Some people object to the @code{__attribute__} feature, suggesting that
2603 ISO C's @code{#pragma} should be used instead. At the time
2604 @code{__attribute__} was designed, there were two reasons for not doing
2609 It is impossible to generate @code{#pragma} commands from a macro.
2612 There is no telling what the same @code{#pragma} might mean in another
2616 These two reasons applied to almost any application that might have been
2617 proposed for @code{#pragma}. It was basically a mistake to use
2618 @code{#pragma} for @emph{anything}.
2620 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2621 to be generated from macros. In addition, a @code{#pragma GCC}
2622 namespace is now in use for GCC-specific pragmas. However, it has been
2623 found convenient to use @code{__attribute__} to achieve a natural
2624 attachment of attributes to their corresponding declarations, whereas
2625 @code{#pragma GCC} is of use for constructs that do not naturally form
2626 part of the grammar. @xref{Other Directives,,Miscellaneous
2627 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2629 @node Attribute Syntax
2630 @section Attribute Syntax
2631 @cindex attribute syntax
2633 This section describes the syntax with which @code{__attribute__} may be
2634 used, and the constructs to which attribute specifiers bind, for the C
2635 language. Some details may vary for C++. Because of infelicities in
2636 the grammar for attributes, some forms described here may not be
2637 successfully parsed in all cases.
2639 There are some problems with the semantics of attributes in C++. For
2640 example, there are no manglings for attributes, although they may affect
2641 code generation, so problems may arise when attributed types are used in
2642 conjunction with templates or overloading. Similarly, @code{typeid}
2643 does not distinguish between types with different attributes. Support
2644 for attributes in C++ may be restricted in future to attributes on
2645 declarations only, but not on nested declarators.
2647 @xref{Function Attributes}, for details of the semantics of attributes
2648 applying to functions. @xref{Variable Attributes}, for details of the
2649 @c APPLE LOCAL begin for-fsf-4_4 3274130 5295549
2650 semantics of attributes applying to variables. @xref{Type
2651 Attributes}, for details of the semantics of attributes applying to
2652 structure, union and enumerated types. @xref{Label Attributes}, for
2653 details of the semantics of attributes applying to labels and
2656 @c APPLE LOCAL end for-fsf-4_4 3274130 5295549
2657 An @dfn{attribute specifier} is of the form
2658 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2659 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2660 each attribute is one of the following:
2664 Empty. Empty attributes are ignored.
2667 A word (which may be an identifier such as @code{unused}, or a reserved
2668 word such as @code{const}).
2671 A word, followed by, in parentheses, parameters for the attribute.
2672 These parameters take one of the following forms:
2676 An identifier. For example, @code{mode} attributes use this form.
2679 An identifier followed by a comma and a non-empty comma-separated list
2680 of expressions. For example, @code{format} attributes use this form.
2683 A possibly empty comma-separated list of expressions. For example,
2684 @code{format_arg} attributes use this form with the list being a single
2685 integer constant expression, and @code{alias} attributes use this form
2686 with the list being a single string constant.
2690 An @dfn{attribute specifier list} is a sequence of one or more attribute
2691 specifiers, not separated by any other tokens.
2693 @c APPLE LOCAL begin for-fsf-4_4 3274130 5295549
2694 In GNU C, an attribute specifier list may appear after the colon
2695 following a label, other than a @code{case} or @code{default} label.
2696 GNU C++ does not permit such placement of attribute lists, as it is
2697 permissible for a declaration, which could begin with an attribute
2698 list, to be labelled in C++. Declarations cannot be labelled in C90
2699 or C99, so the ambiguity does not arise there.
2701 In GNU C an attribute specifier list may also appear after the keyword
2702 @code{while} in a while loop, after @code{do} and after @code{for}.
2704 @c APPLE LOCAL end for-fsf-4_4 3274130 5295549
2705 An attribute specifier list may appear as part of a @code{struct},
2706 @code{union} or @code{enum} specifier. It may go either immediately
2707 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2708 the closing brace. The former syntax is preferred.
2709 Where attribute specifiers follow the closing brace, they are considered
2710 to relate to the structure, union or enumerated type defined, not to any
2711 enclosing declaration the type specifier appears in, and the type
2712 defined is not complete until after the attribute specifiers.
2713 @c Otherwise, there would be the following problems: a shift/reduce
2714 @c conflict between attributes binding the struct/union/enum and
2715 @c binding to the list of specifiers/qualifiers; and "aligned"
2716 @c attributes could use sizeof for the structure, but the size could be
2717 @c changed later by "packed" attributes.
2719 Otherwise, an attribute specifier appears as part of a declaration,
2720 counting declarations of unnamed parameters and type names, and relates
2721 to that declaration (which may be nested in another declaration, for
2722 example in the case of a parameter declaration), or to a particular declarator
2723 within a declaration. Where an
2724 attribute specifier is applied to a parameter declared as a function or
2725 an array, it should apply to the function or array rather than the
2726 pointer to which the parameter is implicitly converted, but this is not
2727 yet correctly implemented.
2729 Any list of specifiers and qualifiers at the start of a declaration may
2730 contain attribute specifiers, whether or not such a list may in that
2731 context contain storage class specifiers. (Some attributes, however,
2732 are essentially in the nature of storage class specifiers, and only make
2733 sense where storage class specifiers may be used; for example,
2734 @code{section}.) There is one necessary limitation to this syntax: the
2735 first old-style parameter declaration in a function definition cannot
2736 begin with an attribute specifier, because such an attribute applies to
2737 the function instead by syntax described below (which, however, is not
2738 yet implemented in this case). In some other cases, attribute
2739 specifiers are permitted by this grammar but not yet supported by the
2740 compiler. All attribute specifiers in this place relate to the
2741 declaration as a whole. In the obsolescent usage where a type of
2742 @code{int} is implied by the absence of type specifiers, such a list of
2743 specifiers and qualifiers may be an attribute specifier list with no
2744 other specifiers or qualifiers.
2746 At present, the first parameter in a function prototype must have some
2747 type specifier which is not an attribute specifier; this resolves an
2748 ambiguity in the interpretation of @code{void f(int
2749 (__attribute__((foo)) x))}, but is subject to change. At present, if
2750 the parentheses of a function declarator contain only attributes then
2751 those attributes are ignored, rather than yielding an error or warning
2752 or implying a single parameter of type int, but this is subject to
2755 An attribute specifier list may appear immediately before a declarator
2756 (other than the first) in a comma-separated list of declarators in a
2757 declaration of more than one identifier using a single list of
2758 specifiers and qualifiers. Such attribute specifiers apply
2759 only to the identifier before whose declarator they appear. For
2763 __attribute__((noreturn)) void d0 (void),
2764 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2769 the @code{noreturn} attribute applies to all the functions
2770 declared; the @code{format} attribute only applies to @code{d1}.
2772 An attribute specifier list may appear immediately before the comma,
2773 @code{=} or semicolon terminating the declaration of an identifier other
2774 than a function definition. At present, such attribute specifiers apply
2775 to the declared object or function, but in future they may attach to the
2776 outermost adjacent declarator. In simple cases there is no difference,
2777 but, for example, in
2780 void (****f)(void) __attribute__((noreturn));
2784 at present the @code{noreturn} attribute applies to @code{f}, which
2785 causes a warning since @code{f} is not a function, but in future it may
2786 apply to the function @code{****f}. The precise semantics of what
2787 attributes in such cases will apply to are not yet specified. Where an
2788 assembler name for an object or function is specified (@pxref{Asm
2789 Labels}), at present the attribute must follow the @code{asm}
2790 specification; in future, attributes before the @code{asm} specification
2791 may apply to the adjacent declarator, and those after it to the declared
2794 An attribute specifier list may, in future, be permitted to appear after
2795 the declarator in a function definition (before any old-style parameter
2796 declarations or the function body).
2798 Attribute specifiers may be mixed with type qualifiers appearing inside
2799 the @code{[]} of a parameter array declarator, in the C99 construct by
2800 which such qualifiers are applied to the pointer to which the array is
2801 implicitly converted. Such attribute specifiers apply to the pointer,
2802 not to the array, but at present this is not implemented and they are
2805 An attribute specifier list may appear at the start of a nested
2806 declarator. At present, there are some limitations in this usage: the
2807 attributes correctly apply to the declarator, but for most individual
2808 attributes the semantics this implies are not implemented.
2809 When attribute specifiers follow the @code{*} of a pointer
2810 declarator, they may be mixed with any type qualifiers present.
2811 The following describes the formal semantics of this syntax. It will make the
2812 most sense if you are familiar with the formal specification of
2813 declarators in the ISO C standard.
2815 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2816 D1}, where @code{T} contains declaration specifiers that specify a type
2817 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2818 contains an identifier @var{ident}. The type specified for @var{ident}
2819 for derived declarators whose type does not include an attribute
2820 specifier is as in the ISO C standard.
2822 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2823 and the declaration @code{T D} specifies the type
2824 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2825 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2826 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2828 If @code{D1} has the form @code{*
2829 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2830 declaration @code{T D} specifies the type
2831 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2832 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2833 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2839 void (__attribute__((noreturn)) ****f) (void);
2843 specifies the type ``pointer to pointer to pointer to pointer to
2844 non-returning function returning @code{void}''. As another example,
2847 char *__attribute__((aligned(8))) *f;
2851 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2852 Note again that this does not work with most attributes; for example,
2853 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2854 is not yet supported.
2856 For compatibility with existing code written for compiler versions that
2857 did not implement attributes on nested declarators, some laxity is
2858 allowed in the placing of attributes. If an attribute that only applies
2859 to types is applied to a declaration, it will be treated as applying to
2860 the type of that declaration. If an attribute that only applies to
2861 declarations is applied to the type of a declaration, it will be treated
2862 as applying to that declaration; and, for compatibility with code
2863 placing the attributes immediately before the identifier declared, such
2864 an attribute applied to a function return type will be treated as
2865 applying to the function type, and such an attribute applied to an array
2866 element type will be treated as applying to the array type. If an
2867 attribute that only applies to function types is applied to a
2868 pointer-to-function type, it will be treated as applying to the pointer
2869 target type; if such an attribute is applied to a function return type
2870 that is not a pointer-to-function type, it will be treated as applying
2871 to the function type.
2873 @node Function Prototypes
2874 @section Prototypes and Old-Style Function Definitions
2875 @cindex function prototype declarations
2876 @cindex old-style function definitions
2877 @cindex promotion of formal parameters
2879 GNU C extends ISO C to allow a function prototype to override a later
2880 old-style non-prototype definition. Consider the following example:
2883 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2890 /* @r{Prototype function declaration.} */
2891 int isroot P((uid_t));
2893 /* @r{Old-style function definition.} */
2895 isroot (x) /* @r{??? lossage here ???} */
2902 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2903 not allow this example, because subword arguments in old-style
2904 non-prototype definitions are promoted. Therefore in this example the
2905 function definition's argument is really an @code{int}, which does not
2906 match the prototype argument type of @code{short}.
2908 This restriction of ISO C makes it hard to write code that is portable
2909 to traditional C compilers, because the programmer does not know
2910 whether the @code{uid_t} type is @code{short}, @code{int}, or
2911 @code{long}. Therefore, in cases like these GNU C allows a prototype
2912 to override a later old-style definition. More precisely, in GNU C, a
2913 function prototype argument type overrides the argument type specified
2914 by a later old-style definition if the former type is the same as the
2915 latter type before promotion. Thus in GNU C the above example is
2916 equivalent to the following:
2929 GNU C++ does not support old-style function definitions, so this
2930 extension is irrelevant.
2933 @section C++ Style Comments
2935 @cindex C++ comments
2936 @cindex comments, C++ style
2938 In GNU C, you may use C++ style comments, which start with @samp{//} and
2939 continue until the end of the line. Many other C implementations allow
2940 such comments, and they are included in the 1999 C standard. However,
2941 C++ style comments are not recognized if you specify an @option{-std}
2942 option specifying a version of ISO C before C99, or @option{-ansi}
2943 (equivalent to @option{-std=c89}).
2946 @section Dollar Signs in Identifier Names
2948 @cindex dollar signs in identifier names
2949 @cindex identifier names, dollar signs in
2951 In GNU C, you may normally use dollar signs in identifier names.
2952 This is because many traditional C implementations allow such identifiers.
2953 However, dollar signs in identifiers are not supported on a few target
2954 machines, typically because the target assembler does not allow them.
2956 @node Character Escapes
2957 @section The Character @key{ESC} in Constants
2959 You can use the sequence @samp{\e} in a string or character constant to
2960 stand for the ASCII character @key{ESC}.
2963 @section Inquiring on Alignment of Types or Variables
2965 @cindex type alignment
2966 @cindex variable alignment
2968 The keyword @code{__alignof__} allows you to inquire about how an object
2969 is aligned, or the minimum alignment usually required by a type. Its
2970 syntax is just like @code{sizeof}.
2972 For example, if the target machine requires a @code{double} value to be
2973 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2974 This is true on many RISC machines. On more traditional machine
2975 designs, @code{__alignof__ (double)} is 4 or even 2.
2977 Some machines never actually require alignment; they allow reference to any
2978 data type even at an odd address. For these machines, @code{__alignof__}
2979 reports the @emph{recommended} alignment of a type.
2981 If the operand of @code{__alignof__} is an lvalue rather than a type,
2982 its value is the required alignment for its type, taking into account
2983 any minimum alignment specified with GCC's @code{__attribute__}
2984 extension (@pxref{Variable Attributes}). For example, after this
2988 struct foo @{ int x; char y; @} foo1;
2992 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2993 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2995 It is an error to ask for the alignment of an incomplete type.
2997 @node Variable Attributes
2998 @section Specifying Attributes of Variables
2999 @cindex attribute of variables
3000 @cindex variable attributes
3002 The keyword @code{__attribute__} allows you to specify special
3003 attributes of variables or structure fields. This keyword is followed
3004 by an attribute specification inside double parentheses. Some
3005 attributes are currently defined generically for variables.
3006 Other attributes are defined for variables on particular target
3007 systems. Other attributes are available for functions
3008 @c APPLE LOCAL begin for-fsf-4_4 3274130 5295549
3009 (@pxref{Function Attributes}), types (@pxref{Type Attributes}) and
3010 labels (@pxref{Label Attributes}). Other front ends might define
3011 more attributes (@pxref{C++ Extensions,,Extensions to the C++ Language}).
3013 @c APPLE LOCAL end for-fsf-4_4 3274130 5295549
3014 You may also specify attributes with @samp{__} preceding and following
3015 each keyword. This allows you to use them in header files without
3016 being concerned about a possible macro of the same name. For example,
3017 you may use @code{__aligned__} instead of @code{aligned}.
3019 @xref{Attribute Syntax}, for details of the exact syntax for using
3023 @cindex @code{aligned} attribute
3024 @item aligned (@var{alignment})
3025 This attribute specifies a minimum alignment for the variable or
3026 structure field, measured in bytes. For example, the declaration:
3029 int x __attribute__ ((aligned (16))) = 0;
3033 causes the compiler to allocate the global variable @code{x} on a
3034 16-byte boundary. On a 68040, this could be used in conjunction with
3035 an @code{asm} expression to access the @code{move16} instruction which
3036 requires 16-byte aligned operands.
3038 You can also specify the alignment of structure fields. For example, to
3039 create a double-word aligned @code{int} pair, you could write:
3042 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
3046 This is an alternative to creating a union with a @code{double} member
3047 that forces the union to be double-word aligned.
3049 As in the preceding examples, you can explicitly specify the alignment
3050 (in bytes) that you wish the compiler to use for a given variable or
3051 structure field. Alternatively, you can leave out the alignment factor
3052 and just ask the compiler to align a variable or field to the maximum
3053 useful alignment for the target machine you are compiling for. For
3054 example, you could write:
3057 short array[3] __attribute__ ((aligned));
3060 Whenever you leave out the alignment factor in an @code{aligned} attribute
3061 specification, the compiler automatically sets the alignment for the declared
3062 variable or field to the largest alignment which is ever used for any data
3063 type on the target machine you are compiling for. Doing this can often make
3064 copy operations more efficient, because the compiler can use whatever
3065 instructions copy the biggest chunks of memory when performing copies to
3066 or from the variables or fields that you have aligned this way.
3068 The @code{aligned} attribute can only increase the alignment; but you
3069 can decrease it by specifying @code{packed} as well. See below.
3071 Note that the effectiveness of @code{aligned} attributes may be limited
3072 by inherent limitations in your linker. On many systems, the linker is
3073 only able to arrange for variables to be aligned up to a certain maximum
3074 alignment. (For some linkers, the maximum supported alignment may
3075 be very very small.) If your linker is only able to align variables
3076 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3077 in an @code{__attribute__} will still only provide you with 8 byte
3078 alignment. See your linker documentation for further information.
3080 The @code{aligned} attribute can also be used for functions
3081 (@pxref{Function Attributes}.)
3083 @item cleanup (@var{cleanup_function})
3084 @cindex @code{cleanup} attribute
3085 The @code{cleanup} attribute runs a function when the variable goes
3086 out of scope. This attribute can only be applied to auto function
3087 scope variables; it may not be applied to parameters or variables
3088 with static storage duration. The function must take one parameter,
3089 a pointer to a type compatible with the variable. The return value
3090 of the function (if any) is ignored.
3092 If @option{-fexceptions} is enabled, then @var{cleanup_function}
3093 will be run during the stack unwinding that happens during the
3094 processing of the exception. Note that the @code{cleanup} attribute
3095 does not allow the exception to be caught, only to perform an action.
3096 It is undefined what happens if @var{cleanup_function} does not
3101 @cindex @code{common} attribute
3102 @cindex @code{nocommon} attribute
3105 The @code{common} attribute requests GCC to place a variable in
3106 ``common'' storage. The @code{nocommon} attribute requests the
3107 opposite---to allocate space for it directly.
3109 These attributes override the default chosen by the
3110 @option{-fno-common} and @option{-fcommon} flags respectively.
3113 @cindex @code{deprecated} attribute
3114 The @code{deprecated} attribute results in a warning if the variable
3115 is used anywhere in the source file. This is useful when identifying
3116 variables that are expected to be removed in a future version of a
3117 program. The warning also includes the location of the declaration
3118 of the deprecated variable, to enable users to easily find further
3119 information about why the variable is deprecated, or what they should
3120 do instead. Note that the warning only occurs for uses:
3123 extern int old_var __attribute__ ((deprecated));
3125 int new_fn () @{ return old_var; @}
3128 results in a warning on line 3 but not line 2.
3130 The @code{deprecated} attribute can also be used for functions and
3131 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3133 @item mode (@var{mode})
3134 @cindex @code{mode} attribute
3135 This attribute specifies the data type for the declaration---whichever
3136 type corresponds to the mode @var{mode}. This in effect lets you
3137 request an integer or floating point type according to its width.
3139 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3140 indicate the mode corresponding to a one-byte integer, @samp{word} or
3141 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3142 or @samp{__pointer__} for the mode used to represent pointers.
3145 @cindex @code{packed} attribute
3146 The @code{packed} attribute specifies that a variable or structure field
3147 should have the smallest possible alignment---one byte for a variable,
3148 and one bit for a field, unless you specify a larger value with the
3149 @code{aligned} attribute.
3151 Here is a structure in which the field @code{x} is packed, so that it
3152 immediately follows @code{a}:
3158 int x[2] __attribute__ ((packed));
3162 @item section ("@var{section-name}")
3163 @cindex @code{section} variable attribute
3164 Normally, the compiler places the objects it generates in sections like
3165 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3166 or you need certain particular variables to appear in special sections,
3167 for example to map to special hardware. The @code{section}
3168 attribute specifies that a variable (or function) lives in a particular
3169 section. For example, this small program uses several specific section names:
3172 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3173 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3174 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3175 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3179 /* @r{Initialize stack pointer} */
3180 init_sp (stack + sizeof (stack));
3182 /* @r{Initialize initialized data} */
3183 memcpy (&init_data, &data, &edata - &data);
3185 /* @r{Turn on the serial ports} */
3192 Use the @code{section} attribute with an @emph{initialized} definition
3193 of a @emph{global} variable, as shown in the example. GCC issues
3194 a warning and otherwise ignores the @code{section} attribute in
3195 uninitialized variable declarations.
3197 You may only use the @code{section} attribute with a fully initialized
3198 global definition because of the way linkers work. The linker requires
3199 each object be defined once, with the exception that uninitialized
3200 variables tentatively go in the @code{common} (or @code{bss}) section
3201 and can be multiply ``defined''. You can force a variable to be
3202 initialized with the @option{-fno-common} flag or the @code{nocommon}
3205 Some file formats do not support arbitrary sections so the @code{section}
3206 attribute is not available on all platforms.
3207 If you need to map the entire contents of a module to a particular
3208 section, consider using the facilities of the linker instead.
3211 @cindex @code{shared} variable attribute
3212 On Microsoft Windows, in addition to putting variable definitions in a named
3213 section, the section can also be shared among all running copies of an
3214 executable or DLL@. For example, this small program defines shared data
3215 by putting it in a named section @code{shared} and marking the section
3219 int foo __attribute__((section ("shared"), shared)) = 0;
3224 /* @r{Read and write foo. All running
3225 copies see the same value.} */
3231 You may only use the @code{shared} attribute along with @code{section}
3232 attribute with a fully initialized global definition because of the way
3233 linkers work. See @code{section} attribute for more information.
3235 The @code{shared} attribute is only available on Microsoft Windows@.
3237 @item tls_model ("@var{tls_model}")
3238 @cindex @code{tls_model} attribute
3239 The @code{tls_model} attribute sets thread-local storage model
3240 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3241 overriding @option{-ftls-model=} command line switch on a per-variable
3243 The @var{tls_model} argument should be one of @code{global-dynamic},
3244 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3246 Not all targets support this attribute.
3249 This attribute, attached to a variable, means that the variable is meant
3250 to be possibly unused. GCC will not produce a warning for this
3254 This attribute, attached to a variable, means that the variable must be
3255 emitted even if it appears that the variable is not referenced.
3257 @item vector_size (@var{bytes})
3258 This attribute specifies the vector size for the variable, measured in
3259 bytes. For example, the declaration:
3262 int foo __attribute__ ((vector_size (16)));
3266 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3267 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3268 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3270 This attribute is only applicable to integral and float scalars,
3271 although arrays, pointers, and function return values are allowed in
3272 conjunction with this construct.
3274 Aggregates with this attribute are invalid, even if they are of the same
3275 size as a corresponding scalar. For example, the declaration:
3278 struct S @{ int a; @};
3279 struct S __attribute__ ((vector_size (16))) foo;
3283 is invalid even if the size of the structure is the same as the size of
3287 The @code{selectany} attribute causes an initialized global variable to
3288 have link-once semantics. When multiple definitions of the variable are
3289 encountered by the linker, the first is selected and the remainder are
3290 discarded. Following usage by the Microsoft compiler, the linker is told
3291 @emph{not} to warn about size or content differences of the multiple
3294 Although the primary usage of this attribute is for POD types, the
3295 attribute can also be applied to global C++ objects that are initialized
3296 by a constructor. In this case, the static initialization and destruction
3297 code for the object is emitted in each translation defining the object,
3298 but the calls to the constructor and destructor are protected by a
3299 link-once guard variable.
3301 The @code{selectany} attribute is only available on Microsoft Windows
3302 targets. You can use @code{__declspec (selectany)} as a synonym for
3303 @code{__attribute__ ((selectany))} for compatibility with other
3307 The @code{weak} attribute is described in @xref{Function Attributes}.
3310 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3313 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3317 @subsection M32R/D Variable Attributes
3319 One attribute is currently defined for the M32R/D@.
3322 @item model (@var{model-name})
3323 @cindex variable addressability on the M32R/D
3324 Use this attribute on the M32R/D to set the addressability of an object.
3325 The identifier @var{model-name} is one of @code{small}, @code{medium},
3326 or @code{large}, representing each of the code models.
3328 Small model objects live in the lower 16MB of memory (so that their
3329 addresses can be loaded with the @code{ld24} instruction).
3331 Medium and large model objects may live anywhere in the 32-bit address space
3332 (the compiler will generate @code{seth/add3} instructions to load their
3336 @anchor{i386 Variable Attributes}
3337 @subsection i386 Variable Attributes
3339 Two attributes are currently defined for i386 configurations:
3340 @code{ms_struct} and @code{gcc_struct}
3345 @cindex @code{ms_struct} attribute
3346 @cindex @code{gcc_struct} attribute
3348 If @code{packed} is used on a structure, or if bit-fields are used
3349 it may be that the Microsoft ABI packs them differently
3350 than GCC would normally pack them. Particularly when moving packed
3351 data between functions compiled with GCC and the native Microsoft compiler
3352 (either via function call or as data in a file), it may be necessary to access
3355 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3356 compilers to match the native Microsoft compiler.
3358 The Microsoft structure layout algorithm is fairly simple with the exception
3359 of the bitfield packing:
3361 The padding and alignment of members of structures and whether a bit field
3362 can straddle a storage-unit boundary
3365 @item Structure members are stored sequentially in the order in which they are
3366 declared: the first member has the lowest memory address and the last member
3369 @item Every data object has an alignment-requirement. The alignment-requirement
3370 for all data except structures, unions, and arrays is either the size of the
3371 object or the current packing size (specified with either the aligned attribute
3372 or the pack pragma), whichever is less. For structures, unions, and arrays,
3373 the alignment-requirement is the largest alignment-requirement of its members.
3374 Every object is allocated an offset so that:
3376 offset % alignment-requirement == 0
3378 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
3379 unit if the integral types are the same size and if the next bit field fits
3380 into the current allocation unit without crossing the boundary imposed by the
3381 common alignment requirements of the bit fields.
3384 Handling of zero-length bitfields:
3386 MSVC interprets zero-length bitfields in the following ways:
3389 @item If a zero-length bitfield is inserted between two bitfields that would
3390 normally be coalesced, the bitfields will not be coalesced.
3397 unsigned long bf_1 : 12;
3399 unsigned long bf_2 : 12;
3403 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
3404 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
3406 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
3407 alignment of the zero-length bitfield is greater than the member that follows it,
3408 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
3428 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
3429 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
3430 bitfield will not affect the alignment of @code{bar} or, as a result, the size
3433 Taking this into account, it is important to note the following:
3436 @item If a zero-length bitfield follows a normal bitfield, the type of the
3437 zero-length bitfield may affect the alignment of the structure as whole. For
3438 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
3439 normal bitfield, and is of type short.
3441 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
3442 still affect the alignment of the structure:
3452 Here, @code{t4} will take up 4 bytes.
3455 @item Zero-length bitfields following non-bitfield members are ignored:
3466 Here, @code{t5} will take up 2 bytes.
3470 @subsection PowerPC Variable Attributes
3472 Three attributes currently are defined for PowerPC configurations:
3473 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
3475 For full documentation of the struct attributes please see the
3476 documentation in the @xref{i386 Variable Attributes}, section.
3478 For documentation of @code{altivec} attribute please see the
3479 documentation in the @xref{PowerPC Type Attributes}, section.
3481 @subsection Xstormy16 Variable Attributes
3483 One attribute is currently defined for xstormy16 configurations:
3488 @cindex @code{below100} attribute
3490 If a variable has the @code{below100} attribute (@code{BELOW100} is
3491 allowed also), GCC will place the variable in the first 0x100 bytes of
3492 memory and use special opcodes to access it. Such variables will be
3493 placed in either the @code{.bss_below100} section or the
3494 @code{.data_below100} section.
3498 @node Type Attributes
3499 @section Specifying Attributes of Types
3500 @cindex attribute of types
3501 @cindex type attributes
3503 The keyword @code{__attribute__} allows you to specify special
3504 attributes of @code{struct} and @code{union} types when you define
3505 such types. This keyword is followed by an attribute specification
3506 inside double parentheses. Seven attributes are currently defined for
3507 types: @code{aligned}, @code{packed}, @code{transparent_union},
3508 @code{unused}, @code{deprecated}, @code{visibility}, and
3509 @code{may_alias}. Other attributes are defined for functions
3510 @c APPLE LOCAL begin for-fsf-4_4 3274130 5295549
3511 (@pxref{Function Attributes}), variables (@pxref{Variable
3512 Attributes}), and labels (@pxref{Label Attributes}).
3514 @c APPLE LOCAL end for-fsf-4_4 3274130 5295549
3515 You may also specify any one of these attributes with @samp{__}
3516 preceding and following its keyword. This allows you to use these
3517 attributes in header files without being concerned about a possible
3518 macro of the same name. For example, you may use @code{__aligned__}
3519 instead of @code{aligned}.
3521 You may specify type attributes either in a @code{typedef} declaration
3522 or in an enum, struct or union type declaration or definition.
3524 For an enum, struct or union type, you may specify attributes either
3525 between the enum, struct or union tag and the name of the type, or
3526 just past the closing curly brace of the @emph{definition}. The
3527 former syntax is preferred.
3529 @xref{Attribute Syntax}, for details of the exact syntax for using
3533 @cindex @code{aligned} attribute
3534 @item aligned (@var{alignment})
3535 This attribute specifies a minimum alignment (in bytes) for variables
3536 of the specified type. For example, the declarations:
3539 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3540 typedef int more_aligned_int __attribute__ ((aligned (8)));
3544 force the compiler to insure (as far as it can) that each variable whose
3545 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3546 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3547 variables of type @code{struct S} aligned to 8-byte boundaries allows
3548 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3549 store) instructions when copying one variable of type @code{struct S} to
3550 another, thus improving run-time efficiency.
3552 Note that the alignment of any given @code{struct} or @code{union} type
3553 is required by the ISO C standard to be at least a perfect multiple of
3554 the lowest common multiple of the alignments of all of the members of
3555 the @code{struct} or @code{union} in question. This means that you @emph{can}
3556 effectively adjust the alignment of a @code{struct} or @code{union}
3557 type by attaching an @code{aligned} attribute to any one of the members
3558 of such a type, but the notation illustrated in the example above is a
3559 more obvious, intuitive, and readable way to request the compiler to
3560 adjust the alignment of an entire @code{struct} or @code{union} type.
3562 As in the preceding example, you can explicitly specify the alignment
3563 (in bytes) that you wish the compiler to use for a given @code{struct}
3564 or @code{union} type. Alternatively, you can leave out the alignment factor
3565 and just ask the compiler to align a type to the maximum
3566 useful alignment for the target machine you are compiling for. For
3567 example, you could write:
3570 struct S @{ short f[3]; @} __attribute__ ((aligned));
3573 Whenever you leave out the alignment factor in an @code{aligned}
3574 attribute specification, the compiler automatically sets the alignment
3575 for the type to the largest alignment which is ever used for any data
3576 type on the target machine you are compiling for. Doing this can often
3577 make copy operations more efficient, because the compiler can use
3578 whatever instructions copy the biggest chunks of memory when performing
3579 copies to or from the variables which have types that you have aligned
3582 In the example above, if the size of each @code{short} is 2 bytes, then
3583 the size of the entire @code{struct S} type is 6 bytes. The smallest
3584 power of two which is greater than or equal to that is 8, so the
3585 compiler sets the alignment for the entire @code{struct S} type to 8
3588 Note that although you can ask the compiler to select a time-efficient
3589 alignment for a given type and then declare only individual stand-alone
3590 objects of that type, the compiler's ability to select a time-efficient
3591 alignment is primarily useful only when you plan to create arrays of
3592 variables having the relevant (efficiently aligned) type. If you
3593 declare or use arrays of variables of an efficiently-aligned type, then
3594 it is likely that your program will also be doing pointer arithmetic (or
3595 subscripting, which amounts to the same thing) on pointers to the
3596 relevant type, and the code that the compiler generates for these
3597 pointer arithmetic operations will often be more efficient for
3598 efficiently-aligned types than for other types.
3600 The @code{aligned} attribute can only increase the alignment; but you
3601 can decrease it by specifying @code{packed} as well. See below.
3603 Note that the effectiveness of @code{aligned} attributes may be limited
3604 by inherent limitations in your linker. On many systems, the linker is
3605 only able to arrange for variables to be aligned up to a certain maximum
3606 alignment. (For some linkers, the maximum supported alignment may
3607 be very very small.) If your linker is only able to align variables
3608 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3609 in an @code{__attribute__} will still only provide you with 8 byte
3610 alignment. See your linker documentation for further information.
3613 This attribute, attached to @code{struct} or @code{union} type
3614 definition, specifies that each member (other than zero-width bitfields)
3615 of the structure or union is placed to minimize the memory required. When
3616 attached to an @code{enum} definition, it indicates that the smallest
3617 integral type should be used.
3619 @opindex fshort-enums
3620 Specifying this attribute for @code{struct} and @code{union} types is
3621 equivalent to specifying the @code{packed} attribute on each of the
3622 structure or union members. Specifying the @option{-fshort-enums}
3623 flag on the line is equivalent to specifying the @code{packed}
3624 attribute on all @code{enum} definitions.
3626 In the following example @code{struct my_packed_struct}'s members are
3627 packed closely together, but the internal layout of its @code{s} member
3628 is not packed---to do that, @code{struct my_unpacked_struct} would need to
3632 struct my_unpacked_struct
3638 struct __attribute__ ((__packed__)) my_packed_struct
3642 struct my_unpacked_struct s;
3646 You may only specify this attribute on the definition of a @code{enum},
3647 @code{struct} or @code{union}, not on a @code{typedef} which does not
3648 also define the enumerated type, structure or union.
3650 @item transparent_union
3651 This attribute, attached to a @code{union} type definition, indicates
3652 that any function parameter having that union type causes calls to that
3653 function to be treated in a special way.
3655 First, the argument corresponding to a transparent union type can be of
3656 any type in the union; no cast is required. Also, if the union contains
3657 a pointer type, the corresponding argument can be a null pointer
3658 constant or a void pointer expression; and if the union contains a void
3659 pointer type, the corresponding argument can be any pointer expression.
3660 If the union member type is a pointer, qualifiers like @code{const} on
3661 the referenced type must be respected, just as with normal pointer
3664 Second, the argument is passed to the function using the calling
3665 conventions of the first member of the transparent union, not the calling
3666 conventions of the union itself. All members of the union must have the
3667 same machine representation; this is necessary for this argument passing
3670 Transparent unions are designed for library functions that have multiple
3671 interfaces for compatibility reasons. For example, suppose the
3672 @code{wait} function must accept either a value of type @code{int *} to
3673 comply with Posix, or a value of type @code{union wait *} to comply with
3674 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3675 @code{wait} would accept both kinds of arguments, but it would also
3676 accept any other pointer type and this would make argument type checking
3677 less useful. Instead, @code{<sys/wait.h>} might define the interface
3685 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3687 pid_t wait (wait_status_ptr_t);
3690 This interface allows either @code{int *} or @code{union wait *}
3691 arguments to be passed, using the @code{int *} calling convention.
3692 The program can call @code{wait} with arguments of either type:
3695 int w1 () @{ int w; return wait (&w); @}
3696 int w2 () @{ union wait w; return wait (&w); @}
3699 With this interface, @code{wait}'s implementation might look like this:
3702 pid_t wait (wait_status_ptr_t p)
3704 return waitpid (-1, p.__ip, 0);
3709 When attached to a type (including a @code{union} or a @code{struct}),
3710 this attribute means that variables of that type are meant to appear
3711 possibly unused. GCC will not produce a warning for any variables of
3712 that type, even if the variable appears to do nothing. This is often
3713 the case with lock or thread classes, which are usually defined and then
3714 not referenced, but contain constructors and destructors that have
3715 nontrivial bookkeeping functions.
3718 The @code{deprecated} attribute results in a warning if the type
3719 is used anywhere in the source file. This is useful when identifying
3720 types that are expected to be removed in a future version of a program.
3721 If possible, the warning also includes the location of the declaration
3722 of the deprecated type, to enable users to easily find further
3723 information about why the type is deprecated, or what they should do
3724 instead. Note that the warnings only occur for uses and then only
3725 if the type is being applied to an identifier that itself is not being
3726 declared as deprecated.
3729 typedef int T1 __attribute__ ((deprecated));
3733 typedef T1 T3 __attribute__ ((deprecated));
3734 T3 z __attribute__ ((deprecated));
3737 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3738 warning is issued for line 4 because T2 is not explicitly
3739 deprecated. Line 5 has no warning because T3 is explicitly
3740 deprecated. Similarly for line 6.
3742 The @code{deprecated} attribute can also be used for functions and
3743 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3746 Accesses to objects with types with this attribute are not subjected to
3747 type-based alias analysis, but are instead assumed to be able to alias
3748 any other type of objects, just like the @code{char} type. See
3749 @option{-fstrict-aliasing} for more information on aliasing issues.
3754 typedef short __attribute__((__may_alias__)) short_a;
3760 short_a *b = (short_a *) &a;
3764 if (a == 0x12345678)
3771 If you replaced @code{short_a} with @code{short} in the variable
3772 declaration, the above program would abort when compiled with
3773 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3774 above in recent GCC versions.
3777 In C++, attribute visibility (@pxref{Function Attributes}) can also be
3778 applied to class, struct, union and enum types. Unlike other type
3779 attributes, the attribute must appear between the initial keyword and
3780 the name of the type; it cannot appear after the body of the type.
3782 Note that the type visibility is applied to vague linkage entities
3783 associated with the class (vtable, typeinfo node, etc.). In
3784 particular, if a class is thrown as an exception in one shared object
3785 and caught in another, the class must have default visibility.
3786 Otherwise the two shared objects will be unable to use the same
3787 typeinfo node and exception handling will break.
3789 @subsection ARM Type Attributes
3791 On those ARM targets that support @code{dllimport} (such as Symbian
3792 OS), you can use the @code{notshared} attribute to indicate that the
3793 virtual table and other similar data for a class should not be
3794 exported from a DLL@. For example:
3797 class __declspec(notshared) C @{
3799 __declspec(dllimport) C();
3803 __declspec(dllexport)
3807 In this code, @code{C::C} is exported from the current DLL, but the
3808 virtual table for @code{C} is not exported. (You can use
3809 @code{__attribute__} instead of @code{__declspec} if you prefer, but
3810 most Symbian OS code uses @code{__declspec}.)
3812 @anchor{i386 Type Attributes}
3813 @subsection i386 Type Attributes
3815 Two attributes are currently defined for i386 configurations:
3816 @code{ms_struct} and @code{gcc_struct}
3820 @cindex @code{ms_struct}
3821 @cindex @code{gcc_struct}
3823 If @code{packed} is used on a structure, or if bit-fields are used
3824 it may be that the Microsoft ABI packs them differently
3825 than GCC would normally pack them. Particularly when moving packed
3826 data between functions compiled with GCC and the native Microsoft compiler
3827 (either via function call or as data in a file), it may be necessary to access
3830 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3831 compilers to match the native Microsoft compiler.
3834 To specify multiple attributes, separate them by commas within the
3835 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3838 @anchor{PowerPC Type Attributes}
3839 @subsection PowerPC Type Attributes
3841 Three attributes currently are defined for PowerPC configurations:
3842 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
3844 For full documentation of the struct attributes please see the
3845 documentation in the @xref{i386 Type Attributes}, section.
3847 The @code{altivec} attribute allows one to declare AltiVec vector data
3848 types supported by the AltiVec Programming Interface Manual. The
3849 attribute requires an argument to specify one of three vector types:
3850 @code{vector__}, @code{pixel__} (always followed by unsigned short),
3851 and @code{bool__} (always followed by unsigned).
3854 __attribute__((altivec(vector__)))
3855 __attribute__((altivec(pixel__))) unsigned short
3856 __attribute__((altivec(bool__))) unsigned
3859 These attributes mainly are intended to support the @code{__vector},
3860 @code{__pixel}, and @code{__bool} AltiVec keywords.
3862 @c APPLE LOCAL begin for-fsf-4_4 3274130 5295549
3863 @node Label Attributes
3864 @section Specifying Attributes of Labels and Statements
3865 @cindex attribute of labels
3866 @cindex label attributes
3867 @cindex attribute of statements
3868 @cindex statement attributes
3870 The keyword @code{__attribute__} allows you to specify special
3871 attributes of labels and statements.
3873 Some attributes are currently defined generically for variables.
3874 Other attributes are defined for variables on particular target
3875 systems. Other attributes are available for functions
3876 (@pxref{Function Attributes}), types (@pxref{Type Attributes}) and
3877 variables (@pxref{Variable Attributes}).
3879 You may also specify attributes with @samp{__} preceding and following
3880 each keyword. This allows you to use them in header files without
3881 being concerned about a possible macro of the same name. For example,
3882 you may use @code{__aligned__} instead of @code{aligned}.
3884 @xref{Attribute Syntax}, for details of the exact syntax for using
3888 @cindex @code{aligned} attribute
3889 @item aligned (@var{alignment})
3890 This attribute specifies a minimum alignment for the label,
3891 measured in bytes. For example, the declaration:
3894 some_label: __attribute__((aligned(16)))
3898 requests the compiler to align the label, inserting @code{nop}s as necessary,
3899 to a 16-byte boundary.
3901 The alignment is only a request. The compiler will usually be able to
3902 honour it but sometimes the label will be eliminated by the compiler,
3903 in which case its alignment will be eliminated too.
3905 When applied to loops, the @code{aligned} attribute causes the loop to
3909 When attached to a label this attribute means that the label might not
3910 be used. GCC will not produce a warning for the label, even if the
3911 label doesn't seem to be referenced. This feature is intended for
3912 code generated by programs which contains labels that may be unused
3913 but which is compiled with @option{-Wall}. It would not normally be
3914 appropriate to use in it human-written code, though it could be useful
3915 in cases where the code that jumps to the label is contained within an
3916 @code{#ifdef} conditional.
3918 This attribute can only be applied to labels, not statements, because
3919 there is no warning if a statement is removed.
3922 @c APPLE LOCAL end for-fsf-4_4 3274130 5295549
3924 @section An Inline Function is As Fast As a Macro
3925 @cindex inline functions
3926 @cindex integrating function code
3928 @cindex macros, inline alternative
3930 By declaring a function inline, you can direct GCC to make
3931 calls to that function faster. One way GCC can achieve this is to
3932 integrate that function's code into the code for its callers. This
3933 makes execution faster by eliminating the function-call overhead; in
3934 addition, if any of the actual argument values are constant, their
3935 known values may permit simplifications at compile time so that not
3936 all of the inline function's code needs to be included. The effect on
3937 code size is less predictable; object code may be larger or smaller
3938 with function inlining, depending on the particular case. You can
3939 also direct GCC to try to integrate all ``simple enough'' functions
3940 into their callers with the option @option{-finline-functions}.
3942 GCC implements three different semantics of declaring a function
3943 inline. One is available with @option{-std=gnu89}, another when
3944 @option{-std=c99} or @option{-std=gnu99}, and the third is used when
3947 To declare a function inline, use the @code{inline} keyword in its
3948 declaration, like this:
3958 If you are writing a header file to be included in ISO C89 programs, write
3959 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
3961 The three types of inlining behave similarly in two important cases:
3962 when the @code{inline} keyword is used on a @code{static} function,
3963 like the example above, and when a function is first declared without
3964 using the @code{inline} keyword and then is defined with
3965 @code{inline}, like this:
3968 extern int inc (int *a);
3976 In both of these common cases, the program behaves the same as if you
3977 had not used the @code{inline} keyword, except for its speed.
3979 @cindex inline functions, omission of
3980 @opindex fkeep-inline-functions
3981 When a function is both inline and @code{static}, if all calls to the
3982 function are integrated into the caller, and the function's address is
3983 never used, then the function's own assembler code is never referenced.
3984 In this case, GCC does not actually output assembler code for the
3985 function, unless you specify the option @option{-fkeep-inline-functions}.
3986 Some calls cannot be integrated for various reasons (in particular,
3987 calls that precede the function's definition cannot be integrated, and
3988 neither can recursive calls within the definition). If there is a
3989 nonintegrated call, then the function is compiled to assembler code as
3990 usual. The function must also be compiled as usual if the program
3991 refers to its address, because that can't be inlined.
3993 @cindex automatic @code{inline} for C++ member fns
3994 @cindex @code{inline} automatic for C++ member fns
3995 @cindex member fns, automatically @code{inline}
3996 @cindex C++ member fns, automatically @code{inline}
3997 @opindex fno-default-inline
3998 As required by ISO C++, GCC considers member functions defined within
3999 the body of a class to be marked inline even if they are
4000 not explicitly declared with the @code{inline} keyword. You can
4001 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
4002 Options,,Options Controlling C++ Dialect}.
4004 GCC does not inline any functions when not optimizing unless you specify
4005 the @samp{always_inline} attribute for the function, like this:
4008 /* @r{Prototype.} */
4009 inline void foo (const char) __attribute__((always_inline));
4012 The remainder of this section is specific to GNU C89 inlining.
4014 @cindex non-static inline function
4015 When an inline function is not @code{static}, then the compiler must assume
4016 that there may be calls from other source files; since a global symbol can
4017 be defined only once in any program, the function must not be defined in
4018 the other source files, so the calls therein cannot be integrated.
4019 Therefore, a non-@code{static} inline function is always compiled on its
4020 own in the usual fashion.
4022 If you specify both @code{inline} and @code{extern} in the function
4023 definition, then the definition is used only for inlining. In no case
4024 is the function compiled on its own, not even if you refer to its
4025 address explicitly. Such an address becomes an external reference, as
4026 if you had only declared the function, and had not defined it.
4028 This combination of @code{inline} and @code{extern} has almost the
4029 effect of a macro. The way to use it is to put a function definition in
4030 a header file with these keywords, and put another copy of the
4031 definition (lacking @code{inline} and @code{extern}) in a library file.
4032 The definition in the header file will cause most calls to the function
4033 to be inlined. If any uses of the function remain, they will refer to
4034 the single copy in the library.
4037 @section Assembler Instructions with C Expression Operands
4038 @cindex extended @code{asm}
4039 @cindex @code{asm} expressions
4040 @cindex assembler instructions
4043 In an assembler instruction using @code{asm}, you can specify the
4044 operands of the instruction using C expressions. This means you need not
4045 guess which registers or memory locations will contain the data you want
4048 You must specify an assembler instruction template much like what
4049 appears in a machine description, plus an operand constraint string for
4052 For example, here is how to use the 68881's @code{fsinx} instruction:
4055 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
4059 Here @code{angle} is the C expression for the input operand while
4060 @code{result} is that of the output operand. Each has @samp{"f"} as its
4061 operand constraint, saying that a floating point register is required.
4062 The @samp{=} in @samp{=f} indicates that the operand is an output; all
4063 output operands' constraints must use @samp{=}. The constraints use the
4064 same language used in the machine description (@pxref{Constraints}).
4066 Each operand is described by an operand-constraint string followed by
4067 the C expression in parentheses. A colon separates the assembler
4068 template from the first output operand and another separates the last
4069 output operand from the first input, if any. Commas separate the
4070 operands within each group. The total number of operands is currently
4071 limited to 30; this limitation may be lifted in some future version of
4074 If there are no output operands but there are input operands, you must
4075 place two consecutive colons surrounding the place where the output
4078 As of GCC version 3.1, it is also possible to specify input and output
4079 operands using symbolic names which can be referenced within the
4080 assembler code. These names are specified inside square brackets
4081 preceding the constraint string, and can be referenced inside the
4082 assembler code using @code{%[@var{name}]} instead of a percentage sign
4083 followed by the operand number. Using named operands the above example
4087 asm ("fsinx %[angle],%[output]"
4088 : [output] "=f" (result)
4089 : [angle] "f" (angle));
4093 Note that the symbolic operand names have no relation whatsoever to
4094 other C identifiers. You may use any name you like, even those of
4095 existing C symbols, but you must ensure that no two operands within the same
4096 assembler construct use the same symbolic name.
4098 Output operand expressions must be lvalues; the compiler can check this.
4099 The input operands need not be lvalues. The compiler cannot check
4100 whether the operands have data types that are reasonable for the
4101 instruction being executed. It does not parse the assembler instruction
4102 template and does not know what it means or even whether it is valid
4103 assembler input. The extended @code{asm} feature is most often used for
4104 machine instructions the compiler itself does not know exist. If
4105 the output expression cannot be directly addressed (for example, it is a
4106 bit-field), your constraint must allow a register. In that case, GCC
4107 will use the register as the output of the @code{asm}, and then store
4108 that register into the output.
4110 The ordinary output operands must be write-only; GCC will assume that
4111 the values in these operands before the instruction are dead and need
4112 not be generated. Extended asm supports input-output or read-write
4113 operands. Use the constraint character @samp{+} to indicate such an
4114 operand and list it with the output operands. You should only use
4115 read-write operands when the constraints for the operand (or the
4116 operand in which only some of the bits are to be changed) allow a
4119 You may, as an alternative, logically split its function into two
4120 separate operands, one input operand and one write-only output
4121 operand. The connection between them is expressed by constraints
4122 which say they need to be in the same location when the instruction
4123 executes. You can use the same C expression for both operands, or
4124 different expressions. For example, here we write the (fictitious)
4125 @samp{combine} instruction with @code{bar} as its read-only source
4126 operand and @code{foo} as its read-write destination:
4129 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
4133 The constraint @samp{"0"} for operand 1 says that it must occupy the
4134 same location as operand 0. A number in constraint is allowed only in
4135 an input operand and it must refer to an output operand.
4137 Only a number in the constraint can guarantee that one operand will be in
4138 the same place as another. The mere fact that @code{foo} is the value
4139 of both operands is not enough to guarantee that they will be in the
4140 same place in the generated assembler code. The following would not
4144 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
4147 Various optimizations or reloading could cause operands 0 and 1 to be in
4148 different registers; GCC knows no reason not to do so. For example, the
4149 compiler might find a copy of the value of @code{foo} in one register and
4150 use it for operand 1, but generate the output operand 0 in a different
4151 register (copying it afterward to @code{foo}'s own address). Of course,
4152 since the register for operand 1 is not even mentioned in the assembler
4153 code, the result will not work, but GCC can't tell that.
4155 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
4156 the operand number for a matching constraint. For example:
4159 asm ("cmoveq %1,%2,%[result]"
4160 : [result] "=r"(result)
4161 : "r" (test), "r"(new), "[result]"(old));
4164 Sometimes you need to make an @code{asm} operand be a specific register,
4165 but there's no matching constraint letter for that register @emph{by
4166 itself}. To force the operand into that register, use a local variable
4167 for the operand and specify the register in the variable declaration.
4168 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
4169 register constraint letter that matches the register:
4172 register int *p1 asm ("r0") = @dots{};
4173 register int *p2 asm ("r1") = @dots{};
4174 register int *result asm ("r0");
4175 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4178 @anchor{Example of asm with clobbered asm reg}
4179 In the above example, beware that a register that is call-clobbered by
4180 the target ABI will be overwritten by any function call in the
4181 assignment, including library calls for arithmetic operators.
4182 Assuming it is a call-clobbered register, this may happen to @code{r0}
4183 above by the assignment to @code{p2}. If you have to use such a
4184 register, use temporary variables for expressions between the register
4189 register int *p1 asm ("r0") = @dots{};
4190 register int *p2 asm ("r1") = t1;
4191 register int *result asm ("r0");
4192 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4195 Some instructions clobber specific hard registers. To describe this,
4196 write a third colon after the input operands, followed by the names of
4197 the clobbered hard registers (given as strings). Here is a realistic
4198 example for the VAX:
4201 asm volatile ("movc3 %0,%1,%2"
4202 : /* @r{no outputs} */
4203 : "g" (from), "g" (to), "g" (count)
4204 : "r0", "r1", "r2", "r3", "r4", "r5");
4207 You may not write a clobber description in a way that overlaps with an
4208 input or output operand. For example, you may not have an operand
4209 describing a register class with one member if you mention that register
4210 in the clobber list. Variables declared to live in specific registers
4211 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
4212 have no part mentioned in the clobber description.
4213 There is no way for you to specify that an input
4214 operand is modified without also specifying it as an output
4215 operand. Note that if all the output operands you specify are for this
4216 purpose (and hence unused), you will then also need to specify
4217 @code{volatile} for the @code{asm} construct, as described below, to
4218 prevent GCC from deleting the @code{asm} statement as unused.
4220 If you refer to a particular hardware register from the assembler code,
4221 you will probably have to list the register after the third colon to
4222 tell the compiler the register's value is modified. In some assemblers,
4223 the register names begin with @samp{%}; to produce one @samp{%} in the
4224 assembler code, you must write @samp{%%} in the input.
4226 If your assembler instruction can alter the condition code register, add
4227 @samp{cc} to the list of clobbered registers. GCC on some machines
4228 represents the condition codes as a specific hardware register;
4229 @samp{cc} serves to name this register. On other machines, the
4230 condition code is handled differently, and specifying @samp{cc} has no
4231 effect. But it is valid no matter what the machine.
4233 If your assembler instructions access memory in an unpredictable
4234 fashion, add @samp{memory} to the list of clobbered registers. This
4235 will cause GCC to not keep memory values cached in registers across the
4236 assembler instruction and not optimize stores or loads to that memory.
4237 You will also want to add the @code{volatile} keyword if the memory
4238 affected is not listed in the inputs or outputs of the @code{asm}, as
4239 the @samp{memory} clobber does not count as a side-effect of the
4240 @code{asm}. If you know how large the accessed memory is, you can add
4241 it as input or output but if this is not known, you should add
4242 @samp{memory}. As an example, if you access ten bytes of a string, you
4243 can use a memory input like:
4246 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
4249 Note that in the following example the memory input is necessary,
4250 otherwise GCC might optimize the store to @code{x} away:
4257 asm ("magic stuff accessing an 'int' pointed to by '%1'"
4258 "=&d" (r) : "a" (y), "m" (*y));
4263 You can put multiple assembler instructions together in a single
4264 @code{asm} template, separated by the characters normally used in assembly
4265 code for the system. A combination that works in most places is a newline
4266 to break the line, plus a tab character to move to the instruction field
4267 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
4268 assembler allows semicolons as a line-breaking character. Note that some
4269 assembler dialects use semicolons to start a comment.
4270 The input operands are guaranteed not to use any of the clobbered
4271 registers, and neither will the output operands' addresses, so you can
4272 read and write the clobbered registers as many times as you like. Here
4273 is an example of multiple instructions in a template; it assumes the
4274 subroutine @code{_foo} accepts arguments in registers 9 and 10:
4277 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
4279 : "g" (from), "g" (to)
4283 Unless an output operand has the @samp{&} constraint modifier, GCC
4284 may allocate it in the same register as an unrelated input operand, on
4285 the assumption the inputs are consumed before the outputs are produced.
4286 This assumption may be false if the assembler code actually consists of
4287 more than one instruction. In such a case, use @samp{&} for each output
4288 operand that may not overlap an input. @xref{Modifiers}.
4290 If you want to test the condition code produced by an assembler
4291 instruction, you must include a branch and a label in the @code{asm}
4292 construct, as follows:
4295 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
4301 This assumes your assembler supports local labels, as the GNU assembler
4302 and most Unix assemblers do.
4304 Speaking of labels, jumps from one @code{asm} to another are not
4305 supported. The compiler's optimizers do not know about these jumps, and
4306 therefore they cannot take account of them when deciding how to
4309 @cindex macros containing @code{asm}
4310 Usually the most convenient way to use these @code{asm} instructions is to
4311 encapsulate them in macros that look like functions. For example,
4315 (@{ double __value, __arg = (x); \
4316 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
4321 Here the variable @code{__arg} is used to make sure that the instruction
4322 operates on a proper @code{double} value, and to accept only those
4323 arguments @code{x} which can convert automatically to a @code{double}.
4325 Another way to make sure the instruction operates on the correct data
4326 type is to use a cast in the @code{asm}. This is different from using a
4327 variable @code{__arg} in that it converts more different types. For
4328 example, if the desired type were @code{int}, casting the argument to
4329 @code{int} would accept a pointer with no complaint, while assigning the
4330 argument to an @code{int} variable named @code{__arg} would warn about
4331 using a pointer unless the caller explicitly casts it.
4333 If an @code{asm} has output operands, GCC assumes for optimization
4334 purposes the instruction has no side effects except to change the output
4335 operands. This does not mean instructions with a side effect cannot be
4336 used, but you must be careful, because the compiler may eliminate them
4337 if the output operands aren't used, or move them out of loops, or
4338 replace two with one if they constitute a common subexpression. Also,
4339 if your instruction does have a side effect on a variable that otherwise
4340 appears not to change, the old value of the variable may be reused later
4341 if it happens to be found in a register.
4343 You can prevent an @code{asm} instruction from being deleted
4344 by writing the keyword @code{volatile} after
4345 the @code{asm}. For example:
4348 #define get_and_set_priority(new) \
4350 asm volatile ("get_and_set_priority %0, %1" \
4351 : "=g" (__old) : "g" (new)); \
4356 The @code{volatile} keyword indicates that the instruction has
4357 important side-effects. GCC will not delete a volatile @code{asm} if
4358 it is reachable. (The instruction can still be deleted if GCC can
4359 prove that control-flow will never reach the location of the
4360 instruction.) Note that even a volatile @code{asm} instruction
4361 can be moved relative to other code, including across jump
4362 instructions. For example, on many targets there is a system
4363 register which can be set to control the rounding mode of
4364 floating point operations. You might try
4365 setting it with a volatile @code{asm}, like this PowerPC example:
4368 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
4373 This will not work reliably, as the compiler may move the addition back
4374 before the volatile @code{asm}. To make it work you need to add an
4375 artificial dependency to the @code{asm} referencing a variable in the code
4376 you don't want moved, for example:
4379 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
4383 Similarly, you can't expect a
4384 sequence of volatile @code{asm} instructions to remain perfectly
4385 consecutive. If you want consecutive output, use a single @code{asm}.
4386 Also, GCC will perform some optimizations across a volatile @code{asm}
4387 instruction; GCC does not ``forget everything'' when it encounters
4388 a volatile @code{asm} instruction the way some other compilers do.
4390 An @code{asm} instruction without any output operands will be treated
4391 identically to a volatile @code{asm} instruction.
4393 It is a natural idea to look for a way to give access to the condition
4394 code left by the assembler instruction. However, when we attempted to
4395 implement this, we found no way to make it work reliably. The problem
4396 is that output operands might need reloading, which would result in
4397 additional following ``store'' instructions. On most machines, these
4398 instructions would alter the condition code before there was time to
4399 test it. This problem doesn't arise for ordinary ``test'' and
4400 ``compare'' instructions because they don't have any output operands.
4402 For reasons similar to those described above, it is not possible to give
4403 an assembler instruction access to the condition code left by previous
4406 If you are writing a header file that should be includable in ISO C
4407 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
4410 @subsection Size of an @code{asm}
4412 Some targets require that GCC track the size of each instruction used in
4413 order to generate correct code. Because the final length of an
4414 @code{asm} is only known by the assembler, GCC must make an estimate as
4415 to how big it will be. The estimate is formed by counting the number of
4416 statements in the pattern of the @code{asm} and multiplying that by the
4417 length of the longest instruction on that processor. Statements in the
4418 @code{asm} are identified by newline characters and whatever statement
4419 separator characters are supported by the assembler; on most processors
4420 this is the `@code{;}' character.
4422 Normally, GCC's estimate is perfectly adequate to ensure that correct
4423 code is generated, but it is possible to confuse the compiler if you use
4424 pseudo instructions or assembler macros that expand into multiple real
4425 instructions or if you use assembler directives that expand to more
4426 space in the object file than would be needed for a single instruction.
4427 If this happens then the assembler will produce a diagnostic saying that
4428 a label is unreachable.
4430 @subsection i386 floating point asm operands
4432 There are several rules on the usage of stack-like regs in
4433 asm_operands insns. These rules apply only to the operands that are
4438 Given a set of input regs that die in an asm_operands, it is
4439 necessary to know which are implicitly popped by the asm, and
4440 which must be explicitly popped by gcc.
4442 An input reg that is implicitly popped by the asm must be
4443 explicitly clobbered, unless it is constrained to match an
4447 For any input reg that is implicitly popped by an asm, it is
4448 necessary to know how to adjust the stack to compensate for the pop.
4449 If any non-popped input is closer to the top of the reg-stack than
4450 the implicitly popped reg, it would not be possible to know what the
4451 stack looked like---it's not clear how the rest of the stack ``slides
4454 All implicitly popped input regs must be closer to the top of
4455 the reg-stack than any input that is not implicitly popped.
4457 It is possible that if an input dies in an insn, reload might
4458 use the input reg for an output reload. Consider this example:
4461 asm ("foo" : "=t" (a) : "f" (b));
4464 This asm says that input B is not popped by the asm, and that
4465 the asm pushes a result onto the reg-stack, i.e., the stack is one
4466 deeper after the asm than it was before. But, it is possible that
4467 reload will think that it can use the same reg for both the input and
4468 the output, if input B dies in this insn.
4470 If any input operand uses the @code{f} constraint, all output reg
4471 constraints must use the @code{&} earlyclobber.
4473 The asm above would be written as
4476 asm ("foo" : "=&t" (a) : "f" (b));
4480 Some operands need to be in particular places on the stack. All
4481 output operands fall in this category---there is no other way to
4482 know which regs the outputs appear in unless the user indicates
4483 this in the constraints.
4485 Output operands must specifically indicate which reg an output
4486 appears in after an asm. @code{=f} is not allowed: the operand
4487 constraints must select a class with a single reg.
4490 Output operands may not be ``inserted'' between existing stack regs.
4491 Since no 387 opcode uses a read/write operand, all output operands
4492 are dead before the asm_operands, and are pushed by the asm_operands.
4493 It makes no sense to push anywhere but the top of the reg-stack.
4495 Output operands must start at the top of the reg-stack: output
4496 operands may not ``skip'' a reg.
4499 Some asm statements may need extra stack space for internal
4500 calculations. This can be guaranteed by clobbering stack registers
4501 unrelated to the inputs and outputs.
4505 Here are a couple of reasonable asms to want to write. This asm
4506 takes one input, which is internally popped, and produces two outputs.
4509 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4512 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4513 and replaces them with one output. The user must code the @code{st(1)}
4514 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4517 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4523 @section Controlling Names Used in Assembler Code
4524 @cindex assembler names for identifiers
4525 @cindex names used in assembler code
4526 @cindex identifiers, names in assembler code
4528 You can specify the name to be used in the assembler code for a C
4529 function or variable by writing the @code{asm} (or @code{__asm__})
4530 keyword after the declarator as follows:
4533 int foo asm ("myfoo") = 2;
4537 This specifies that the name to be used for the variable @code{foo} in
4538 the assembler code should be @samp{myfoo} rather than the usual
4541 On systems where an underscore is normally prepended to the name of a C
4542 function or variable, this feature allows you to define names for the
4543 linker that do not start with an underscore.
4545 It does not make sense to use this feature with a non-static local
4546 variable since such variables do not have assembler names. If you are
4547 trying to put the variable in a particular register, see @ref{Explicit
4548 Reg Vars}. GCC presently accepts such code with a warning, but will
4549 probably be changed to issue an error, rather than a warning, in the
4552 You cannot use @code{asm} in this way in a function @emph{definition}; but
4553 you can get the same effect by writing a declaration for the function
4554 before its definition and putting @code{asm} there, like this:
4557 extern func () asm ("FUNC");
4564 It is up to you to make sure that the assembler names you choose do not
4565 conflict with any other assembler symbols. Also, you must not use a
4566 register name; that would produce completely invalid assembler code. GCC
4567 does not as yet have the ability to store static variables in registers.
4568 Perhaps that will be added.
4570 @node Explicit Reg Vars
4571 @section Variables in Specified Registers
4572 @cindex explicit register variables
4573 @cindex variables in specified registers
4574 @cindex specified registers
4575 @cindex registers, global allocation
4577 GNU C allows you to put a few global variables into specified hardware
4578 registers. You can also specify the register in which an ordinary
4579 register variable should be allocated.
4583 Global register variables reserve registers throughout the program.
4584 This may be useful in programs such as programming language
4585 interpreters which have a couple of global variables that are accessed
4589 Local register variables in specific registers do not reserve the
4590 registers, except at the point where they are used as input or output
4591 operands in an @code{asm} statement and the @code{asm} statement itself is
4592 not deleted. The compiler's data flow analysis is capable of determining
4593 where the specified registers contain live values, and where they are
4594 available for other uses. Stores into local register variables may be deleted
4595 when they appear to be dead according to dataflow analysis. References
4596 to local register variables may be deleted or moved or simplified.
4598 These local variables are sometimes convenient for use with the extended
4599 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4600 output of the assembler instruction directly into a particular register.
4601 (This will work provided the register you specify fits the constraints
4602 specified for that operand in the @code{asm}.)
4610 @node Global Reg Vars
4611 @subsection Defining Global Register Variables
4612 @cindex global register variables
4613 @cindex registers, global variables in
4615 You can define a global register variable in GNU C like this:
4618 register int *foo asm ("a5");
4622 Here @code{a5} is the name of the register which should be used. Choose a
4623 register which is normally saved and restored by function calls on your
4624 machine, so that library routines will not clobber it.
4626 Naturally the register name is cpu-dependent, so you would need to
4627 conditionalize your program according to cpu type. The register
4628 @code{a5} would be a good choice on a 68000 for a variable of pointer
4629 type. On machines with register windows, be sure to choose a ``global''
4630 register that is not affected magically by the function call mechanism.
4632 In addition, operating systems on one type of cpu may differ in how they
4633 name the registers; then you would need additional conditionals. For
4634 example, some 68000 operating systems call this register @code{%a5}.
4636 Eventually there may be a way of asking the compiler to choose a register
4637 automatically, but first we need to figure out how it should choose and
4638 how to enable you to guide the choice. No solution is evident.
4640 Defining a global register variable in a certain register reserves that
4641 register entirely for this use, at least within the current compilation.
4642 The register will not be allocated for any other purpose in the functions
4643 in the current compilation. The register will not be saved and restored by
4644 these functions. Stores into this register are never deleted even if they
4645 would appear to be dead, but references may be deleted or moved or
4648 It is not safe to access the global register variables from signal
4649 handlers, or from more than one thread of control, because the system
4650 library routines may temporarily use the register for other things (unless
4651 you recompile them specially for the task at hand).
4653 @cindex @code{qsort}, and global register variables
4654 It is not safe for one function that uses a global register variable to
4655 call another such function @code{foo} by way of a third function
4656 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4657 different source file in which the variable wasn't declared). This is
4658 because @code{lose} might save the register and put some other value there.
4659 For example, you can't expect a global register variable to be available in
4660 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4661 might have put something else in that register. (If you are prepared to
4662 recompile @code{qsort} with the same global register variable, you can
4663 solve this problem.)
4665 If you want to recompile @code{qsort} or other source files which do not
4666 actually use your global register variable, so that they will not use that
4667 register for any other purpose, then it suffices to specify the compiler
4668 option @option{-ffixed-@var{reg}}. You need not actually add a global
4669 register declaration to their source code.
4671 A function which can alter the value of a global register variable cannot
4672 safely be called from a function compiled without this variable, because it
4673 could clobber the value the caller expects to find there on return.
4674 Therefore, the function which is the entry point into the part of the
4675 program that uses the global register variable must explicitly save and
4676 restore the value which belongs to its caller.
4678 @cindex register variable after @code{longjmp}
4679 @cindex global register after @code{longjmp}
4680 @cindex value after @code{longjmp}
4683 On most machines, @code{longjmp} will restore to each global register
4684 variable the value it had at the time of the @code{setjmp}. On some
4685 machines, however, @code{longjmp} will not change the value of global
4686 register variables. To be portable, the function that called @code{setjmp}
4687 should make other arrangements to save the values of the global register
4688 variables, and to restore them in a @code{longjmp}. This way, the same
4689 thing will happen regardless of what @code{longjmp} does.
4691 All global register variable declarations must precede all function
4692 definitions. If such a declaration could appear after function
4693 definitions, the declaration would be too late to prevent the register from
4694 being used for other purposes in the preceding functions.
4696 Global register variables may not have initial values, because an
4697 executable file has no means to supply initial contents for a register.
4699 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4700 registers, but certain library functions, such as @code{getwd}, as well
4701 as the subroutines for division and remainder, modify g3 and g4. g1 and
4702 g2 are local temporaries.
4704 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4705 Of course, it will not do to use more than a few of those.
4707 @node Local Reg Vars
4708 @subsection Specifying Registers for Local Variables
4709 @cindex local variables, specifying registers
4710 @cindex specifying registers for local variables
4711 @cindex registers for local variables
4713 You can define a local register variable with a specified register
4717 register int *foo asm ("a5");
4721 Here @code{a5} is the name of the register which should be used. Note
4722 that this is the same syntax used for defining global register
4723 variables, but for a local variable it would appear within a function.
4725 Naturally the register name is cpu-dependent, but this is not a
4726 problem, since specific registers are most often useful with explicit
4727 assembler instructions (@pxref{Extended Asm}). Both of these things
4728 generally require that you conditionalize your program according to
4731 In addition, operating systems on one type of cpu may differ in how they
4732 name the registers; then you would need additional conditionals. For
4733 example, some 68000 operating systems call this register @code{%a5}.
4735 Defining such a register variable does not reserve the register; it
4736 remains available for other uses in places where flow control determines
4737 the variable's value is not live.
4739 This option does not guarantee that GCC will generate code that has
4740 this variable in the register you specify at all times. You may not
4741 code an explicit reference to this register in the @emph{assembler
4742 instruction template} part of an @code{asm} statement and assume it will
4743 always refer to this variable. However, using the variable as an
4744 @code{asm} @emph{operand} guarantees that the specified register is used
4747 Stores into local register variables may be deleted when they appear to be dead
4748 according to dataflow analysis. References to local register variables may
4749 be deleted or moved or simplified.
4751 As for global register variables, it's recommended that you choose a
4752 register which is normally saved and restored by function calls on
4753 your machine, so that library routines will not clobber it. A common
4754 pitfall is to initialize multiple call-clobbered registers with
4755 arbitrary expressions, where a function call or library call for an
4756 arithmetic operator will overwrite a register value from a previous
4757 assignment, for example @code{r0} below:
4759 register int *p1 asm ("r0") = @dots{};
4760 register int *p2 asm ("r1") = @dots{};
4762 In those cases, a solution is to use a temporary variable for
4763 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
4765 @node Alternate Keywords
4766 @section Alternate Keywords
4767 @cindex alternate keywords
4768 @cindex keywords, alternate
4770 @option{-ansi} and the various @option{-std} options disable certain
4771 keywords. This causes trouble when you want to use GNU C extensions, or
4772 a general-purpose header file that should be usable by all programs,
4773 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4774 @code{inline} are not available in programs compiled with
4775 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4776 program compiled with @option{-std=c99}). The ISO C99 keyword
4777 @code{restrict} is only available when @option{-std=gnu99} (which will
4778 eventually be the default) or @option{-std=c99} (or the equivalent
4779 @option{-std=iso9899:1999}) is used.
4781 The way to solve these problems is to put @samp{__} at the beginning and
4782 end of each problematical keyword. For example, use @code{__asm__}
4783 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4785 Other C compilers won't accept these alternative keywords; if you want to
4786 compile with another compiler, you can define the alternate keywords as
4787 macros to replace them with the customary keywords. It looks like this:
4795 @findex __extension__
4797 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4799 prevent such warnings within one expression by writing
4800 @code{__extension__} before the expression. @code{__extension__} has no
4801 effect aside from this.
4803 @node Incomplete Enums
4804 @section Incomplete @code{enum} Types
4806 You can define an @code{enum} tag without specifying its possible values.
4807 This results in an incomplete type, much like what you get if you write
4808 @code{struct foo} without describing the elements. A later declaration
4809 which does specify the possible values completes the type.
4811 You can't allocate variables or storage using the type while it is
4812 incomplete. However, you can work with pointers to that type.
4814 This extension may not be very useful, but it makes the handling of
4815 @code{enum} more consistent with the way @code{struct} and @code{union}
4818 This extension is not supported by GNU C++.
4820 @node Function Names
4821 @section Function Names as Strings
4822 @cindex @code{__func__} identifier
4823 @cindex @code{__FUNCTION__} identifier
4824 @cindex @code{__PRETTY_FUNCTION__} identifier
4826 GCC provides three magic variables which hold the name of the current
4827 function, as a string. The first of these is @code{__func__}, which
4828 is part of the C99 standard:
4831 The identifier @code{__func__} is implicitly declared by the translator
4832 as if, immediately following the opening brace of each function
4833 definition, the declaration
4836 static const char __func__[] = "function-name";
4839 appeared, where function-name is the name of the lexically-enclosing
4840 function. This name is the unadorned name of the function.
4843 @code{__FUNCTION__} is another name for @code{__func__}. Older
4844 versions of GCC recognize only this name. However, it is not
4845 standardized. For maximum portability, we recommend you use
4846 @code{__func__}, but provide a fallback definition with the
4850 #if __STDC_VERSION__ < 199901L
4852 # define __func__ __FUNCTION__
4854 # define __func__ "<unknown>"
4859 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4860 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4861 the type signature of the function as well as its bare name. For
4862 example, this program:
4866 extern int printf (char *, ...);
4873 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4874 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4892 __PRETTY_FUNCTION__ = void a::sub(int)
4895 These identifiers are not preprocessor macros. In GCC 3.3 and
4896 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4897 were treated as string literals; they could be used to initialize
4898 @code{char} arrays, and they could be concatenated with other string
4899 literals. GCC 3.4 and later treat them as variables, like
4900 @code{__func__}. In C++, @code{__FUNCTION__} and
4901 @code{__PRETTY_FUNCTION__} have always been variables.
4903 @node Return Address
4904 @section Getting the Return or Frame Address of a Function
4906 These functions may be used to get information about the callers of a
4909 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4910 This function returns the return address of the current function, or of
4911 one of its callers. The @var{level} argument is number of frames to
4912 scan up the call stack. A value of @code{0} yields the return address
4913 of the current function, a value of @code{1} yields the return address
4914 of the caller of the current function, and so forth. When inlining
4915 the expected behavior is that the function will return the address of
4916 the function that will be returned to. To work around this behavior use
4917 the @code{noinline} function attribute.
4919 The @var{level} argument must be a constant integer.
4921 On some machines it may be impossible to determine the return address of
4922 any function other than the current one; in such cases, or when the top
4923 of the stack has been reached, this function will return @code{0} or a
4924 random value. In addition, @code{__builtin_frame_address} may be used
4925 to determine if the top of the stack has been reached.
4927 This function should only be used with a nonzero argument for debugging
4931 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4932 This function is similar to @code{__builtin_return_address}, but it
4933 returns the address of the function frame rather than the return address
4934 of the function. Calling @code{__builtin_frame_address} with a value of
4935 @code{0} yields the frame address of the current function, a value of
4936 @code{1} yields the frame address of the caller of the current function,
4939 The frame is the area on the stack which holds local variables and saved
4940 registers. The frame address is normally the address of the first word
4941 pushed on to the stack by the function. However, the exact definition
4942 depends upon the processor and the calling convention. If the processor
4943 has a dedicated frame pointer register, and the function has a frame,
4944 then @code{__builtin_frame_address} will return the value of the frame
4947 On some machines it may be impossible to determine the frame address of
4948 any function other than the current one; in such cases, or when the top
4949 of the stack has been reached, this function will return @code{0} if
4950 the first frame pointer is properly initialized by the startup code.
4952 This function should only be used with a nonzero argument for debugging
4956 @node Vector Extensions
4957 @section Using vector instructions through built-in functions
4959 On some targets, the instruction set contains SIMD vector instructions that
4960 operate on multiple values contained in one large register at the same time.
4961 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4964 The first step in using these extensions is to provide the necessary data
4965 types. This should be done using an appropriate @code{typedef}:
4968 typedef int v4si __attribute__ ((vector_size (16)));
4971 The @code{int} type specifies the base type, while the attribute specifies
4972 the vector size for the variable, measured in bytes. For example, the
4973 declaration above causes the compiler to set the mode for the @code{v4si}
4974 type to be 16 bytes wide and divided into @code{int} sized units. For
4975 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4976 corresponding mode of @code{foo} will be @acronym{V4SI}.
4978 The @code{vector_size} attribute is only applicable to integral and
4979 float scalars, although arrays, pointers, and function return values
4980 are allowed in conjunction with this construct.
4982 All the basic integer types can be used as base types, both as signed
4983 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4984 @code{long long}. In addition, @code{float} and @code{double} can be
4985 used to build floating-point vector types.
4987 Specifying a combination that is not valid for the current architecture
4988 will cause GCC to synthesize the instructions using a narrower mode.
4989 For example, if you specify a variable of type @code{V4SI} and your
4990 architecture does not allow for this specific SIMD type, GCC will
4991 produce code that uses 4 @code{SIs}.
4993 The types defined in this manner can be used with a subset of normal C
4994 operations. Currently, GCC will allow using the following operators
4995 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4997 The operations behave like C++ @code{valarrays}. Addition is defined as
4998 the addition of the corresponding elements of the operands. For
4999 example, in the code below, each of the 4 elements in @var{a} will be
5000 added to the corresponding 4 elements in @var{b} and the resulting
5001 vector will be stored in @var{c}.
5004 typedef int v4si __attribute__ ((vector_size (16)));
5011 Subtraction, multiplication, division, and the logical operations
5012 operate in a similar manner. Likewise, the result of using the unary
5013 minus or complement operators on a vector type is a vector whose
5014 elements are the negative or complemented values of the corresponding
5015 elements in the operand.
5017 You can declare variables and use them in function calls and returns, as
5018 well as in assignments and some casts. You can specify a vector type as
5019 a return type for a function. Vector types can also be used as function
5020 arguments. It is possible to cast from one vector type to another,
5021 provided they are of the same size (in fact, you can also cast vectors
5022 to and from other datatypes of the same size).
5024 You cannot operate between vectors of different lengths or different
5025 signedness without a cast.
5027 A port that supports hardware vector operations, usually provides a set
5028 of built-in functions that can be used to operate on vectors. For
5029 example, a function to add two vectors and multiply the result by a
5030 third could look like this:
5033 v4si f (v4si a, v4si b, v4si c)
5035 v4si tmp = __builtin_addv4si (a, b);
5036 return __builtin_mulv4si (tmp, c);
5043 @findex __builtin_offsetof
5045 GCC implements for both C and C++ a syntactic extension to implement
5046 the @code{offsetof} macro.
5050 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
5052 offsetof_member_designator:
5054 | offsetof_member_designator "." @code{identifier}
5055 | offsetof_member_designator "[" @code{expr} "]"
5058 This extension is sufficient such that
5061 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
5064 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
5065 may be dependent. In either case, @var{member} may consist of a single
5066 identifier, or a sequence of member accesses and array references.
5068 @node Atomic Builtins
5069 @section Built-in functions for atomic memory access
5071 The following builtins are intended to be compatible with those described
5072 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
5073 section 7.4. As such, they depart from the normal GCC practice of using
5074 the ``__builtin_'' prefix, and further that they are overloaded such that
5075 they work on multiple types.
5077 The definition given in the Intel documentation allows only for the use of
5078 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
5079 counterparts. GCC will allow any integral scalar or pointer type that is
5080 1, 2, 4 or 8 bytes in length.
5082 Not all operations are supported by all target processors. If a particular
5083 operation cannot be implemented on the target processor, a warning will be
5084 generated and a call an external function will be generated. The external
5085 function will carry the same name as the builtin, with an additional suffix
5086 @samp{_@var{n}} where @var{n} is the size of the data type.
5088 @c ??? Should we have a mechanism to suppress this warning? This is almost
5089 @c useful for implementing the operation under the control of an external
5092 In most cases, these builtins are considered a @dfn{full barrier}. That is,
5093 no memory operand will be moved across the operation, either forward or
5094 backward. Further, instructions will be issued as necessary to prevent the
5095 processor from speculating loads across the operation and from queuing stores
5096 after the operation.
5098 All of the routines are are described in the Intel documentation to take
5099 ``an optional list of variables protected by the memory barrier''. It's
5100 not clear what is meant by that; it could mean that @emph{only} the
5101 following variables are protected, or it could mean that these variables
5102 should in addition be protected. At present GCC ignores this list and
5103 protects all variables which are globally accessible. If in the future
5104 we make some use of this list, an empty list will continue to mean all
5105 globally accessible variables.
5108 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
5109 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
5110 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
5111 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
5112 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
5113 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
5114 @findex __sync_fetch_and_add
5115 @findex __sync_fetch_and_sub
5116 @findex __sync_fetch_and_or
5117 @findex __sync_fetch_and_and
5118 @findex __sync_fetch_and_xor
5119 @findex __sync_fetch_and_nand
5120 These builtins perform the operation suggested by the name, and
5121 returns the value that had previously been in memory. That is,
5124 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
5125 @{ tmp = *ptr; *ptr = ~tmp & value; return tmp; @} // nand
5128 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
5129 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
5130 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
5131 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
5132 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
5133 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
5134 @findex __sync_add_and_fetch
5135 @findex __sync_sub_and_fetch
5136 @findex __sync_or_and_fetch
5137 @findex __sync_and_and_fetch
5138 @findex __sync_xor_and_fetch
5139 @findex __sync_nand_and_fetch
5140 These builtins perform the operation suggested by the name, and
5141 return the new value. That is,
5144 @{ *ptr @var{op}= value; return *ptr; @}
5145 @{ *ptr = ~*ptr & value; return *ptr; @} // nand
5148 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5149 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5150 @findex __sync_bool_compare_and_swap
5151 @findex __sync_val_compare_and_swap
5152 These builtins perform an atomic compare and swap. That is, if the current
5153 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
5156 The ``bool'' version returns true if the comparison is successful and
5157 @var{newval} was written. The ``val'' version returns the contents
5158 of @code{*@var{ptr}} before the operation.
5160 @item __sync_synchronize (...)
5161 @findex __sync_synchronize
5162 This builtin issues a full memory barrier.
5164 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
5165 @findex __sync_lock_test_and_set
5166 This builtin, as described by Intel, is not a traditional test-and-set
5167 operation, but rather an atomic exchange operation. It writes @var{value}
5168 into @code{*@var{ptr}}, and returns the previous contents of
5171 Many targets have only minimal support for such locks, and do not support
5172 a full exchange operation. In this case, a target may support reduced
5173 functionality here by which the @emph{only} valid value to store is the
5174 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
5175 is implementation defined.
5177 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
5178 This means that references after the builtin cannot move to (or be
5179 speculated to) before the builtin, but previous memory stores may not
5180 be globally visible yet, and previous memory loads may not yet be
5183 @item void __sync_lock_release (@var{type} *ptr, ...)
5184 @findex __sync_lock_release
5185 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
5186 Normally this means writing the constant 0 to @code{*@var{ptr}}.
5188 This builtin is not a full barrier, but rather a @dfn{release barrier}.
5189 This means that all previous memory stores are globally visible, and all
5190 previous memory loads have been satisfied, but following memory reads
5191 are not prevented from being speculated to before the barrier.
5194 @node Object Size Checking
5195 @section Object Size Checking Builtins
5196 @findex __builtin_object_size
5197 @findex __builtin___memcpy_chk
5198 @findex __builtin___mempcpy_chk
5199 @findex __builtin___memmove_chk
5200 @findex __builtin___memset_chk
5201 @findex __builtin___strcpy_chk
5202 @findex __builtin___stpcpy_chk
5203 @findex __builtin___strncpy_chk
5204 @findex __builtin___strcat_chk
5205 @findex __builtin___strncat_chk
5206 @findex __builtin___sprintf_chk
5207 @findex __builtin___snprintf_chk
5208 @findex __builtin___vsprintf_chk
5209 @findex __builtin___vsnprintf_chk
5210 @findex __builtin___printf_chk
5211 @findex __builtin___vprintf_chk
5212 @findex __builtin___fprintf_chk
5213 @findex __builtin___vfprintf_chk
5215 GCC implements a limited buffer overflow protection mechanism
5216 that can prevent some buffer overflow attacks.
5218 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
5219 is a built-in construct that returns a constant number of bytes from
5220 @var{ptr} to the end of the object @var{ptr} pointer points to
5221 (if known at compile time). @code{__builtin_object_size} never evaluates
5222 its arguments for side-effects. If there are any side-effects in them, it
5223 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5224 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
5225 point to and all of them are known at compile time, the returned number
5226 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
5227 0 and minimum if nonzero. If it is not possible to determine which objects
5228 @var{ptr} points to at compile time, @code{__builtin_object_size} should
5229 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5230 for @var{type} 2 or 3.
5232 @var{type} is an integer constant from 0 to 3. If the least significant
5233 bit is clear, objects are whole variables, if it is set, a closest
5234 surrounding subobject is considered the object a pointer points to.
5235 The second bit determines if maximum or minimum of remaining bytes
5239 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
5240 char *p = &var.buf1[1], *q = &var.b;
5242 /* Here the object p points to is var. */
5243 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
5244 /* The subobject p points to is var.buf1. */
5245 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
5246 /* The object q points to is var. */
5247 assert (__builtin_object_size (q, 0)
5248 == (char *) (&var + 1) - (char *) &var.b);
5249 /* The subobject q points to is var.b. */
5250 assert (__builtin_object_size (q, 1) == sizeof (var.b));
5254 There are built-in functions added for many common string operation
5255 functions, e.g. for @code{memcpy} @code{__builtin___memcpy_chk}
5256 built-in is provided. This built-in has an additional last argument,
5257 which is the number of bytes remaining in object the @var{dest}
5258 argument points to or @code{(size_t) -1} if the size is not known.
5260 The built-in functions are optimized into the normal string functions
5261 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
5262 it is known at compile time that the destination object will not
5263 be overflown. If the compiler can determine at compile time the
5264 object will be always overflown, it issues a warning.
5266 The intended use can be e.g.
5270 #define bos0(dest) __builtin_object_size (dest, 0)
5271 #define memcpy(dest, src, n) \
5272 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
5276 /* It is unknown what object p points to, so this is optimized
5277 into plain memcpy - no checking is possible. */
5278 memcpy (p, "abcde", n);
5279 /* Destination is known and length too. It is known at compile
5280 time there will be no overflow. */
5281 memcpy (&buf[5], "abcde", 5);
5282 /* Destination is known, but the length is not known at compile time.
5283 This will result in __memcpy_chk call that can check for overflow
5285 memcpy (&buf[5], "abcde", n);
5286 /* Destination is known and it is known at compile time there will
5287 be overflow. There will be a warning and __memcpy_chk call that
5288 will abort the program at runtime. */
5289 memcpy (&buf[6], "abcde", 5);
5292 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
5293 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
5294 @code{strcat} and @code{strncat}.
5296 There are also checking built-in functions for formatted output functions.
5298 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
5299 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5300 const char *fmt, ...);
5301 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
5303 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5304 const char *fmt, va_list ap);
5307 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
5308 etc. functions and can contain implementation specific flags on what
5309 additional security measures the checking function might take, such as
5310 handling @code{%n} differently.
5312 The @var{os} argument is the object size @var{s} points to, like in the
5313 other built-in functions. There is a small difference in the behavior
5314 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
5315 optimized into the non-checking functions only if @var{flag} is 0, otherwise
5316 the checking function is called with @var{os} argument set to
5319 In addition to this, there are checking built-in functions
5320 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
5321 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
5322 These have just one additional argument, @var{flag}, right before
5323 format string @var{fmt}. If the compiler is able to optimize them to
5324 @code{fputc} etc. functions, it will, otherwise the checking function
5325 should be called and the @var{flag} argument passed to it.
5327 @node Other Builtins
5328 @section Other built-in functions provided by GCC
5329 @cindex built-in functions
5330 @findex __builtin_isgreater
5331 @findex __builtin_isgreaterequal
5332 @findex __builtin_isless
5333 @findex __builtin_islessequal
5334 @findex __builtin_islessgreater
5335 @findex __builtin_isunordered
5336 @findex __builtin_powi
5337 @findex __builtin_powif
5338 @findex __builtin_powil
5496 @findex fprintf_unlocked
5498 @findex fputs_unlocked
5608 @findex printf_unlocked
5637 @findex significandf
5638 @findex significandl
5709 GCC provides a large number of built-in functions other than the ones
5710 mentioned above. Some of these are for internal use in the processing
5711 of exceptions or variable-length argument lists and will not be
5712 documented here because they may change from time to time; we do not
5713 recommend general use of these functions.
5715 The remaining functions are provided for optimization purposes.
5717 @opindex fno-builtin
5718 GCC includes built-in versions of many of the functions in the standard
5719 C library. The versions prefixed with @code{__builtin_} will always be
5720 treated as having the same meaning as the C library function even if you
5721 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
5722 Many of these functions are only optimized in certain cases; if they are
5723 not optimized in a particular case, a call to the library function will
5728 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
5729 @option{-std=c99}), the functions
5730 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
5731 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
5732 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
5733 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
5734 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
5735 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
5736 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
5737 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
5738 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
5739 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
5740 @code{significandf}, @code{significandl}, @code{significand},
5741 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
5742 @code{stpncpy}, @code{strcasecmp}, @code{strdup}, @code{strfmon},
5743 @code{strncasecmp}, @code{strndup}, @code{toascii}, @code{y0f},
5744 @code{y0l}, @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf},
5745 @code{ynl} and @code{yn}
5746 may be handled as built-in functions.
5747 All these functions have corresponding versions
5748 prefixed with @code{__builtin_}, which may be used even in strict C89
5751 The ISO C99 functions
5752 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
5753 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
5754 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
5755 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
5756 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
5757 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
5758 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
5759 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
5760 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
5761 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
5762 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
5763 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
5764 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
5765 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
5766 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
5767 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
5768 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
5769 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
5770 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
5771 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
5772 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
5773 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
5774 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
5775 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
5776 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
5777 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
5778 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
5779 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
5780 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
5781 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
5782 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
5783 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
5784 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
5785 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
5786 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
5787 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
5788 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
5789 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
5790 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
5791 are handled as built-in functions
5792 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5794 There are also built-in versions of the ISO C99 functions
5795 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5796 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5797 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5798 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5799 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5800 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5801 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5802 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5803 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5804 that are recognized in any mode since ISO C90 reserves these names for
5805 the purpose to which ISO C99 puts them. All these functions have
5806 corresponding versions prefixed with @code{__builtin_}.
5808 The ISO C94 functions
5809 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
5810 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
5811 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
5813 are handled as built-in functions
5814 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5816 The ISO C90 functions
5817 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5818 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5819 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5820 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
5821 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
5822 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
5823 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
5824 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
5825 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
5826 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
5827 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
5828 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
5829 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
5830 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
5831 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
5832 @code{vprintf} and @code{vsprintf}
5833 are all recognized as built-in functions unless
5834 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5835 is specified for an individual function). All of these functions have
5836 corresponding versions prefixed with @code{__builtin_}.
5838 GCC provides built-in versions of the ISO C99 floating point comparison
5839 macros that avoid raising exceptions for unordered operands. They have
5840 the same names as the standard macros ( @code{isgreater},
5841 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5842 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5843 prefixed. We intend for a library implementor to be able to simply
5844 @code{#define} each standard macro to its built-in equivalent.
5846 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
5848 You can use the built-in function @code{__builtin_types_compatible_p} to
5849 determine whether two types are the same.
5851 This built-in function returns 1 if the unqualified versions of the
5852 types @var{type1} and @var{type2} (which are types, not expressions) are
5853 compatible, 0 otherwise. The result of this built-in function can be
5854 used in integer constant expressions.
5856 This built-in function ignores top level qualifiers (e.g., @code{const},
5857 @code{volatile}). For example, @code{int} is equivalent to @code{const
5860 The type @code{int[]} and @code{int[5]} are compatible. On the other
5861 hand, @code{int} and @code{char *} are not compatible, even if the size
5862 of their types, on the particular architecture are the same. Also, the
5863 amount of pointer indirection is taken into account when determining
5864 similarity. Consequently, @code{short *} is not similar to
5865 @code{short **}. Furthermore, two types that are typedefed are
5866 considered compatible if their underlying types are compatible.
5868 An @code{enum} type is not considered to be compatible with another
5869 @code{enum} type even if both are compatible with the same integer
5870 type; this is what the C standard specifies.
5871 For example, @code{enum @{foo, bar@}} is not similar to
5872 @code{enum @{hot, dog@}}.
5874 You would typically use this function in code whose execution varies
5875 depending on the arguments' types. For example:
5880 typeof (x) tmp = (x); \
5881 if (__builtin_types_compatible_p (typeof (x), long double)) \
5882 tmp = foo_long_double (tmp); \
5883 else if (__builtin_types_compatible_p (typeof (x), double)) \
5884 tmp = foo_double (tmp); \
5885 else if (__builtin_types_compatible_p (typeof (x), float)) \
5886 tmp = foo_float (tmp); \
5893 @emph{Note:} This construct is only available for C@.
5897 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
5899 You can use the built-in function @code{__builtin_choose_expr} to
5900 evaluate code depending on the value of a constant expression. This
5901 built-in function returns @var{exp1} if @var{const_exp}, which is a
5902 constant expression that must be able to be determined at compile time,
5903 is nonzero. Otherwise it returns 0.
5905 This built-in function is analogous to the @samp{? :} operator in C,
5906 except that the expression returned has its type unaltered by promotion
5907 rules. Also, the built-in function does not evaluate the expression
5908 that was not chosen. For example, if @var{const_exp} evaluates to true,
5909 @var{exp2} is not evaluated even if it has side-effects.
5911 This built-in function can return an lvalue if the chosen argument is an
5914 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5915 type. Similarly, if @var{exp2} is returned, its return type is the same
5922 __builtin_choose_expr ( \
5923 __builtin_types_compatible_p (typeof (x), double), \
5925 __builtin_choose_expr ( \
5926 __builtin_types_compatible_p (typeof (x), float), \
5928 /* @r{The void expression results in a compile-time error} \
5929 @r{when assigning the result to something.} */ \
5933 @emph{Note:} This construct is only available for C@. Furthermore, the
5934 unused expression (@var{exp1} or @var{exp2} depending on the value of
5935 @var{const_exp}) may still generate syntax errors. This may change in
5940 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5941 You can use the built-in function @code{__builtin_constant_p} to
5942 determine if a value is known to be constant at compile-time and hence
5943 that GCC can perform constant-folding on expressions involving that
5944 value. The argument of the function is the value to test. The function
5945 returns the integer 1 if the argument is known to be a compile-time
5946 constant and 0 if it is not known to be a compile-time constant. A
5947 return of 0 does not indicate that the value is @emph{not} a constant,
5948 but merely that GCC cannot prove it is a constant with the specified
5949 value of the @option{-O} option.
5951 You would typically use this function in an embedded application where
5952 memory was a critical resource. If you have some complex calculation,
5953 you may want it to be folded if it involves constants, but need to call
5954 a function if it does not. For example:
5957 #define Scale_Value(X) \
5958 (__builtin_constant_p (X) \
5959 ? ((X) * SCALE + OFFSET) : Scale (X))
5962 You may use this built-in function in either a macro or an inline
5963 function. However, if you use it in an inlined function and pass an
5964 argument of the function as the argument to the built-in, GCC will
5965 never return 1 when you call the inline function with a string constant
5966 or compound literal (@pxref{Compound Literals}) and will not return 1
5967 when you pass a constant numeric value to the inline function unless you
5968 specify the @option{-O} option.
5970 You may also use @code{__builtin_constant_p} in initializers for static
5971 data. For instance, you can write
5974 static const int table[] = @{
5975 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5981 This is an acceptable initializer even if @var{EXPRESSION} is not a
5982 constant expression. GCC must be more conservative about evaluating the
5983 built-in in this case, because it has no opportunity to perform
5986 Previous versions of GCC did not accept this built-in in data
5987 initializers. The earliest version where it is completely safe is
5991 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5992 @opindex fprofile-arcs
5993 You may use @code{__builtin_expect} to provide the compiler with
5994 branch prediction information. In general, you should prefer to
5995 use actual profile feedback for this (@option{-fprofile-arcs}), as
5996 programmers are notoriously bad at predicting how their programs
5997 actually perform. However, there are applications in which this
5998 data is hard to collect.
6000 The return value is the value of @var{exp}, which should be an
6001 integral expression. The value of @var{c} must be a compile-time
6002 constant. The semantics of the built-in are that it is expected
6003 that @var{exp} == @var{c}. For example:
6006 if (__builtin_expect (x, 0))
6011 would indicate that we do not expect to call @code{foo}, since
6012 we expect @code{x} to be zero. Since you are limited to integral
6013 expressions for @var{exp}, you should use constructions such as
6016 if (__builtin_expect (ptr != NULL, 1))
6021 when testing pointer or floating-point values.
6024 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
6025 This function is used to minimize cache-miss latency by moving data into
6026 a cache before it is accessed.
6027 You can insert calls to @code{__builtin_prefetch} into code for which
6028 you know addresses of data in memory that is likely to be accessed soon.
6029 If the target supports them, data prefetch instructions will be generated.
6030 If the prefetch is done early enough before the access then the data will
6031 be in the cache by the time it is accessed.
6033 The value of @var{addr} is the address of the memory to prefetch.
6034 There are two optional arguments, @var{rw} and @var{locality}.
6035 The value of @var{rw} is a compile-time constant one or zero; one
6036 means that the prefetch is preparing for a write to the memory address
6037 and zero, the default, means that the prefetch is preparing for a read.
6038 The value @var{locality} must be a compile-time constant integer between
6039 zero and three. A value of zero means that the data has no temporal
6040 locality, so it need not be left in the cache after the access. A value
6041 of three means that the data has a high degree of temporal locality and
6042 should be left in all levels of cache possible. Values of one and two
6043 mean, respectively, a low or moderate degree of temporal locality. The
6047 for (i = 0; i < n; i++)
6050 __builtin_prefetch (&a[i+j], 1, 1);
6051 __builtin_prefetch (&b[i+j], 0, 1);
6056 Data prefetch does not generate faults if @var{addr} is invalid, but
6057 the address expression itself must be valid. For example, a prefetch
6058 of @code{p->next} will not fault if @code{p->next} is not a valid
6059 address, but evaluation will fault if @code{p} is not a valid address.
6061 If the target does not support data prefetch, the address expression
6062 is evaluated if it includes side effects but no other code is generated
6063 and GCC does not issue a warning.
6066 @deftypefn {Built-in Function} double __builtin_huge_val (void)
6067 Returns a positive infinity, if supported by the floating-point format,
6068 else @code{DBL_MAX}. This function is suitable for implementing the
6069 ISO C macro @code{HUGE_VAL}.
6072 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
6073 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
6076 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
6077 Similar to @code{__builtin_huge_val}, except the return
6078 type is @code{long double}.
6081 @deftypefn {Built-in Function} double __builtin_inf (void)
6082 Similar to @code{__builtin_huge_val}, except a warning is generated
6083 if the target floating-point format does not support infinities.
6086 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
6087 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
6090 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
6091 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
6094 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
6095 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
6098 @deftypefn {Built-in Function} float __builtin_inff (void)
6099 Similar to @code{__builtin_inf}, except the return type is @code{float}.
6100 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
6103 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
6104 Similar to @code{__builtin_inf}, except the return
6105 type is @code{long double}.
6108 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
6109 This is an implementation of the ISO C99 function @code{nan}.
6111 Since ISO C99 defines this function in terms of @code{strtod}, which we
6112 do not implement, a description of the parsing is in order. The string
6113 is parsed as by @code{strtol}; that is, the base is recognized by
6114 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
6115 in the significand such that the least significant bit of the number
6116 is at the least significant bit of the significand. The number is
6117 truncated to fit the significand field provided. The significand is
6118 forced to be a quiet NaN@.
6120 This function, if given a string literal all of which would have been
6121 consumed by strtol, is evaluated early enough that it is considered a
6122 compile-time constant.
6125 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
6126 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
6129 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
6130 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
6133 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
6134 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
6137 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
6138 Similar to @code{__builtin_nan}, except the return type is @code{float}.
6141 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
6142 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
6145 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
6146 Similar to @code{__builtin_nan}, except the significand is forced
6147 to be a signaling NaN@. The @code{nans} function is proposed by
6148 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
6151 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
6152 Similar to @code{__builtin_nans}, except the return type is @code{float}.
6155 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
6156 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
6159 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
6160 Returns one plus the index of the least significant 1-bit of @var{x}, or
6161 if @var{x} is zero, returns zero.
6164 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
6165 Returns the number of leading 0-bits in @var{x}, starting at the most
6166 significant bit position. If @var{x} is 0, the result is undefined.
6169 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
6170 Returns the number of trailing 0-bits in @var{x}, starting at the least
6171 significant bit position. If @var{x} is 0, the result is undefined.
6174 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
6175 Returns the number of 1-bits in @var{x}.
6178 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
6179 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
6183 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
6184 Similar to @code{__builtin_ffs}, except the argument type is
6185 @code{unsigned long}.
6188 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
6189 Similar to @code{__builtin_clz}, except the argument type is
6190 @code{unsigned long}.
6193 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
6194 Similar to @code{__builtin_ctz}, except the argument type is
6195 @code{unsigned long}.
6198 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
6199 Similar to @code{__builtin_popcount}, except the argument type is
6200 @code{unsigned long}.
6203 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
6204 Similar to @code{__builtin_parity}, except the argument type is
6205 @code{unsigned long}.
6208 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
6209 Similar to @code{__builtin_ffs}, except the argument type is
6210 @code{unsigned long long}.
6213 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
6214 Similar to @code{__builtin_clz}, except the argument type is
6215 @code{unsigned long long}.
6218 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
6219 Similar to @code{__builtin_ctz}, except the argument type is
6220 @code{unsigned long long}.
6223 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
6224 Similar to @code{__builtin_popcount}, except the argument type is
6225 @code{unsigned long long}.
6228 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
6229 Similar to @code{__builtin_parity}, except the argument type is
6230 @code{unsigned long long}.
6233 @deftypefn {Built-in Function} double __builtin_powi (double, int)
6234 Returns the first argument raised to the power of the second. Unlike the
6235 @code{pow} function no guarantees about precision and rounding are made.
6238 @deftypefn {Built-in Function} float __builtin_powif (float, int)
6239 Similar to @code{__builtin_powi}, except the argument and return types
6243 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
6244 Similar to @code{__builtin_powi}, except the argument and return types
6245 are @code{long double}.
6248 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
6249 Returns @var{x} with the order of the bytes reversed; for example,
6250 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
6254 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
6255 Similar to @code{__builtin_bswap32}, except the argument and return types
6259 @node Target Builtins
6260 @section Built-in Functions Specific to Particular Target Machines
6262 On some target machines, GCC supports many built-in functions specific
6263 to those machines. Generally these generate calls to specific machine
6264 instructions, but allow the compiler to schedule those calls.
6267 * Alpha Built-in Functions::
6268 * ARM Built-in Functions::
6269 * Blackfin Built-in Functions::
6270 * FR-V Built-in Functions::
6271 * X86 Built-in Functions::
6272 * MIPS DSP Built-in Functions::
6273 * MIPS Paired-Single Support::
6274 * PowerPC AltiVec Built-in Functions::
6275 * SPARC VIS Built-in Functions::
6278 @node Alpha Built-in Functions
6279 @subsection Alpha Built-in Functions
6281 These built-in functions are available for the Alpha family of
6282 processors, depending on the command-line switches used.
6284 The following built-in functions are always available. They
6285 all generate the machine instruction that is part of the name.
6288 long __builtin_alpha_implver (void)
6289 long __builtin_alpha_rpcc (void)
6290 long __builtin_alpha_amask (long)
6291 long __builtin_alpha_cmpbge (long, long)
6292 long __builtin_alpha_extbl (long, long)
6293 long __builtin_alpha_extwl (long, long)
6294 long __builtin_alpha_extll (long, long)
6295 long __builtin_alpha_extql (long, long)
6296 long __builtin_alpha_extwh (long, long)
6297 long __builtin_alpha_extlh (long, long)
6298 long __builtin_alpha_extqh (long, long)
6299 long __builtin_alpha_insbl (long, long)
6300 long __builtin_alpha_inswl (long, long)
6301 long __builtin_alpha_insll (long, long)
6302 long __builtin_alpha_insql (long, long)
6303 long __builtin_alpha_inswh (long, long)
6304 long __builtin_alpha_inslh (long, long)
6305 long __builtin_alpha_insqh (long, long)
6306 long __builtin_alpha_mskbl (long, long)
6307 long __builtin_alpha_mskwl (long, long)
6308 long __builtin_alpha_mskll (long, long)
6309 long __builtin_alpha_mskql (long, long)
6310 long __builtin_alpha_mskwh (long, long)
6311 long __builtin_alpha_msklh (long, long)
6312 long __builtin_alpha_mskqh (long, long)
6313 long __builtin_alpha_umulh (long, long)
6314 long __builtin_alpha_zap (long, long)
6315 long __builtin_alpha_zapnot (long, long)
6318 The following built-in functions are always with @option{-mmax}
6319 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
6320 later. They all generate the machine instruction that is part
6324 long __builtin_alpha_pklb (long)
6325 long __builtin_alpha_pkwb (long)
6326 long __builtin_alpha_unpkbl (long)
6327 long __builtin_alpha_unpkbw (long)
6328 long __builtin_alpha_minub8 (long, long)
6329 long __builtin_alpha_minsb8 (long, long)
6330 long __builtin_alpha_minuw4 (long, long)
6331 long __builtin_alpha_minsw4 (long, long)
6332 long __builtin_alpha_maxub8 (long, long)
6333 long __builtin_alpha_maxsb8 (long, long)
6334 long __builtin_alpha_maxuw4 (long, long)
6335 long __builtin_alpha_maxsw4 (long, long)
6336 long __builtin_alpha_perr (long, long)
6339 The following built-in functions are always with @option{-mcix}
6340 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
6341 later. They all generate the machine instruction that is part
6345 long __builtin_alpha_cttz (long)
6346 long __builtin_alpha_ctlz (long)
6347 long __builtin_alpha_ctpop (long)
6350 The following builtins are available on systems that use the OSF/1
6351 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
6352 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
6353 @code{rdval} and @code{wrval}.
6356 void *__builtin_thread_pointer (void)
6357 void __builtin_set_thread_pointer (void *)
6360 @node ARM Built-in Functions
6361 @subsection ARM Built-in Functions
6363 These built-in functions are available for the ARM family of
6364 processors, when the @option{-mcpu=iwmmxt} switch is used:
6367 typedef int v2si __attribute__ ((vector_size (8)));
6368 typedef short v4hi __attribute__ ((vector_size (8)));
6369 typedef char v8qi __attribute__ ((vector_size (8)));
6371 int __builtin_arm_getwcx (int)
6372 void __builtin_arm_setwcx (int, int)
6373 int __builtin_arm_textrmsb (v8qi, int)
6374 int __builtin_arm_textrmsh (v4hi, int)
6375 int __builtin_arm_textrmsw (v2si, int)
6376 int __builtin_arm_textrmub (v8qi, int)
6377 int __builtin_arm_textrmuh (v4hi, int)
6378 int __builtin_arm_textrmuw (v2si, int)
6379 v8qi __builtin_arm_tinsrb (v8qi, int)
6380 v4hi __builtin_arm_tinsrh (v4hi, int)
6381 v2si __builtin_arm_tinsrw (v2si, int)
6382 long long __builtin_arm_tmia (long long, int, int)
6383 long long __builtin_arm_tmiabb (long long, int, int)
6384 long long __builtin_arm_tmiabt (long long, int, int)
6385 long long __builtin_arm_tmiaph (long long, int, int)
6386 long long __builtin_arm_tmiatb (long long, int, int)
6387 long long __builtin_arm_tmiatt (long long, int, int)
6388 int __builtin_arm_tmovmskb (v8qi)
6389 int __builtin_arm_tmovmskh (v4hi)
6390 int __builtin_arm_tmovmskw (v2si)
6391 long long __builtin_arm_waccb (v8qi)
6392 long long __builtin_arm_wacch (v4hi)
6393 long long __builtin_arm_waccw (v2si)
6394 v8qi __builtin_arm_waddb (v8qi, v8qi)
6395 v8qi __builtin_arm_waddbss (v8qi, v8qi)
6396 v8qi __builtin_arm_waddbus (v8qi, v8qi)
6397 v4hi __builtin_arm_waddh (v4hi, v4hi)
6398 v4hi __builtin_arm_waddhss (v4hi, v4hi)
6399 v4hi __builtin_arm_waddhus (v4hi, v4hi)
6400 v2si __builtin_arm_waddw (v2si, v2si)
6401 v2si __builtin_arm_waddwss (v2si, v2si)
6402 v2si __builtin_arm_waddwus (v2si, v2si)
6403 v8qi __builtin_arm_walign (v8qi, v8qi, int)
6404 long long __builtin_arm_wand(long long, long long)
6405 long long __builtin_arm_wandn (long long, long long)
6406 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
6407 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
6408 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
6409 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
6410 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
6411 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
6412 v2si __builtin_arm_wcmpeqw (v2si, v2si)
6413 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
6414 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
6415 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
6416 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
6417 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
6418 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
6419 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
6420 long long __builtin_arm_wmacsz (v4hi, v4hi)
6421 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
6422 long long __builtin_arm_wmacuz (v4hi, v4hi)
6423 v4hi __builtin_arm_wmadds (v4hi, v4hi)
6424 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
6425 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
6426 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
6427 v2si __builtin_arm_wmaxsw (v2si, v2si)
6428 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
6429 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
6430 v2si __builtin_arm_wmaxuw (v2si, v2si)
6431 v8qi __builtin_arm_wminsb (v8qi, v8qi)
6432 v4hi __builtin_arm_wminsh (v4hi, v4hi)
6433 v2si __builtin_arm_wminsw (v2si, v2si)
6434 v8qi __builtin_arm_wminub (v8qi, v8qi)
6435 v4hi __builtin_arm_wminuh (v4hi, v4hi)
6436 v2si __builtin_arm_wminuw (v2si, v2si)
6437 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
6438 v4hi __builtin_arm_wmulul (v4hi, v4hi)
6439 v4hi __builtin_arm_wmulum (v4hi, v4hi)
6440 long long __builtin_arm_wor (long long, long long)
6441 v2si __builtin_arm_wpackdss (long long, long long)
6442 v2si __builtin_arm_wpackdus (long long, long long)
6443 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
6444 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
6445 v4hi __builtin_arm_wpackwss (v2si, v2si)
6446 v4hi __builtin_arm_wpackwus (v2si, v2si)
6447 long long __builtin_arm_wrord (long long, long long)
6448 long long __builtin_arm_wrordi (long long, int)
6449 v4hi __builtin_arm_wrorh (v4hi, long long)
6450 v4hi __builtin_arm_wrorhi (v4hi, int)
6451 v2si __builtin_arm_wrorw (v2si, long long)
6452 v2si __builtin_arm_wrorwi (v2si, int)
6453 v2si __builtin_arm_wsadb (v8qi, v8qi)
6454 v2si __builtin_arm_wsadbz (v8qi, v8qi)
6455 v2si __builtin_arm_wsadh (v4hi, v4hi)
6456 v2si __builtin_arm_wsadhz (v4hi, v4hi)
6457 v4hi __builtin_arm_wshufh (v4hi, int)
6458 long long __builtin_arm_wslld (long long, long long)
6459 long long __builtin_arm_wslldi (long long, int)
6460 v4hi __builtin_arm_wsllh (v4hi, long long)
6461 v4hi __builtin_arm_wsllhi (v4hi, int)
6462 v2si __builtin_arm_wsllw (v2si, long long)
6463 v2si __builtin_arm_wsllwi (v2si, int)
6464 long long __builtin_arm_wsrad (long long, long long)
6465 long long __builtin_arm_wsradi (long long, int)
6466 v4hi __builtin_arm_wsrah (v4hi, long long)
6467 v4hi __builtin_arm_wsrahi (v4hi, int)
6468 v2si __builtin_arm_wsraw (v2si, long long)
6469 v2si __builtin_arm_wsrawi (v2si, int)
6470 long long __builtin_arm_wsrld (long long, long long)
6471 long long __builtin_arm_wsrldi (long long, int)
6472 v4hi __builtin_arm_wsrlh (v4hi, long long)
6473 v4hi __builtin_arm_wsrlhi (v4hi, int)
6474 v2si __builtin_arm_wsrlw (v2si, long long)
6475 v2si __builtin_arm_wsrlwi (v2si, int)
6476 v8qi __builtin_arm_wsubb (v8qi, v8qi)
6477 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
6478 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
6479 v4hi __builtin_arm_wsubh (v4hi, v4hi)
6480 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
6481 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
6482 v2si __builtin_arm_wsubw (v2si, v2si)
6483 v2si __builtin_arm_wsubwss (v2si, v2si)
6484 v2si __builtin_arm_wsubwus (v2si, v2si)
6485 v4hi __builtin_arm_wunpckehsb (v8qi)
6486 v2si __builtin_arm_wunpckehsh (v4hi)
6487 long long __builtin_arm_wunpckehsw (v2si)
6488 v4hi __builtin_arm_wunpckehub (v8qi)
6489 v2si __builtin_arm_wunpckehuh (v4hi)
6490 long long __builtin_arm_wunpckehuw (v2si)
6491 v4hi __builtin_arm_wunpckelsb (v8qi)
6492 v2si __builtin_arm_wunpckelsh (v4hi)
6493 long long __builtin_arm_wunpckelsw (v2si)
6494 v4hi __builtin_arm_wunpckelub (v8qi)
6495 v2si __builtin_arm_wunpckeluh (v4hi)
6496 long long __builtin_arm_wunpckeluw (v2si)
6497 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
6498 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
6499 v2si __builtin_arm_wunpckihw (v2si, v2si)
6500 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
6501 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
6502 v2si __builtin_arm_wunpckilw (v2si, v2si)
6503 long long __builtin_arm_wxor (long long, long long)
6504 long long __builtin_arm_wzero ()
6507 @node Blackfin Built-in Functions
6508 @subsection Blackfin Built-in Functions
6510 Currently, there are two Blackfin-specific built-in functions. These are
6511 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
6512 using inline assembly; by using these built-in functions the compiler can
6513 automatically add workarounds for hardware errata involving these
6514 instructions. These functions are named as follows:
6517 void __builtin_bfin_csync (void)
6518 void __builtin_bfin_ssync (void)
6521 @node FR-V Built-in Functions
6522 @subsection FR-V Built-in Functions
6524 GCC provides many FR-V-specific built-in functions. In general,
6525 these functions are intended to be compatible with those described
6526 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
6527 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
6528 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
6529 pointer rather than by value.
6531 Most of the functions are named after specific FR-V instructions.
6532 Such functions are said to be ``directly mapped'' and are summarized
6533 here in tabular form.
6537 * Directly-mapped Integer Functions::
6538 * Directly-mapped Media Functions::
6539 * Raw read/write Functions::
6540 * Other Built-in Functions::
6543 @node Argument Types
6544 @subsubsection Argument Types
6546 The arguments to the built-in functions can be divided into three groups:
6547 register numbers, compile-time constants and run-time values. In order
6548 to make this classification clear at a glance, the arguments and return
6549 values are given the following pseudo types:
6551 @multitable @columnfractions .20 .30 .15 .35
6552 @item Pseudo type @tab Real C type @tab Constant? @tab Description
6553 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
6554 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
6555 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
6556 @item @code{uw2} @tab @code{unsigned long long} @tab No
6557 @tab an unsigned doubleword
6558 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
6559 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
6560 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
6561 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
6564 These pseudo types are not defined by GCC, they are simply a notational
6565 convenience used in this manual.
6567 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
6568 and @code{sw2} are evaluated at run time. They correspond to
6569 register operands in the underlying FR-V instructions.
6571 @code{const} arguments represent immediate operands in the underlying
6572 FR-V instructions. They must be compile-time constants.
6574 @code{acc} arguments are evaluated at compile time and specify the number
6575 of an accumulator register. For example, an @code{acc} argument of 2
6576 will select the ACC2 register.
6578 @code{iacc} arguments are similar to @code{acc} arguments but specify the
6579 number of an IACC register. See @pxref{Other Built-in Functions}
6582 @node Directly-mapped Integer Functions
6583 @subsubsection Directly-mapped Integer Functions
6585 The functions listed below map directly to FR-V I-type instructions.
6587 @multitable @columnfractions .45 .32 .23
6588 @item Function prototype @tab Example usage @tab Assembly output
6589 @item @code{sw1 __ADDSS (sw1, sw1)}
6590 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
6591 @tab @code{ADDSS @var{a},@var{b},@var{c}}
6592 @item @code{sw1 __SCAN (sw1, sw1)}
6593 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
6594 @tab @code{SCAN @var{a},@var{b},@var{c}}
6595 @item @code{sw1 __SCUTSS (sw1)}
6596 @tab @code{@var{b} = __SCUTSS (@var{a})}
6597 @tab @code{SCUTSS @var{a},@var{b}}
6598 @item @code{sw1 __SLASS (sw1, sw1)}
6599 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
6600 @tab @code{SLASS @var{a},@var{b},@var{c}}
6601 @item @code{void __SMASS (sw1, sw1)}
6602 @tab @code{__SMASS (@var{a}, @var{b})}
6603 @tab @code{SMASS @var{a},@var{b}}
6604 @item @code{void __SMSSS (sw1, sw1)}
6605 @tab @code{__SMSSS (@var{a}, @var{b})}
6606 @tab @code{SMSSS @var{a},@var{b}}
6607 @item @code{void __SMU (sw1, sw1)}
6608 @tab @code{__SMU (@var{a}, @var{b})}
6609 @tab @code{SMU @var{a},@var{b}}
6610 @item @code{sw2 __SMUL (sw1, sw1)}
6611 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
6612 @tab @code{SMUL @var{a},@var{b},@var{c}}
6613 @item @code{sw1 __SUBSS (sw1, sw1)}
6614 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
6615 @tab @code{SUBSS @var{a},@var{b},@var{c}}
6616 @item @code{uw2 __UMUL (uw1, uw1)}
6617 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
6618 @tab @code{UMUL @var{a},@var{b},@var{c}}
6621 @node Directly-mapped Media Functions
6622 @subsubsection Directly-mapped Media Functions
6624 The functions listed below map directly to FR-V M-type instructions.
6626 @multitable @columnfractions .45 .32 .23
6627 @item Function prototype @tab Example usage @tab Assembly output
6628 @item @code{uw1 __MABSHS (sw1)}
6629 @tab @code{@var{b} = __MABSHS (@var{a})}
6630 @tab @code{MABSHS @var{a},@var{b}}
6631 @item @code{void __MADDACCS (acc, acc)}
6632 @tab @code{__MADDACCS (@var{b}, @var{a})}
6633 @tab @code{MADDACCS @var{a},@var{b}}
6634 @item @code{sw1 __MADDHSS (sw1, sw1)}
6635 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
6636 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
6637 @item @code{uw1 __MADDHUS (uw1, uw1)}
6638 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
6639 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
6640 @item @code{uw1 __MAND (uw1, uw1)}
6641 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
6642 @tab @code{MAND @var{a},@var{b},@var{c}}
6643 @item @code{void __MASACCS (acc, acc)}
6644 @tab @code{__MASACCS (@var{b}, @var{a})}
6645 @tab @code{MASACCS @var{a},@var{b}}
6646 @item @code{uw1 __MAVEH (uw1, uw1)}
6647 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
6648 @tab @code{MAVEH @var{a},@var{b},@var{c}}
6649 @item @code{uw2 __MBTOH (uw1)}
6650 @tab @code{@var{b} = __MBTOH (@var{a})}
6651 @tab @code{MBTOH @var{a},@var{b}}
6652 @item @code{void __MBTOHE (uw1 *, uw1)}
6653 @tab @code{__MBTOHE (&@var{b}, @var{a})}
6654 @tab @code{MBTOHE @var{a},@var{b}}
6655 @item @code{void __MCLRACC (acc)}
6656 @tab @code{__MCLRACC (@var{a})}
6657 @tab @code{MCLRACC @var{a}}
6658 @item @code{void __MCLRACCA (void)}
6659 @tab @code{__MCLRACCA ()}
6660 @tab @code{MCLRACCA}
6661 @item @code{uw1 __Mcop1 (uw1, uw1)}
6662 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
6663 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
6664 @item @code{uw1 __Mcop2 (uw1, uw1)}
6665 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
6666 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
6667 @item @code{uw1 __MCPLHI (uw2, const)}
6668 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
6669 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
6670 @item @code{uw1 __MCPLI (uw2, const)}
6671 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
6672 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
6673 @item @code{void __MCPXIS (acc, sw1, sw1)}
6674 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
6675 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
6676 @item @code{void __MCPXIU (acc, uw1, uw1)}
6677 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
6678 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
6679 @item @code{void __MCPXRS (acc, sw1, sw1)}
6680 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
6681 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
6682 @item @code{void __MCPXRU (acc, uw1, uw1)}
6683 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
6684 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
6685 @item @code{uw1 __MCUT (acc, uw1)}
6686 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
6687 @tab @code{MCUT @var{a},@var{b},@var{c}}
6688 @item @code{uw1 __MCUTSS (acc, sw1)}
6689 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
6690 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
6691 @item @code{void __MDADDACCS (acc, acc)}
6692 @tab @code{__MDADDACCS (@var{b}, @var{a})}
6693 @tab @code{MDADDACCS @var{a},@var{b}}
6694 @item @code{void __MDASACCS (acc, acc)}
6695 @tab @code{__MDASACCS (@var{b}, @var{a})}
6696 @tab @code{MDASACCS @var{a},@var{b}}
6697 @item @code{uw2 __MDCUTSSI (acc, const)}
6698 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
6699 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
6700 @item @code{uw2 __MDPACKH (uw2, uw2)}
6701 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
6702 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
6703 @item @code{uw2 __MDROTLI (uw2, const)}
6704 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
6705 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
6706 @item @code{void __MDSUBACCS (acc, acc)}
6707 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
6708 @tab @code{MDSUBACCS @var{a},@var{b}}
6709 @item @code{void __MDUNPACKH (uw1 *, uw2)}
6710 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
6711 @tab @code{MDUNPACKH @var{a},@var{b}}
6712 @item @code{uw2 __MEXPDHD (uw1, const)}
6713 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
6714 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
6715 @item @code{uw1 __MEXPDHW (uw1, const)}
6716 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
6717 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
6718 @item @code{uw1 __MHDSETH (uw1, const)}
6719 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
6720 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
6721 @item @code{sw1 __MHDSETS (const)}
6722 @tab @code{@var{b} = __MHDSETS (@var{a})}
6723 @tab @code{MHDSETS #@var{a},@var{b}}
6724 @item @code{uw1 __MHSETHIH (uw1, const)}
6725 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
6726 @tab @code{MHSETHIH #@var{a},@var{b}}
6727 @item @code{sw1 __MHSETHIS (sw1, const)}
6728 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
6729 @tab @code{MHSETHIS #@var{a},@var{b}}
6730 @item @code{uw1 __MHSETLOH (uw1, const)}
6731 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
6732 @tab @code{MHSETLOH #@var{a},@var{b}}
6733 @item @code{sw1 __MHSETLOS (sw1, const)}
6734 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
6735 @tab @code{MHSETLOS #@var{a},@var{b}}
6736 @item @code{uw1 __MHTOB (uw2)}
6737 @tab @code{@var{b} = __MHTOB (@var{a})}
6738 @tab @code{MHTOB @var{a},@var{b}}
6739 @item @code{void __MMACHS (acc, sw1, sw1)}
6740 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
6741 @tab @code{MMACHS @var{a},@var{b},@var{c}}
6742 @item @code{void __MMACHU (acc, uw1, uw1)}
6743 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
6744 @tab @code{MMACHU @var{a},@var{b},@var{c}}
6745 @item @code{void __MMRDHS (acc, sw1, sw1)}
6746 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
6747 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
6748 @item @code{void __MMRDHU (acc, uw1, uw1)}
6749 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
6750 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
6751 @item @code{void __MMULHS (acc, sw1, sw1)}
6752 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
6753 @tab @code{MMULHS @var{a},@var{b},@var{c}}
6754 @item @code{void __MMULHU (acc, uw1, uw1)}
6755 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
6756 @tab @code{MMULHU @var{a},@var{b},@var{c}}
6757 @item @code{void __MMULXHS (acc, sw1, sw1)}
6758 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
6759 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
6760 @item @code{void __MMULXHU (acc, uw1, uw1)}
6761 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
6762 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
6763 @item @code{uw1 __MNOT (uw1)}
6764 @tab @code{@var{b} = __MNOT (@var{a})}
6765 @tab @code{MNOT @var{a},@var{b}}
6766 @item @code{uw1 __MOR (uw1, uw1)}
6767 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
6768 @tab @code{MOR @var{a},@var{b},@var{c}}
6769 @item @code{uw1 __MPACKH (uh, uh)}
6770 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
6771 @tab @code{MPACKH @var{a},@var{b},@var{c}}
6772 @item @code{sw2 __MQADDHSS (sw2, sw2)}
6773 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
6774 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
6775 @item @code{uw2 __MQADDHUS (uw2, uw2)}
6776 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
6777 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
6778 @item @code{void __MQCPXIS (acc, sw2, sw2)}
6779 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
6780 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
6781 @item @code{void __MQCPXIU (acc, uw2, uw2)}
6782 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
6783 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
6784 @item @code{void __MQCPXRS (acc, sw2, sw2)}
6785 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
6786 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
6787 @item @code{void __MQCPXRU (acc, uw2, uw2)}
6788 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
6789 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
6790 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
6791 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
6792 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
6793 @item @code{sw2 __MQLMTHS (sw2, sw2)}
6794 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
6795 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
6796 @item @code{void __MQMACHS (acc, sw2, sw2)}
6797 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
6798 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
6799 @item @code{void __MQMACHU (acc, uw2, uw2)}
6800 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
6801 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
6802 @item @code{void __MQMACXHS (acc, sw2, sw2)}
6803 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
6804 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
6805 @item @code{void __MQMULHS (acc, sw2, sw2)}
6806 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
6807 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
6808 @item @code{void __MQMULHU (acc, uw2, uw2)}
6809 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
6810 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
6811 @item @code{void __MQMULXHS (acc, sw2, sw2)}
6812 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
6813 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
6814 @item @code{void __MQMULXHU (acc, uw2, uw2)}
6815 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
6816 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
6817 @item @code{sw2 __MQSATHS (sw2, sw2)}
6818 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
6819 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
6820 @item @code{uw2 __MQSLLHI (uw2, int)}
6821 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
6822 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
6823 @item @code{sw2 __MQSRAHI (sw2, int)}
6824 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
6825 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
6826 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
6827 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
6828 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
6829 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
6830 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
6831 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
6832 @item @code{void __MQXMACHS (acc, sw2, sw2)}
6833 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
6834 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
6835 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
6836 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
6837 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
6838 @item @code{uw1 __MRDACC (acc)}
6839 @tab @code{@var{b} = __MRDACC (@var{a})}
6840 @tab @code{MRDACC @var{a},@var{b}}
6841 @item @code{uw1 __MRDACCG (acc)}
6842 @tab @code{@var{b} = __MRDACCG (@var{a})}
6843 @tab @code{MRDACCG @var{a},@var{b}}
6844 @item @code{uw1 __MROTLI (uw1, const)}
6845 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
6846 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
6847 @item @code{uw1 __MROTRI (uw1, const)}
6848 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
6849 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
6850 @item @code{sw1 __MSATHS (sw1, sw1)}
6851 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
6852 @tab @code{MSATHS @var{a},@var{b},@var{c}}
6853 @item @code{uw1 __MSATHU (uw1, uw1)}
6854 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
6855 @tab @code{MSATHU @var{a},@var{b},@var{c}}
6856 @item @code{uw1 __MSLLHI (uw1, const)}
6857 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
6858 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
6859 @item @code{sw1 __MSRAHI (sw1, const)}
6860 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
6861 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
6862 @item @code{uw1 __MSRLHI (uw1, const)}
6863 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
6864 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
6865 @item @code{void __MSUBACCS (acc, acc)}
6866 @tab @code{__MSUBACCS (@var{b}, @var{a})}
6867 @tab @code{MSUBACCS @var{a},@var{b}}
6868 @item @code{sw1 __MSUBHSS (sw1, sw1)}
6869 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
6870 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
6871 @item @code{uw1 __MSUBHUS (uw1, uw1)}
6872 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
6873 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
6874 @item @code{void __MTRAP (void)}
6875 @tab @code{__MTRAP ()}
6877 @item @code{uw2 __MUNPACKH (uw1)}
6878 @tab @code{@var{b} = __MUNPACKH (@var{a})}
6879 @tab @code{MUNPACKH @var{a},@var{b}}
6880 @item @code{uw1 __MWCUT (uw2, uw1)}
6881 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
6882 @tab @code{MWCUT @var{a},@var{b},@var{c}}
6883 @item @code{void __MWTACC (acc, uw1)}
6884 @tab @code{__MWTACC (@var{b}, @var{a})}
6885 @tab @code{MWTACC @var{a},@var{b}}
6886 @item @code{void __MWTACCG (acc, uw1)}
6887 @tab @code{__MWTACCG (@var{b}, @var{a})}
6888 @tab @code{MWTACCG @var{a},@var{b}}
6889 @item @code{uw1 __MXOR (uw1, uw1)}
6890 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
6891 @tab @code{MXOR @var{a},@var{b},@var{c}}
6894 @node Raw read/write Functions
6895 @subsubsection Raw read/write Functions
6897 This sections describes built-in functions related to read and write
6898 instructions to access memory. These functions generate
6899 @code{membar} instructions to flush the I/O load and stores where
6900 appropriate, as described in Fujitsu's manual described above.
6904 @item unsigned char __builtin_read8 (void *@var{data})
6905 @item unsigned short __builtin_read16 (void *@var{data})
6906 @item unsigned long __builtin_read32 (void *@var{data})
6907 @item unsigned long long __builtin_read64 (void *@var{data})
6909 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
6910 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
6911 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
6912 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
6915 @node Other Built-in Functions
6916 @subsubsection Other Built-in Functions
6918 This section describes built-in functions that are not named after
6919 a specific FR-V instruction.
6922 @item sw2 __IACCreadll (iacc @var{reg})
6923 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
6924 for future expansion and must be 0.
6926 @item sw1 __IACCreadl (iacc @var{reg})
6927 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
6928 Other values of @var{reg} are rejected as invalid.
6930 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
6931 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
6932 is reserved for future expansion and must be 0.
6934 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
6935 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
6936 is 1. Other values of @var{reg} are rejected as invalid.
6938 @item void __data_prefetch0 (const void *@var{x})
6939 Use the @code{dcpl} instruction to load the contents of address @var{x}
6940 into the data cache.
6942 @item void __data_prefetch (const void *@var{x})
6943 Use the @code{nldub} instruction to load the contents of address @var{x}
6944 into the data cache. The instruction will be issued in slot I1@.
6947 @node X86 Built-in Functions
6948 @subsection X86 Built-in Functions
6950 These built-in functions are available for the i386 and x86-64 family
6951 of computers, depending on the command-line switches used.
6953 Note that, if you specify command-line switches such as @option{-msse},
6954 the compiler could use the extended instruction sets even if the built-ins
6955 are not used explicitly in the program. For this reason, applications
6956 which perform runtime CPU detection must compile separate files for each
6957 supported architecture, using the appropriate flags. In particular,
6958 the file containing the CPU detection code should be compiled without
6961 The following machine modes are available for use with MMX built-in functions
6962 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
6963 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
6964 vector of eight 8-bit integers. Some of the built-in functions operate on
6965 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
6967 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
6968 of two 32-bit floating point values.
6970 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
6971 floating point values. Some instructions use a vector of four 32-bit
6972 integers, these use @code{V4SI}. Finally, some instructions operate on an
6973 entire vector register, interpreting it as a 128-bit integer, these use mode
6976 The following built-in functions are made available by @option{-mmmx}.
6977 All of them generate the machine instruction that is part of the name.
6980 v8qi __builtin_ia32_paddb (v8qi, v8qi)
6981 v4hi __builtin_ia32_paddw (v4hi, v4hi)
6982 v2si __builtin_ia32_paddd (v2si, v2si)
6983 v8qi __builtin_ia32_psubb (v8qi, v8qi)
6984 v4hi __builtin_ia32_psubw (v4hi, v4hi)
6985 v2si __builtin_ia32_psubd (v2si, v2si)
6986 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
6987 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
6988 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
6989 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
6990 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
6991 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
6992 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
6993 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
6994 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
6995 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
6996 di __builtin_ia32_pand (di, di)
6997 di __builtin_ia32_pandn (di,di)
6998 di __builtin_ia32_por (di, di)
6999 di __builtin_ia32_pxor (di, di)
7000 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7001 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7002 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7003 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7004 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7005 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7006 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7007 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7008 v2si __builtin_ia32_punpckhdq (v2si, v2si)
7009 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7010 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7011 v2si __builtin_ia32_punpckldq (v2si, v2si)
7012 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7013 v4hi __builtin_ia32_packssdw (v2si, v2si)
7014 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7017 The following built-in functions are made available either with
7018 @option{-msse}, or with a combination of @option{-m3dnow} and
7019 @option{-march=athlon}. All of them generate the machine
7020 instruction that is part of the name.
7023 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7024 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7025 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7026 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
7027 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7028 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7029 v8qi __builtin_ia32_pminub (v8qi, v8qi)
7030 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7031 int __builtin_ia32_pextrw (v4hi, int)
7032 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7033 int __builtin_ia32_pmovmskb (v8qi)
7034 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7035 void __builtin_ia32_movntq (di *, di)
7036 void __builtin_ia32_sfence (void)
7039 The following built-in functions are available when @option{-msse} is used.
7040 All of them generate the machine instruction that is part of the name.
7043 int __builtin_ia32_comieq (v4sf, v4sf)
7044 int __builtin_ia32_comineq (v4sf, v4sf)
7045 int __builtin_ia32_comilt (v4sf, v4sf)
7046 int __builtin_ia32_comile (v4sf, v4sf)
7047 int __builtin_ia32_comigt (v4sf, v4sf)
7048 int __builtin_ia32_comige (v4sf, v4sf)
7049 int __builtin_ia32_ucomieq (v4sf, v4sf)
7050 int __builtin_ia32_ucomineq (v4sf, v4sf)
7051 int __builtin_ia32_ucomilt (v4sf, v4sf)
7052 int __builtin_ia32_ucomile (v4sf, v4sf)
7053 int __builtin_ia32_ucomigt (v4sf, v4sf)
7054 int __builtin_ia32_ucomige (v4sf, v4sf)
7055 v4sf __builtin_ia32_addps (v4sf, v4sf)
7056 v4sf __builtin_ia32_subps (v4sf, v4sf)
7057 v4sf __builtin_ia32_mulps (v4sf, v4sf)
7058 v4sf __builtin_ia32_divps (v4sf, v4sf)
7059 v4sf __builtin_ia32_addss (v4sf, v4sf)
7060 v4sf __builtin_ia32_subss (v4sf, v4sf)
7061 v4sf __builtin_ia32_mulss (v4sf, v4sf)
7062 v4sf __builtin_ia32_divss (v4sf, v4sf)
7063 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
7064 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
7065 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
7066 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
7067 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
7068 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
7069 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
7070 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
7071 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
7072 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
7073 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
7074 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
7075 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
7076 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
7077 v4si __builtin_ia32_cmpless (v4sf, v4sf)
7078 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
7079 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
7080 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
7081 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
7082 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
7083 v4sf __builtin_ia32_maxps (v4sf, v4sf)
7084 v4sf __builtin_ia32_maxss (v4sf, v4sf)
7085 v4sf __builtin_ia32_minps (v4sf, v4sf)
7086 v4sf __builtin_ia32_minss (v4sf, v4sf)
7087 v4sf __builtin_ia32_andps (v4sf, v4sf)
7088 v4sf __builtin_ia32_andnps (v4sf, v4sf)
7089 v4sf __builtin_ia32_orps (v4sf, v4sf)
7090 v4sf __builtin_ia32_xorps (v4sf, v4sf)
7091 v4sf __builtin_ia32_movss (v4sf, v4sf)
7092 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
7093 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
7094 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
7095 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
7096 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
7097 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
7098 v2si __builtin_ia32_cvtps2pi (v4sf)
7099 int __builtin_ia32_cvtss2si (v4sf)
7100 v2si __builtin_ia32_cvttps2pi (v4sf)
7101 int __builtin_ia32_cvttss2si (v4sf)
7102 v4sf __builtin_ia32_rcpps (v4sf)
7103 v4sf __builtin_ia32_rsqrtps (v4sf)
7104 v4sf __builtin_ia32_sqrtps (v4sf)
7105 v4sf __builtin_ia32_rcpss (v4sf)
7106 v4sf __builtin_ia32_rsqrtss (v4sf)
7107 v4sf __builtin_ia32_sqrtss (v4sf)
7108 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
7109 void __builtin_ia32_movntps (float *, v4sf)
7110 int __builtin_ia32_movmskps (v4sf)
7113 The following built-in functions are available when @option{-msse} is used.
7116 @item v4sf __builtin_ia32_loadaps (float *)
7117 Generates the @code{movaps} machine instruction as a load from memory.
7118 @item void __builtin_ia32_storeaps (float *, v4sf)
7119 Generates the @code{movaps} machine instruction as a store to memory.
7120 @item v4sf __builtin_ia32_loadups (float *)
7121 Generates the @code{movups} machine instruction as a load from memory.
7122 @item void __builtin_ia32_storeups (float *, v4sf)
7123 Generates the @code{movups} machine instruction as a store to memory.
7124 @item v4sf __builtin_ia32_loadsss (float *)
7125 Generates the @code{movss} machine instruction as a load from memory.
7126 @item void __builtin_ia32_storess (float *, v4sf)
7127 Generates the @code{movss} machine instruction as a store to memory.
7128 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
7129 Generates the @code{movhps} machine instruction as a load from memory.
7130 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
7131 Generates the @code{movlps} machine instruction as a load from memory
7132 @item void __builtin_ia32_storehps (v4sf, v2si *)
7133 Generates the @code{movhps} machine instruction as a store to memory.
7134 @item void __builtin_ia32_storelps (v4sf, v2si *)
7135 Generates the @code{movlps} machine instruction as a store to memory.
7138 The following built-in functions are available when @option{-msse2} is used.
7139 All of them generate the machine instruction that is part of the name.
7142 int __builtin_ia32_comisdeq (v2df, v2df)
7143 int __builtin_ia32_comisdlt (v2df, v2df)
7144 int __builtin_ia32_comisdle (v2df, v2df)
7145 int __builtin_ia32_comisdgt (v2df, v2df)
7146 int __builtin_ia32_comisdge (v2df, v2df)
7147 int __builtin_ia32_comisdneq (v2df, v2df)
7148 int __builtin_ia32_ucomisdeq (v2df, v2df)
7149 int __builtin_ia32_ucomisdlt (v2df, v2df)
7150 int __builtin_ia32_ucomisdle (v2df, v2df)
7151 int __builtin_ia32_ucomisdgt (v2df, v2df)
7152 int __builtin_ia32_ucomisdge (v2df, v2df)
7153 int __builtin_ia32_ucomisdneq (v2df, v2df)
7154 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
7155 v2df __builtin_ia32_cmpltpd (v2df, v2df)
7156 v2df __builtin_ia32_cmplepd (v2df, v2df)
7157 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
7158 v2df __builtin_ia32_cmpgepd (v2df, v2df)
7159 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
7160 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
7161 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
7162 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
7163 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
7164 v2df __builtin_ia32_cmpngepd (v2df, v2df)
7165 v2df __builtin_ia32_cmpordpd (v2df, v2df)
7166 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
7167 v2df __builtin_ia32_cmpltsd (v2df, v2df)
7168 v2df __builtin_ia32_cmplesd (v2df, v2df)
7169 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
7170 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
7171 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
7172 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
7173 v2df __builtin_ia32_cmpordsd (v2df, v2df)
7174 v2di __builtin_ia32_paddq (v2di, v2di)
7175 v2di __builtin_ia32_psubq (v2di, v2di)
7176 v2df __builtin_ia32_addpd (v2df, v2df)
7177 v2df __builtin_ia32_subpd (v2df, v2df)
7178 v2df __builtin_ia32_mulpd (v2df, v2df)
7179 v2df __builtin_ia32_divpd (v2df, v2df)
7180 v2df __builtin_ia32_addsd (v2df, v2df)
7181 v2df __builtin_ia32_subsd (v2df, v2df)
7182 v2df __builtin_ia32_mulsd (v2df, v2df)
7183 v2df __builtin_ia32_divsd (v2df, v2df)
7184 v2df __builtin_ia32_minpd (v2df, v2df)
7185 v2df __builtin_ia32_maxpd (v2df, v2df)
7186 v2df __builtin_ia32_minsd (v2df, v2df)
7187 v2df __builtin_ia32_maxsd (v2df, v2df)
7188 v2df __builtin_ia32_andpd (v2df, v2df)
7189 v2df __builtin_ia32_andnpd (v2df, v2df)
7190 v2df __builtin_ia32_orpd (v2df, v2df)
7191 v2df __builtin_ia32_xorpd (v2df, v2df)
7192 v2df __builtin_ia32_movsd (v2df, v2df)
7193 v2df __builtin_ia32_unpckhpd (v2df, v2df)
7194 v2df __builtin_ia32_unpcklpd (v2df, v2df)
7195 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
7196 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
7197 v4si __builtin_ia32_paddd128 (v4si, v4si)
7198 v2di __builtin_ia32_paddq128 (v2di, v2di)
7199 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
7200 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
7201 v4si __builtin_ia32_psubd128 (v4si, v4si)
7202 v2di __builtin_ia32_psubq128 (v2di, v2di)
7203 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
7204 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
7205 v2di __builtin_ia32_pand128 (v2di, v2di)
7206 v2di __builtin_ia32_pandn128 (v2di, v2di)
7207 v2di __builtin_ia32_por128 (v2di, v2di)
7208 v2di __builtin_ia32_pxor128 (v2di, v2di)
7209 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
7210 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
7211 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
7212 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
7213 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
7214 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
7215 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
7216 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
7217 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
7218 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
7219 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
7220 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
7221 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
7222 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
7223 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
7224 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
7225 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
7226 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
7227 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
7228 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
7229 v16qi __builtin_ia32_packsswb128 (v16qi, v16qi)
7230 v8hi __builtin_ia32_packssdw128 (v8hi, v8hi)
7231 v16qi __builtin_ia32_packuswb128 (v16qi, v16qi)
7232 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
7233 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
7234 v2df __builtin_ia32_loadupd (double *)
7235 void __builtin_ia32_storeupd (double *, v2df)
7236 v2df __builtin_ia32_loadhpd (v2df, double *)
7237 v2df __builtin_ia32_loadlpd (v2df, double *)
7238 int __builtin_ia32_movmskpd (v2df)
7239 int __builtin_ia32_pmovmskb128 (v16qi)
7240 void __builtin_ia32_movnti (int *, int)
7241 void __builtin_ia32_movntpd (double *, v2df)
7242 void __builtin_ia32_movntdq (v2df *, v2df)
7243 v4si __builtin_ia32_pshufd (v4si, int)
7244 v8hi __builtin_ia32_pshuflw (v8hi, int)
7245 v8hi __builtin_ia32_pshufhw (v8hi, int)
7246 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
7247 v2df __builtin_ia32_sqrtpd (v2df)
7248 v2df __builtin_ia32_sqrtsd (v2df)
7249 v2df __builtin_ia32_shufpd (v2df, v2df, int)
7250 v2df __builtin_ia32_cvtdq2pd (v4si)
7251 v4sf __builtin_ia32_cvtdq2ps (v4si)
7252 v4si __builtin_ia32_cvtpd2dq (v2df)
7253 v2si __builtin_ia32_cvtpd2pi (v2df)
7254 v4sf __builtin_ia32_cvtpd2ps (v2df)
7255 v4si __builtin_ia32_cvttpd2dq (v2df)
7256 v2si __builtin_ia32_cvttpd2pi (v2df)
7257 v2df __builtin_ia32_cvtpi2pd (v2si)
7258 int __builtin_ia32_cvtsd2si (v2df)
7259 int __builtin_ia32_cvttsd2si (v2df)
7260 long long __builtin_ia32_cvtsd2si64 (v2df)
7261 long long __builtin_ia32_cvttsd2si64 (v2df)
7262 v4si __builtin_ia32_cvtps2dq (v4sf)
7263 v2df __builtin_ia32_cvtps2pd (v4sf)
7264 v4si __builtin_ia32_cvttps2dq (v4sf)
7265 v2df __builtin_ia32_cvtsi2sd (v2df, int)
7266 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
7267 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
7268 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
7269 void __builtin_ia32_clflush (const void *)
7270 void __builtin_ia32_lfence (void)
7271 void __builtin_ia32_mfence (void)
7272 v16qi __builtin_ia32_loaddqu (const char *)
7273 void __builtin_ia32_storedqu (char *, v16qi)
7274 unsigned long long __builtin_ia32_pmuludq (v2si, v2si)
7275 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
7276 v8hi __builtin_ia32_psllw128 (v8hi, v2di)
7277 v4si __builtin_ia32_pslld128 (v4si, v2di)
7278 v2di __builtin_ia32_psllq128 (v4si, v2di)
7279 v8hi __builtin_ia32_psrlw128 (v8hi, v2di)
7280 v4si __builtin_ia32_psrld128 (v4si, v2di)
7281 v2di __builtin_ia32_psrlq128 (v2di, v2di)
7282 v8hi __builtin_ia32_psraw128 (v8hi, v2di)
7283 v4si __builtin_ia32_psrad128 (v4si, v2di)
7284 v2di __builtin_ia32_pslldqi128 (v2di, int)
7285 v8hi __builtin_ia32_psllwi128 (v8hi, int)
7286 v4si __builtin_ia32_pslldi128 (v4si, int)
7287 v2di __builtin_ia32_psllqi128 (v2di, int)
7288 v2di __builtin_ia32_psrldqi128 (v2di, int)
7289 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
7290 v4si __builtin_ia32_psrldi128 (v4si, int)
7291 v2di __builtin_ia32_psrlqi128 (v2di, int)
7292 v8hi __builtin_ia32_psrawi128 (v8hi, int)
7293 v4si __builtin_ia32_psradi128 (v4si, int)
7294 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
7297 The following built-in functions are available when @option{-msse3} is used.
7298 All of them generate the machine instruction that is part of the name.
7301 v2df __builtin_ia32_addsubpd (v2df, v2df)
7302 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
7303 v2df __builtin_ia32_haddpd (v2df, v2df)
7304 v4sf __builtin_ia32_haddps (v4sf, v4sf)
7305 v2df __builtin_ia32_hsubpd (v2df, v2df)
7306 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
7307 v16qi __builtin_ia32_lddqu (char const *)
7308 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
7309 v2df __builtin_ia32_movddup (v2df)
7310 v4sf __builtin_ia32_movshdup (v4sf)
7311 v4sf __builtin_ia32_movsldup (v4sf)
7312 void __builtin_ia32_mwait (unsigned int, unsigned int)
7315 The following built-in functions are available when @option{-msse3} is used.
7318 @item v2df __builtin_ia32_loadddup (double const *)
7319 Generates the @code{movddup} machine instruction as a load from memory.
7322 The following built-in functions are available when @option{-mssse3} is used.
7323 All of them generate the machine instruction that is part of the name
7327 v2si __builtin_ia32_phaddd (v2si, v2si)
7328 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
7329 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
7330 v2si __builtin_ia32_phsubd (v2si, v2si)
7331 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
7332 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
7333 v8qi __builtin_ia32_pmaddubsw (v8qi, v8qi)
7334 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
7335 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
7336 v8qi __builtin_ia32_psignb (v8qi, v8qi)
7337 v2si __builtin_ia32_psignd (v2si, v2si)
7338 v4hi __builtin_ia32_psignw (v4hi, v4hi)
7339 long long __builtin_ia32_palignr (long long, long long, int)
7340 v8qi __builtin_ia32_pabsb (v8qi)
7341 v2si __builtin_ia32_pabsd (v2si)
7342 v4hi __builtin_ia32_pabsw (v4hi)
7345 The following built-in functions are available when @option{-mssse3} is used.
7346 All of them generate the machine instruction that is part of the name
7350 v4si __builtin_ia32_phaddd128 (v4si, v4si)
7351 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
7352 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
7353 v4si __builtin_ia32_phsubd128 (v4si, v4si)
7354 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
7355 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
7356 v16qi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
7357 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
7358 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
7359 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
7360 v4si __builtin_ia32_psignd128 (v4si, v4si)
7361 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
7362 v2di __builtin_ia32_palignr (v2di, v2di, int)
7363 v16qi __builtin_ia32_pabsb128 (v16qi)
7364 v4si __builtin_ia32_pabsd128 (v4si)
7365 v8hi __builtin_ia32_pabsw128 (v8hi)
7368 The following built-in functions are available when @option{-msse4a} is used.
7371 void _mm_stream_sd (double*,__m128d);
7372 Generates the @code{movntsd} machine instruction.
7373 void _mm_stream_ss (float*,__m128);
7374 Generates the @code{movntss} machine instruction.
7375 __m128i _mm_extract_si64 (__m128i, __m128i);
7376 Generates the @code{extrq} machine instruction with only SSE register operands.
7377 __m128i _mm_extracti_si64 (__m128i, int, int);
7378 Generates the @code{extrq} machine instruction with SSE register and immediate operands.
7379 __m128i _mm_insert_si64 (__m128i, __m128i);
7380 Generates the @code{insertq} machine instruction with only SSE register operands.
7381 __m128i _mm_inserti_si64 (__m128i, __m128i, int, int);
7382 Generates the @code{insertq} machine instruction with SSE register and immediate operands.
7385 The following built-in functions are available when @option{-m3dnow} is used.
7386 All of them generate the machine instruction that is part of the name.
7389 void __builtin_ia32_femms (void)
7390 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
7391 v2si __builtin_ia32_pf2id (v2sf)
7392 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
7393 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
7394 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
7395 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
7396 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
7397 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
7398 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
7399 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
7400 v2sf __builtin_ia32_pfrcp (v2sf)
7401 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
7402 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
7403 v2sf __builtin_ia32_pfrsqrt (v2sf)
7404 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
7405 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
7406 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
7407 v2sf __builtin_ia32_pi2fd (v2si)
7408 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
7411 The following built-in functions are available when both @option{-m3dnow}
7412 and @option{-march=athlon} are used. All of them generate the machine
7413 instruction that is part of the name.
7416 v2si __builtin_ia32_pf2iw (v2sf)
7417 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
7418 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
7419 v2sf __builtin_ia32_pi2fw (v2si)
7420 v2sf __builtin_ia32_pswapdsf (v2sf)
7421 v2si __builtin_ia32_pswapdsi (v2si)
7424 @node MIPS DSP Built-in Functions
7425 @subsection MIPS DSP Built-in Functions
7427 The MIPS DSP Application-Specific Extension (ASE) includes new
7428 instructions that are designed to improve the performance of DSP and
7429 media applications. It provides instructions that operate on packed
7430 8-bit integer data, Q15 fractional data and Q31 fractional data.
7432 GCC supports MIPS DSP operations using both the generic
7433 vector extensions (@pxref{Vector Extensions}) and a collection of
7434 MIPS-specific built-in functions. Both kinds of support are
7435 enabled by the @option{-mdsp} command-line option.
7437 At present, GCC only provides support for operations on 32-bit
7438 vectors. The vector type associated with 8-bit integer data is
7439 usually called @code{v4i8} and the vector type associated with Q15 is
7440 usually called @code{v2q15}. They can be defined in C as follows:
7443 typedef char v4i8 __attribute__ ((vector_size(4)));
7444 typedef short v2q15 __attribute__ ((vector_size(4)));
7447 @code{v4i8} and @code{v2q15} values are initialized in the same way as
7448 aggregates. For example:
7451 v4i8 a = @{1, 2, 3, 4@};
7453 b = (v4i8) @{5, 6, 7, 8@};
7455 v2q15 c = @{0x0fcb, 0x3a75@};
7457 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
7460 @emph{Note:} The CPU's endianness determines the order in which values
7461 are packed. On little-endian targets, the first value is the least
7462 significant and the last value is the most significant. The opposite
7463 order applies to big-endian targets. For example, the code above will
7464 set the lowest byte of @code{a} to @code{1} on little-endian targets
7465 and @code{4} on big-endian targets.
7467 @emph{Note:} Q15 and Q31 values must be initialized with their integer
7468 representation. As shown in this example, the integer representation
7469 of a Q15 value can be obtained by multiplying the fractional value by
7470 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
7473 The table below lists the @code{v4i8} and @code{v2q15} operations for which
7474 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
7475 and @code{c} and @code{d} are @code{v2q15} values.
7477 @multitable @columnfractions .50 .50
7478 @item C code @tab MIPS instruction
7479 @item @code{a + b} @tab @code{addu.qb}
7480 @item @code{c + d} @tab @code{addq.ph}
7481 @item @code{a - b} @tab @code{subu.qb}
7482 @item @code{c - d} @tab @code{subq.ph}
7485 It is easier to describe the DSP built-in functions if we first define
7486 the following types:
7491 typedef long long a64;
7494 @code{q31} and @code{i32} are actually the same as @code{int}, but we
7495 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
7496 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
7497 @code{long long}, but we use @code{a64} to indicate values that will
7498 be placed in one of the four DSP accumulators (@code{$ac0},
7499 @code{$ac1}, @code{$ac2} or @code{$ac3}).
7501 Also, some built-in functions prefer or require immediate numbers as
7502 parameters, because the corresponding DSP instructions accept both immediate
7503 numbers and register operands, or accept immediate numbers only. The
7504 immediate parameters are listed as follows.
7512 imm_n32_31: -32 to 31.
7513 imm_n512_511: -512 to 511.
7516 The following built-in functions map directly to a particular MIPS DSP
7517 instruction. Please refer to the architecture specification
7518 for details on what each instruction does.
7521 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
7522 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
7523 q31 __builtin_mips_addq_s_w (q31, q31)
7524 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
7525 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
7526 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
7527 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
7528 q31 __builtin_mips_subq_s_w (q31, q31)
7529 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
7530 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
7531 i32 __builtin_mips_addsc (i32, i32)
7532 i32 __builtin_mips_addwc (i32, i32)
7533 i32 __builtin_mips_modsub (i32, i32)
7534 i32 __builtin_mips_raddu_w_qb (v4i8)
7535 v2q15 __builtin_mips_absq_s_ph (v2q15)
7536 q31 __builtin_mips_absq_s_w (q31)
7537 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
7538 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
7539 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
7540 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
7541 q31 __builtin_mips_preceq_w_phl (v2q15)
7542 q31 __builtin_mips_preceq_w_phr (v2q15)
7543 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
7544 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
7545 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
7546 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
7547 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
7548 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
7549 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
7550 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
7551 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
7552 v4i8 __builtin_mips_shll_qb (v4i8, i32)
7553 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
7554 v2q15 __builtin_mips_shll_ph (v2q15, i32)
7555 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
7556 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
7557 q31 __builtin_mips_shll_s_w (q31, imm0_31)
7558 q31 __builtin_mips_shll_s_w (q31, i32)
7559 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
7560 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
7561 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
7562 v2q15 __builtin_mips_shra_ph (v2q15, i32)
7563 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
7564 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
7565 q31 __builtin_mips_shra_r_w (q31, imm0_31)
7566 q31 __builtin_mips_shra_r_w (q31, i32)
7567 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
7568 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
7569 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
7570 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
7571 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
7572 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
7573 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
7574 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
7575 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
7576 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
7577 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
7578 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
7579 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
7580 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
7581 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
7582 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
7583 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
7584 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
7585 i32 __builtin_mips_bitrev (i32)
7586 i32 __builtin_mips_insv (i32, i32)
7587 v4i8 __builtin_mips_repl_qb (imm0_255)
7588 v4i8 __builtin_mips_repl_qb (i32)
7589 v2q15 __builtin_mips_repl_ph (imm_n512_511)
7590 v2q15 __builtin_mips_repl_ph (i32)
7591 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
7592 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
7593 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
7594 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
7595 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
7596 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
7597 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
7598 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
7599 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
7600 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
7601 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
7602 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
7603 i32 __builtin_mips_extr_w (a64, imm0_31)
7604 i32 __builtin_mips_extr_w (a64, i32)
7605 i32 __builtin_mips_extr_r_w (a64, imm0_31)
7606 i32 __builtin_mips_extr_s_h (a64, i32)
7607 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
7608 i32 __builtin_mips_extr_rs_w (a64, i32)
7609 i32 __builtin_mips_extr_s_h (a64, imm0_31)
7610 i32 __builtin_mips_extr_r_w (a64, i32)
7611 i32 __builtin_mips_extp (a64, imm0_31)
7612 i32 __builtin_mips_extp (a64, i32)
7613 i32 __builtin_mips_extpdp (a64, imm0_31)
7614 i32 __builtin_mips_extpdp (a64, i32)
7615 a64 __builtin_mips_shilo (a64, imm_n32_31)
7616 a64 __builtin_mips_shilo (a64, i32)
7617 a64 __builtin_mips_mthlip (a64, i32)
7618 void __builtin_mips_wrdsp (i32, imm0_63)
7619 i32 __builtin_mips_rddsp (imm0_63)
7620 i32 __builtin_mips_lbux (void *, i32)
7621 i32 __builtin_mips_lhx (void *, i32)
7622 i32 __builtin_mips_lwx (void *, i32)
7623 i32 __builtin_mips_bposge32 (void)
7626 @node MIPS Paired-Single Support
7627 @subsection MIPS Paired-Single Support
7629 The MIPS64 architecture includes a number of instructions that
7630 operate on pairs of single-precision floating-point values.
7631 Each pair is packed into a 64-bit floating-point register,
7632 with one element being designated the ``upper half'' and
7633 the other being designated the ``lower half''.
7635 GCC supports paired-single operations using both the generic
7636 vector extensions (@pxref{Vector Extensions}) and a collection of
7637 MIPS-specific built-in functions. Both kinds of support are
7638 enabled by the @option{-mpaired-single} command-line option.
7640 The vector type associated with paired-single values is usually
7641 called @code{v2sf}. It can be defined in C as follows:
7644 typedef float v2sf __attribute__ ((vector_size (8)));
7647 @code{v2sf} values are initialized in the same way as aggregates.
7651 v2sf a = @{1.5, 9.1@};
7654 b = (v2sf) @{e, f@};
7657 @emph{Note:} The CPU's endianness determines which value is stored in
7658 the upper half of a register and which value is stored in the lower half.
7659 On little-endian targets, the first value is the lower one and the second
7660 value is the upper one. The opposite order applies to big-endian targets.
7661 For example, the code above will set the lower half of @code{a} to
7662 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
7665 * Paired-Single Arithmetic::
7666 * Paired-Single Built-in Functions::
7667 * MIPS-3D Built-in Functions::
7670 @node Paired-Single Arithmetic
7671 @subsubsection Paired-Single Arithmetic
7673 The table below lists the @code{v2sf} operations for which hardware
7674 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
7675 values and @code{x} is an integral value.
7677 @multitable @columnfractions .50 .50
7678 @item C code @tab MIPS instruction
7679 @item @code{a + b} @tab @code{add.ps}
7680 @item @code{a - b} @tab @code{sub.ps}
7681 @item @code{-a} @tab @code{neg.ps}
7682 @item @code{a * b} @tab @code{mul.ps}
7683 @item @code{a * b + c} @tab @code{madd.ps}
7684 @item @code{a * b - c} @tab @code{msub.ps}
7685 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
7686 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
7687 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
7690 Note that the multiply-accumulate instructions can be disabled
7691 using the command-line option @code{-mno-fused-madd}.
7693 @node Paired-Single Built-in Functions
7694 @subsubsection Paired-Single Built-in Functions
7696 The following paired-single functions map directly to a particular
7697 MIPS instruction. Please refer to the architecture specification
7698 for details on what each instruction does.
7701 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
7702 Pair lower lower (@code{pll.ps}).
7704 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
7705 Pair upper lower (@code{pul.ps}).
7707 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
7708 Pair lower upper (@code{plu.ps}).
7710 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
7711 Pair upper upper (@code{puu.ps}).
7713 @item v2sf __builtin_mips_cvt_ps_s (float, float)
7714 Convert pair to paired single (@code{cvt.ps.s}).
7716 @item float __builtin_mips_cvt_s_pl (v2sf)
7717 Convert pair lower to single (@code{cvt.s.pl}).
7719 @item float __builtin_mips_cvt_s_pu (v2sf)
7720 Convert pair upper to single (@code{cvt.s.pu}).
7722 @item v2sf __builtin_mips_abs_ps (v2sf)
7723 Absolute value (@code{abs.ps}).
7725 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
7726 Align variable (@code{alnv.ps}).
7728 @emph{Note:} The value of the third parameter must be 0 or 4
7729 modulo 8, otherwise the result will be unpredictable. Please read the
7730 instruction description for details.
7733 The following multi-instruction functions are also available.
7734 In each case, @var{cond} can be any of the 16 floating-point conditions:
7735 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
7736 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
7737 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
7740 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7741 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7742 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
7743 @code{movt.ps}/@code{movf.ps}).
7745 The @code{movt} functions return the value @var{x} computed by:
7748 c.@var{cond}.ps @var{cc},@var{a},@var{b}
7749 mov.ps @var{x},@var{c}
7750 movt.ps @var{x},@var{d},@var{cc}
7753 The @code{movf} functions are similar but use @code{movf.ps} instead
7756 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7757 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7758 Comparison of two paired-single values (@code{c.@var{cond}.ps},
7759 @code{bc1t}/@code{bc1f}).
7761 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
7762 and return either the upper or lower half of the result. For example:
7766 if (__builtin_mips_upper_c_eq_ps (a, b))
7767 upper_halves_are_equal ();
7769 upper_halves_are_unequal ();
7771 if (__builtin_mips_lower_c_eq_ps (a, b))
7772 lower_halves_are_equal ();
7774 lower_halves_are_unequal ();
7778 @node MIPS-3D Built-in Functions
7779 @subsubsection MIPS-3D Built-in Functions
7781 The MIPS-3D Application-Specific Extension (ASE) includes additional
7782 paired-single instructions that are designed to improve the performance
7783 of 3D graphics operations. Support for these instructions is controlled
7784 by the @option{-mips3d} command-line option.
7786 The functions listed below map directly to a particular MIPS-3D
7787 instruction. Please refer to the architecture specification for
7788 more details on what each instruction does.
7791 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
7792 Reduction add (@code{addr.ps}).
7794 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
7795 Reduction multiply (@code{mulr.ps}).
7797 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
7798 Convert paired single to paired word (@code{cvt.pw.ps}).
7800 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
7801 Convert paired word to paired single (@code{cvt.ps.pw}).
7803 @item float __builtin_mips_recip1_s (float)
7804 @itemx double __builtin_mips_recip1_d (double)
7805 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
7806 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
7808 @item float __builtin_mips_recip2_s (float, float)
7809 @itemx double __builtin_mips_recip2_d (double, double)
7810 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
7811 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
7813 @item float __builtin_mips_rsqrt1_s (float)
7814 @itemx double __builtin_mips_rsqrt1_d (double)
7815 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
7816 Reduced precision reciprocal square root (sequence step 1)
7817 (@code{rsqrt1.@var{fmt}}).
7819 @item float __builtin_mips_rsqrt2_s (float, float)
7820 @itemx double __builtin_mips_rsqrt2_d (double, double)
7821 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
7822 Reduced precision reciprocal square root (sequence step 2)
7823 (@code{rsqrt2.@var{fmt}}).
7826 The following multi-instruction functions are also available.
7827 In each case, @var{cond} can be any of the 16 floating-point conditions:
7828 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
7829 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
7830 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
7833 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
7834 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
7835 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
7836 @code{bc1t}/@code{bc1f}).
7838 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
7839 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
7844 if (__builtin_mips_cabs_eq_s (a, b))
7850 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7851 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7852 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
7853 @code{bc1t}/@code{bc1f}).
7855 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
7856 and return either the upper or lower half of the result. For example:
7860 if (__builtin_mips_upper_cabs_eq_ps (a, b))
7861 upper_halves_are_equal ();
7863 upper_halves_are_unequal ();
7865 if (__builtin_mips_lower_cabs_eq_ps (a, b))
7866 lower_halves_are_equal ();
7868 lower_halves_are_unequal ();
7871 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7872 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7873 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
7874 @code{movt.ps}/@code{movf.ps}).
7876 The @code{movt} functions return the value @var{x} computed by:
7879 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
7880 mov.ps @var{x},@var{c}
7881 movt.ps @var{x},@var{d},@var{cc}
7884 The @code{movf} functions are similar but use @code{movf.ps} instead
7887 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7888 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7889 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7890 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7891 Comparison of two paired-single values
7892 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
7893 @code{bc1any2t}/@code{bc1any2f}).
7895 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
7896 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
7897 result is true and the @code{all} forms return true if both results are true.
7902 if (__builtin_mips_any_c_eq_ps (a, b))
7907 if (__builtin_mips_all_c_eq_ps (a, b))
7913 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7914 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7915 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7916 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7917 Comparison of four paired-single values
7918 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
7919 @code{bc1any4t}/@code{bc1any4f}).
7921 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
7922 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
7923 The @code{any} forms return true if any of the four results are true
7924 and the @code{all} forms return true if all four results are true.
7929 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
7934 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
7941 @node PowerPC AltiVec Built-in Functions
7942 @subsection PowerPC AltiVec Built-in Functions
7944 GCC provides an interface for the PowerPC family of processors to access
7945 the AltiVec operations described in Motorola's AltiVec Programming
7946 Interface Manual. The interface is made available by including
7947 @code{<altivec.h>} and using @option{-maltivec} and
7948 @option{-mabi=altivec}. The interface supports the following vector
7952 vector unsigned char
7956 vector unsigned short
7967 GCC's implementation of the high-level language interface available from
7968 C and C++ code differs from Motorola's documentation in several ways.
7973 A vector constant is a list of constant expressions within curly braces.
7976 A vector initializer requires no cast if the vector constant is of the
7977 same type as the variable it is initializing.
7980 If @code{signed} or @code{unsigned} is omitted, the signedness of the
7981 vector type is the default signedness of the base type. The default
7982 varies depending on the operating system, so a portable program should
7983 always specify the signedness.
7986 Compiling with @option{-maltivec} adds keywords @code{__vector},
7987 @code{__pixel}, and @code{__bool}. Macros @option{vector},
7988 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
7992 GCC allows using a @code{typedef} name as the type specifier for a
7996 For C, overloaded functions are implemented with macros so the following
8000 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
8003 Since @code{vec_add} is a macro, the vector constant in the example
8004 is treated as four separate arguments. Wrap the entire argument in
8005 parentheses for this to work.
8008 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
8009 Internally, GCC uses built-in functions to achieve the functionality in
8010 the aforementioned header file, but they are not supported and are
8011 subject to change without notice.
8013 The following interfaces are supported for the generic and specific
8014 AltiVec operations and the AltiVec predicates. In cases where there
8015 is a direct mapping between generic and specific operations, only the
8016 generic names are shown here, although the specific operations can also
8019 Arguments that are documented as @code{const int} require literal
8020 integral values within the range required for that operation.
8023 vector signed char vec_abs (vector signed char);
8024 vector signed short vec_abs (vector signed short);
8025 vector signed int vec_abs (vector signed int);
8026 vector float vec_abs (vector float);
8028 vector signed char vec_abss (vector signed char);
8029 vector signed short vec_abss (vector signed short);
8030 vector signed int vec_abss (vector signed int);
8032 vector signed char vec_add (vector bool char, vector signed char);
8033 vector signed char vec_add (vector signed char, vector bool char);
8034 vector signed char vec_add (vector signed char, vector signed char);
8035 vector unsigned char vec_add (vector bool char, vector unsigned char);
8036 vector unsigned char vec_add (vector unsigned char, vector bool char);
8037 vector unsigned char vec_add (vector unsigned char,
8038 vector unsigned char);
8039 vector signed short vec_add (vector bool short, vector signed short);
8040 vector signed short vec_add (vector signed short, vector bool short);
8041 vector signed short vec_add (vector signed short, vector signed short);
8042 vector unsigned short vec_add (vector bool short,
8043 vector unsigned short);
8044 vector unsigned short vec_add (vector unsigned short,
8046 vector unsigned short vec_add (vector unsigned short,
8047 vector unsigned short);
8048 vector signed int vec_add (vector bool int, vector signed int);
8049 vector signed int vec_add (vector signed int, vector bool int);
8050 vector signed int vec_add (vector signed int, vector signed int);
8051 vector unsigned int vec_add (vector bool int, vector unsigned int);
8052 vector unsigned int vec_add (vector unsigned int, vector bool int);
8053 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
8054 vector float vec_add (vector float, vector float);
8056 vector float vec_vaddfp (vector float, vector float);
8058 vector signed int vec_vadduwm (vector bool int, vector signed int);
8059 vector signed int vec_vadduwm (vector signed int, vector bool int);
8060 vector signed int vec_vadduwm (vector signed int, vector signed int);
8061 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
8062 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
8063 vector unsigned int vec_vadduwm (vector unsigned int,
8064 vector unsigned int);
8066 vector signed short vec_vadduhm (vector bool short,
8067 vector signed short);
8068 vector signed short vec_vadduhm (vector signed short,
8070 vector signed short vec_vadduhm (vector signed short,
8071 vector signed short);
8072 vector unsigned short vec_vadduhm (vector bool short,
8073 vector unsigned short);
8074 vector unsigned short vec_vadduhm (vector unsigned short,
8076 vector unsigned short vec_vadduhm (vector unsigned short,
8077 vector unsigned short);
8079 vector signed char vec_vaddubm (vector bool char, vector signed char);
8080 vector signed char vec_vaddubm (vector signed char, vector bool char);
8081 vector signed char vec_vaddubm (vector signed char, vector signed char);
8082 vector unsigned char vec_vaddubm (vector bool char,
8083 vector unsigned char);
8084 vector unsigned char vec_vaddubm (vector unsigned char,
8086 vector unsigned char vec_vaddubm (vector unsigned char,
8087 vector unsigned char);
8089 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
8091 vector unsigned char vec_adds (vector bool char, vector unsigned char);
8092 vector unsigned char vec_adds (vector unsigned char, vector bool char);
8093 vector unsigned char vec_adds (vector unsigned char,
8094 vector unsigned char);
8095 vector signed char vec_adds (vector bool char, vector signed char);
8096 vector signed char vec_adds (vector signed char, vector bool char);
8097 vector signed char vec_adds (vector signed char, vector signed char);
8098 vector unsigned short vec_adds (vector bool short,
8099 vector unsigned short);
8100 vector unsigned short vec_adds (vector unsigned short,
8102 vector unsigned short vec_adds (vector unsigned short,
8103 vector unsigned short);
8104 vector signed short vec_adds (vector bool short, vector signed short);
8105 vector signed short vec_adds (vector signed short, vector bool short);
8106 vector signed short vec_adds (vector signed short, vector signed short);
8107 vector unsigned int vec_adds (vector bool int, vector unsigned int);
8108 vector unsigned int vec_adds (vector unsigned int, vector bool int);
8109 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
8110 vector signed int vec_adds (vector bool int, vector signed int);
8111 vector signed int vec_adds (vector signed int, vector bool int);
8112 vector signed int vec_adds (vector signed int, vector signed int);
8114 vector signed int vec_vaddsws (vector bool int, vector signed int);
8115 vector signed int vec_vaddsws (vector signed int, vector bool int);
8116 vector signed int vec_vaddsws (vector signed int, vector signed int);
8118 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
8119 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
8120 vector unsigned int vec_vadduws (vector unsigned int,
8121 vector unsigned int);
8123 vector signed short vec_vaddshs (vector bool short,
8124 vector signed short);
8125 vector signed short vec_vaddshs (vector signed short,
8127 vector signed short vec_vaddshs (vector signed short,
8128 vector signed short);
8130 vector unsigned short vec_vadduhs (vector bool short,
8131 vector unsigned short);
8132 vector unsigned short vec_vadduhs (vector unsigned short,
8134 vector unsigned short vec_vadduhs (vector unsigned short,
8135 vector unsigned short);
8137 vector signed char vec_vaddsbs (vector bool char, vector signed char);
8138 vector signed char vec_vaddsbs (vector signed char, vector bool char);
8139 vector signed char vec_vaddsbs (vector signed char, vector signed char);
8141 vector unsigned char vec_vaddubs (vector bool char,
8142 vector unsigned char);
8143 vector unsigned char vec_vaddubs (vector unsigned char,
8145 vector unsigned char vec_vaddubs (vector unsigned char,
8146 vector unsigned char);
8148 vector float vec_and (vector float, vector float);
8149 vector float vec_and (vector float, vector bool int);
8150 vector float vec_and (vector bool int, vector float);
8151 vector bool int vec_and (vector bool int, vector bool int);
8152 vector signed int vec_and (vector bool int, vector signed int);
8153 vector signed int vec_and (vector signed int, vector bool int);
8154 vector signed int vec_and (vector signed int, vector signed int);
8155 vector unsigned int vec_and (vector bool int, vector unsigned int);
8156 vector unsigned int vec_and (vector unsigned int, vector bool int);
8157 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
8158 vector bool short vec_and (vector bool short, vector bool short);
8159 vector signed short vec_and (vector bool short, vector signed short);
8160 vector signed short vec_and (vector signed short, vector bool short);
8161 vector signed short vec_and (vector signed short, vector signed short);
8162 vector unsigned short vec_and (vector bool short,
8163 vector unsigned short);
8164 vector unsigned short vec_and (vector unsigned short,
8166 vector unsigned short vec_and (vector unsigned short,
8167 vector unsigned short);
8168 vector signed char vec_and (vector bool char, vector signed char);
8169 vector bool char vec_and (vector bool char, vector bool char);
8170 vector signed char vec_and (vector signed char, vector bool char);
8171 vector signed char vec_and (vector signed char, vector signed char);
8172 vector unsigned char vec_and (vector bool char, vector unsigned char);
8173 vector unsigned char vec_and (vector unsigned char, vector bool char);
8174 vector unsigned char vec_and (vector unsigned char,
8175 vector unsigned char);
8177 vector float vec_andc (vector float, vector float);
8178 vector float vec_andc (vector float, vector bool int);
8179 vector float vec_andc (vector bool int, vector float);
8180 vector bool int vec_andc (vector bool int, vector bool int);
8181 vector signed int vec_andc (vector bool int, vector signed int);
8182 vector signed int vec_andc (vector signed int, vector bool int);
8183 vector signed int vec_andc (vector signed int, vector signed int);
8184 vector unsigned int vec_andc (vector bool int, vector unsigned int);
8185 vector unsigned int vec_andc (vector unsigned int, vector bool int);
8186 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
8187 vector bool short vec_andc (vector bool short, vector bool short);
8188 vector signed short vec_andc (vector bool short, vector signed short);
8189 vector signed short vec_andc (vector signed short, vector bool short);
8190 vector signed short vec_andc (vector signed short, vector signed short);
8191 vector unsigned short vec_andc (vector bool short,
8192 vector unsigned short);
8193 vector unsigned short vec_andc (vector unsigned short,
8195 vector unsigned short vec_andc (vector unsigned short,
8196 vector unsigned short);
8197 vector signed char vec_andc (vector bool char, vector signed char);
8198 vector bool char vec_andc (vector bool char, vector bool char);
8199 vector signed char vec_andc (vector signed char, vector bool char);
8200 vector signed char vec_andc (vector signed char, vector signed char);
8201 vector unsigned char vec_andc (vector bool char, vector unsigned char);
8202 vector unsigned char vec_andc (vector unsigned char, vector bool char);
8203 vector unsigned char vec_andc (vector unsigned char,
8204 vector unsigned char);
8206 vector unsigned char vec_avg (vector unsigned char,
8207 vector unsigned char);
8208 vector signed char vec_avg (vector signed char, vector signed char);
8209 vector unsigned short vec_avg (vector unsigned short,
8210 vector unsigned short);
8211 vector signed short vec_avg (vector signed short, vector signed short);
8212 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
8213 vector signed int vec_avg (vector signed int, vector signed int);
8215 vector signed int vec_vavgsw (vector signed int, vector signed int);
8217 vector unsigned int vec_vavguw (vector unsigned int,
8218 vector unsigned int);
8220 vector signed short vec_vavgsh (vector signed short,
8221 vector signed short);
8223 vector unsigned short vec_vavguh (vector unsigned short,
8224 vector unsigned short);
8226 vector signed char vec_vavgsb (vector signed char, vector signed char);
8228 vector unsigned char vec_vavgub (vector unsigned char,
8229 vector unsigned char);
8231 vector float vec_ceil (vector float);
8233 vector signed int vec_cmpb (vector float, vector float);
8235 vector bool char vec_cmpeq (vector signed char, vector signed char);
8236 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
8237 vector bool short vec_cmpeq (vector signed short, vector signed short);
8238 vector bool short vec_cmpeq (vector unsigned short,
8239 vector unsigned short);
8240 vector bool int vec_cmpeq (vector signed int, vector signed int);
8241 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
8242 vector bool int vec_cmpeq (vector float, vector float);
8244 vector bool int vec_vcmpeqfp (vector float, vector float);
8246 vector bool int vec_vcmpequw (vector signed int, vector signed int);
8247 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
8249 vector bool short vec_vcmpequh (vector signed short,
8250 vector signed short);
8251 vector bool short vec_vcmpequh (vector unsigned short,
8252 vector unsigned short);
8254 vector bool char vec_vcmpequb (vector signed char, vector signed char);
8255 vector bool char vec_vcmpequb (vector unsigned char,
8256 vector unsigned char);
8258 vector bool int vec_cmpge (vector float, vector float);
8260 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
8261 vector bool char vec_cmpgt (vector signed char, vector signed char);
8262 vector bool short vec_cmpgt (vector unsigned short,
8263 vector unsigned short);
8264 vector bool short vec_cmpgt (vector signed short, vector signed short);
8265 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
8266 vector bool int vec_cmpgt (vector signed int, vector signed int);
8267 vector bool int vec_cmpgt (vector float, vector float);
8269 vector bool int vec_vcmpgtfp (vector float, vector float);
8271 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
8273 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
8275 vector bool short vec_vcmpgtsh (vector signed short,
8276 vector signed short);
8278 vector bool short vec_vcmpgtuh (vector unsigned short,
8279 vector unsigned short);
8281 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
8283 vector bool char vec_vcmpgtub (vector unsigned char,
8284 vector unsigned char);
8286 vector bool int vec_cmple (vector float, vector float);
8288 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
8289 vector bool char vec_cmplt (vector signed char, vector signed char);
8290 vector bool short vec_cmplt (vector unsigned short,
8291 vector unsigned short);
8292 vector bool short vec_cmplt (vector signed short, vector signed short);
8293 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
8294 vector bool int vec_cmplt (vector signed int, vector signed int);
8295 vector bool int vec_cmplt (vector float, vector float);
8297 vector float vec_ctf (vector unsigned int, const int);
8298 vector float vec_ctf (vector signed int, const int);
8300 vector float vec_vcfsx (vector signed int, const int);
8302 vector float vec_vcfux (vector unsigned int, const int);
8304 vector signed int vec_cts (vector float, const int);
8306 vector unsigned int vec_ctu (vector float, const int);
8308 void vec_dss (const int);
8310 void vec_dssall (void);
8312 void vec_dst (const vector unsigned char *, int, const int);
8313 void vec_dst (const vector signed char *, int, const int);
8314 void vec_dst (const vector bool char *, int, const int);
8315 void vec_dst (const vector unsigned short *, int, const int);
8316 void vec_dst (const vector signed short *, int, const int);
8317 void vec_dst (const vector bool short *, int, const int);
8318 void vec_dst (const vector pixel *, int, const int);
8319 void vec_dst (const vector unsigned int *, int, const int);
8320 void vec_dst (const vector signed int *, int, const int);
8321 void vec_dst (const vector bool int *, int, const int);
8322 void vec_dst (const vector float *, int, const int);
8323 void vec_dst (const unsigned char *, int, const int);
8324 void vec_dst (const signed char *, int, const int);
8325 void vec_dst (const unsigned short *, int, const int);
8326 void vec_dst (const short *, int, const int);
8327 void vec_dst (const unsigned int *, int, const int);
8328 void vec_dst (const int *, int, const int);
8329 void vec_dst (const unsigned long *, int, const int);
8330 void vec_dst (const long *, int, const int);
8331 void vec_dst (const float *, int, const int);
8333 void vec_dstst (const vector unsigned char *, int, const int);
8334 void vec_dstst (const vector signed char *, int, const int);
8335 void vec_dstst (const vector bool char *, int, const int);
8336 void vec_dstst (const vector unsigned short *, int, const int);
8337 void vec_dstst (const vector signed short *, int, const int);
8338 void vec_dstst (const vector bool short *, int, const int);
8339 void vec_dstst (const vector pixel *, int, const int);
8340 void vec_dstst (const vector unsigned int *, int, const int);
8341 void vec_dstst (const vector signed int *, int, const int);
8342 void vec_dstst (const vector bool int *, int, const int);
8343 void vec_dstst (const vector float *, int, const int);
8344 void vec_dstst (const unsigned char *, int, const int);
8345 void vec_dstst (const signed char *, int, const int);
8346 void vec_dstst (const unsigned short *, int, const int);
8347 void vec_dstst (const short *, int, const int);
8348 void vec_dstst (const unsigned int *, int, const int);
8349 void vec_dstst (const int *, int, const int);
8350 void vec_dstst (const unsigned long *, int, const int);
8351 void vec_dstst (const long *, int, const int);
8352 void vec_dstst (const float *, int, const int);
8354 void vec_dststt (const vector unsigned char *, int, const int);
8355 void vec_dststt (const vector signed char *, int, const int);
8356 void vec_dststt (const vector bool char *, int, const int);
8357 void vec_dststt (const vector unsigned short *, int, const int);
8358 void vec_dststt (const vector signed short *, int, const int);
8359 void vec_dststt (const vector bool short *, int, const int);
8360 void vec_dststt (const vector pixel *, int, const int);
8361 void vec_dststt (const vector unsigned int *, int, const int);
8362 void vec_dststt (const vector signed int *, int, const int);
8363 void vec_dststt (const vector bool int *, int, const int);
8364 void vec_dststt (const vector float *, int, const int);
8365 void vec_dststt (const unsigned char *, int, const int);
8366 void vec_dststt (const signed char *, int, const int);
8367 void vec_dststt (const unsigned short *, int, const int);
8368 void vec_dststt (const short *, int, const int);
8369 void vec_dststt (const unsigned int *, int, const int);
8370 void vec_dststt (const int *, int, const int);
8371 void vec_dststt (const unsigned long *, int, const int);
8372 void vec_dststt (const long *, int, const int);
8373 void vec_dststt (const float *, int, const int);
8375 void vec_dstt (const vector unsigned char *, int, const int);
8376 void vec_dstt (const vector signed char *, int, const int);
8377 void vec_dstt (const vector bool char *, int, const int);
8378 void vec_dstt (const vector unsigned short *, int, const int);
8379 void vec_dstt (const vector signed short *, int, const int);
8380 void vec_dstt (const vector bool short *, int, const int);
8381 void vec_dstt (const vector pixel *, int, const int);
8382 void vec_dstt (const vector unsigned int *, int, const int);
8383 void vec_dstt (const vector signed int *, int, const int);
8384 void vec_dstt (const vector bool int *, int, const int);
8385 void vec_dstt (const vector float *, int, const int);
8386 void vec_dstt (const unsigned char *, int, const int);
8387 void vec_dstt (const signed char *, int, const int);
8388 void vec_dstt (const unsigned short *, int, const int);
8389 void vec_dstt (const short *, int, const int);
8390 void vec_dstt (const unsigned int *, int, const int);
8391 void vec_dstt (const int *, int, const int);
8392 void vec_dstt (const unsigned long *, int, const int);
8393 void vec_dstt (const long *, int, const int);
8394 void vec_dstt (const float *, int, const int);
8396 vector float vec_expte (vector float);
8398 vector float vec_floor (vector float);
8400 vector float vec_ld (int, const vector float *);
8401 vector float vec_ld (int, const float *);
8402 vector bool int vec_ld (int, const vector bool int *);
8403 vector signed int vec_ld (int, const vector signed int *);
8404 vector signed int vec_ld (int, const int *);
8405 vector signed int vec_ld (int, const long *);
8406 vector unsigned int vec_ld (int, const vector unsigned int *);
8407 vector unsigned int vec_ld (int, const unsigned int *);
8408 vector unsigned int vec_ld (int, const unsigned long *);
8409 vector bool short vec_ld (int, const vector bool short *);
8410 vector pixel vec_ld (int, const vector pixel *);
8411 vector signed short vec_ld (int, const vector signed short *);
8412 vector signed short vec_ld (int, const short *);
8413 vector unsigned short vec_ld (int, const vector unsigned short *);
8414 vector unsigned short vec_ld (int, const unsigned short *);
8415 vector bool char vec_ld (int, const vector bool char *);
8416 vector signed char vec_ld (int, const vector signed char *);
8417 vector signed char vec_ld (int, const signed char *);
8418 vector unsigned char vec_ld (int, const vector unsigned char *);
8419 vector unsigned char vec_ld (int, const unsigned char *);
8421 vector signed char vec_lde (int, const signed char *);
8422 vector unsigned char vec_lde (int, const unsigned char *);
8423 vector signed short vec_lde (int, const short *);
8424 vector unsigned short vec_lde (int, const unsigned short *);
8425 vector float vec_lde (int, const float *);
8426 vector signed int vec_lde (int, const int *);
8427 vector unsigned int vec_lde (int, const unsigned int *);
8428 vector signed int vec_lde (int, const long *);
8429 vector unsigned int vec_lde (int, const unsigned long *);
8431 vector float vec_lvewx (int, float *);
8432 vector signed int vec_lvewx (int, int *);
8433 vector unsigned int vec_lvewx (int, unsigned int *);
8434 vector signed int vec_lvewx (int, long *);
8435 vector unsigned int vec_lvewx (int, unsigned long *);
8437 vector signed short vec_lvehx (int, short *);
8438 vector unsigned short vec_lvehx (int, unsigned short *);
8440 vector signed char vec_lvebx (int, char *);
8441 vector unsigned char vec_lvebx (int, unsigned char *);
8443 vector float vec_ldl (int, const vector float *);
8444 vector float vec_ldl (int, const float *);
8445 vector bool int vec_ldl (int, const vector bool int *);
8446 vector signed int vec_ldl (int, const vector signed int *);
8447 vector signed int vec_ldl (int, const int *);
8448 vector signed int vec_ldl (int, const long *);
8449 vector unsigned int vec_ldl (int, const vector unsigned int *);
8450 vector unsigned int vec_ldl (int, const unsigned int *);
8451 vector unsigned int vec_ldl (int, const unsigned long *);
8452 vector bool short vec_ldl (int, const vector bool short *);
8453 vector pixel vec_ldl (int, const vector pixel *);
8454 vector signed short vec_ldl (int, const vector signed short *);
8455 vector signed short vec_ldl (int, const short *);
8456 vector unsigned short vec_ldl (int, const vector unsigned short *);
8457 vector unsigned short vec_ldl (int, const unsigned short *);
8458 vector bool char vec_ldl (int, const vector bool char *);
8459 vector signed char vec_ldl (int, const vector signed char *);
8460 vector signed char vec_ldl (int, const signed char *);
8461 vector unsigned char vec_ldl (int, const vector unsigned char *);
8462 vector unsigned char vec_ldl (int, const unsigned char *);
8464 vector float vec_loge (vector float);
8466 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
8467 vector unsigned char vec_lvsl (int, const volatile signed char *);
8468 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
8469 vector unsigned char vec_lvsl (int, const volatile short *);
8470 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
8471 vector unsigned char vec_lvsl (int, const volatile int *);
8472 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
8473 vector unsigned char vec_lvsl (int, const volatile long *);
8474 vector unsigned char vec_lvsl (int, const volatile float *);
8476 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
8477 vector unsigned char vec_lvsr (int, const volatile signed char *);
8478 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
8479 vector unsigned char vec_lvsr (int, const volatile short *);
8480 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
8481 vector unsigned char vec_lvsr (int, const volatile int *);
8482 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
8483 vector unsigned char vec_lvsr (int, const volatile long *);
8484 vector unsigned char vec_lvsr (int, const volatile float *);
8486 vector float vec_madd (vector float, vector float, vector float);
8488 vector signed short vec_madds (vector signed short,
8489 vector signed short,
8490 vector signed short);
8492 vector unsigned char vec_max (vector bool char, vector unsigned char);
8493 vector unsigned char vec_max (vector unsigned char, vector bool char);
8494 vector unsigned char vec_max (vector unsigned char,
8495 vector unsigned char);
8496 vector signed char vec_max (vector bool char, vector signed char);
8497 vector signed char vec_max (vector signed char, vector bool char);
8498 vector signed char vec_max (vector signed char, vector signed char);
8499 vector unsigned short vec_max (vector bool short,
8500 vector unsigned short);
8501 vector unsigned short vec_max (vector unsigned short,
8503 vector unsigned short vec_max (vector unsigned short,
8504 vector unsigned short);
8505 vector signed short vec_max (vector bool short, vector signed short);
8506 vector signed short vec_max (vector signed short, vector bool short);
8507 vector signed short vec_max (vector signed short, vector signed short);
8508 vector unsigned int vec_max (vector bool int, vector unsigned int);
8509 vector unsigned int vec_max (vector unsigned int, vector bool int);
8510 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
8511 vector signed int vec_max (vector bool int, vector signed int);
8512 vector signed int vec_max (vector signed int, vector bool int);
8513 vector signed int vec_max (vector signed int, vector signed int);
8514 vector float vec_max (vector float, vector float);
8516 vector float vec_vmaxfp (vector float, vector float);
8518 vector signed int vec_vmaxsw (vector bool int, vector signed int);
8519 vector signed int vec_vmaxsw (vector signed int, vector bool int);
8520 vector signed int vec_vmaxsw (vector signed int, vector signed int);
8522 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
8523 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
8524 vector unsigned int vec_vmaxuw (vector unsigned int,
8525 vector unsigned int);
8527 vector signed short vec_vmaxsh (vector bool short, vector signed short);
8528 vector signed short vec_vmaxsh (vector signed short, vector bool short);
8529 vector signed short vec_vmaxsh (vector signed short,
8530 vector signed short);
8532 vector unsigned short vec_vmaxuh (vector bool short,
8533 vector unsigned short);
8534 vector unsigned short vec_vmaxuh (vector unsigned short,
8536 vector unsigned short vec_vmaxuh (vector unsigned short,
8537 vector unsigned short);
8539 vector signed char vec_vmaxsb (vector bool char, vector signed char);
8540 vector signed char vec_vmaxsb (vector signed char, vector bool char);
8541 vector signed char vec_vmaxsb (vector signed char, vector signed char);
8543 vector unsigned char vec_vmaxub (vector bool char,
8544 vector unsigned char);
8545 vector unsigned char vec_vmaxub (vector unsigned char,
8547 vector unsigned char vec_vmaxub (vector unsigned char,
8548 vector unsigned char);
8550 vector bool char vec_mergeh (vector bool char, vector bool char);
8551 vector signed char vec_mergeh (vector signed char, vector signed char);
8552 vector unsigned char vec_mergeh (vector unsigned char,
8553 vector unsigned char);
8554 vector bool short vec_mergeh (vector bool short, vector bool short);
8555 vector pixel vec_mergeh (vector pixel, vector pixel);
8556 vector signed short vec_mergeh (vector signed short,
8557 vector signed short);
8558 vector unsigned short vec_mergeh (vector unsigned short,
8559 vector unsigned short);
8560 vector float vec_mergeh (vector float, vector float);
8561 vector bool int vec_mergeh (vector bool int, vector bool int);
8562 vector signed int vec_mergeh (vector signed int, vector signed int);
8563 vector unsigned int vec_mergeh (vector unsigned int,
8564 vector unsigned int);
8566 vector float vec_vmrghw (vector float, vector float);
8567 vector bool int vec_vmrghw (vector bool int, vector bool int);
8568 vector signed int vec_vmrghw (vector signed int, vector signed int);
8569 vector unsigned int vec_vmrghw (vector unsigned int,
8570 vector unsigned int);
8572 vector bool short vec_vmrghh (vector bool short, vector bool short);
8573 vector signed short vec_vmrghh (vector signed short,
8574 vector signed short);
8575 vector unsigned short vec_vmrghh (vector unsigned short,
8576 vector unsigned short);
8577 vector pixel vec_vmrghh (vector pixel, vector pixel);
8579 vector bool char vec_vmrghb (vector bool char, vector bool char);
8580 vector signed char vec_vmrghb (vector signed char, vector signed char);
8581 vector unsigned char vec_vmrghb (vector unsigned char,
8582 vector unsigned char);
8584 vector bool char vec_mergel (vector bool char, vector bool char);
8585 vector signed char vec_mergel (vector signed char, vector signed char);
8586 vector unsigned char vec_mergel (vector unsigned char,
8587 vector unsigned char);
8588 vector bool short vec_mergel (vector bool short, vector bool short);
8589 vector pixel vec_mergel (vector pixel, vector pixel);
8590 vector signed short vec_mergel (vector signed short,
8591 vector signed short);
8592 vector unsigned short vec_mergel (vector unsigned short,
8593 vector unsigned short);
8594 vector float vec_mergel (vector float, vector float);
8595 vector bool int vec_mergel (vector bool int, vector bool int);
8596 vector signed int vec_mergel (vector signed int, vector signed int);
8597 vector unsigned int vec_mergel (vector unsigned int,
8598 vector unsigned int);
8600 vector float vec_vmrglw (vector float, vector float);
8601 vector signed int vec_vmrglw (vector signed int, vector signed int);
8602 vector unsigned int vec_vmrglw (vector unsigned int,
8603 vector unsigned int);
8604 vector bool int vec_vmrglw (vector bool int, vector bool int);
8606 vector bool short vec_vmrglh (vector bool short, vector bool short);
8607 vector signed short vec_vmrglh (vector signed short,
8608 vector signed short);
8609 vector unsigned short vec_vmrglh (vector unsigned short,
8610 vector unsigned short);
8611 vector pixel vec_vmrglh (vector pixel, vector pixel);
8613 vector bool char vec_vmrglb (vector bool char, vector bool char);
8614 vector signed char vec_vmrglb (vector signed char, vector signed char);
8615 vector unsigned char vec_vmrglb (vector unsigned char,
8616 vector unsigned char);
8618 vector unsigned short vec_mfvscr (void);
8620 vector unsigned char vec_min (vector bool char, vector unsigned char);
8621 vector unsigned char vec_min (vector unsigned char, vector bool char);
8622 vector unsigned char vec_min (vector unsigned char,
8623 vector unsigned char);
8624 vector signed char vec_min (vector bool char, vector signed char);
8625 vector signed char vec_min (vector signed char, vector bool char);
8626 vector signed char vec_min (vector signed char, vector signed char);
8627 vector unsigned short vec_min (vector bool short,
8628 vector unsigned short);
8629 vector unsigned short vec_min (vector unsigned short,
8631 vector unsigned short vec_min (vector unsigned short,
8632 vector unsigned short);
8633 vector signed short vec_min (vector bool short, vector signed short);
8634 vector signed short vec_min (vector signed short, vector bool short);
8635 vector signed short vec_min (vector signed short, vector signed short);
8636 vector unsigned int vec_min (vector bool int, vector unsigned int);
8637 vector unsigned int vec_min (vector unsigned int, vector bool int);
8638 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
8639 vector signed int vec_min (vector bool int, vector signed int);
8640 vector signed int vec_min (vector signed int, vector bool int);
8641 vector signed int vec_min (vector signed int, vector signed int);
8642 vector float vec_min (vector float, vector float);
8644 vector float vec_vminfp (vector float, vector float);
8646 vector signed int vec_vminsw (vector bool int, vector signed int);
8647 vector signed int vec_vminsw (vector signed int, vector bool int);
8648 vector signed int vec_vminsw (vector signed int, vector signed int);
8650 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
8651 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
8652 vector unsigned int vec_vminuw (vector unsigned int,
8653 vector unsigned int);
8655 vector signed short vec_vminsh (vector bool short, vector signed short);
8656 vector signed short vec_vminsh (vector signed short, vector bool short);
8657 vector signed short vec_vminsh (vector signed short,
8658 vector signed short);
8660 vector unsigned short vec_vminuh (vector bool short,
8661 vector unsigned short);
8662 vector unsigned short vec_vminuh (vector unsigned short,
8664 vector unsigned short vec_vminuh (vector unsigned short,
8665 vector unsigned short);
8667 vector signed char vec_vminsb (vector bool char, vector signed char);
8668 vector signed char vec_vminsb (vector signed char, vector bool char);
8669 vector signed char vec_vminsb (vector signed char, vector signed char);
8671 vector unsigned char vec_vminub (vector bool char,
8672 vector unsigned char);
8673 vector unsigned char vec_vminub (vector unsigned char,
8675 vector unsigned char vec_vminub (vector unsigned char,
8676 vector unsigned char);
8678 vector signed short vec_mladd (vector signed short,
8679 vector signed short,
8680 vector signed short);
8681 vector signed short vec_mladd (vector signed short,
8682 vector unsigned short,
8683 vector unsigned short);
8684 vector signed short vec_mladd (vector unsigned short,
8685 vector signed short,
8686 vector signed short);
8687 vector unsigned short vec_mladd (vector unsigned short,
8688 vector unsigned short,
8689 vector unsigned short);
8691 vector signed short vec_mradds (vector signed short,
8692 vector signed short,
8693 vector signed short);
8695 vector unsigned int vec_msum (vector unsigned char,
8696 vector unsigned char,
8697 vector unsigned int);
8698 vector signed int vec_msum (vector signed char,
8699 vector unsigned char,
8701 vector unsigned int vec_msum (vector unsigned short,
8702 vector unsigned short,
8703 vector unsigned int);
8704 vector signed int vec_msum (vector signed short,
8705 vector signed short,
8708 vector signed int vec_vmsumshm (vector signed short,
8709 vector signed short,
8712 vector unsigned int vec_vmsumuhm (vector unsigned short,
8713 vector unsigned short,
8714 vector unsigned int);
8716 vector signed int vec_vmsummbm (vector signed char,
8717 vector unsigned char,
8720 vector unsigned int vec_vmsumubm (vector unsigned char,
8721 vector unsigned char,
8722 vector unsigned int);
8724 vector unsigned int vec_msums (vector unsigned short,
8725 vector unsigned short,
8726 vector unsigned int);
8727 vector signed int vec_msums (vector signed short,
8728 vector signed short,
8731 vector signed int vec_vmsumshs (vector signed short,
8732 vector signed short,
8735 vector unsigned int vec_vmsumuhs (vector unsigned short,
8736 vector unsigned short,
8737 vector unsigned int);
8739 void vec_mtvscr (vector signed int);
8740 void vec_mtvscr (vector unsigned int);
8741 void vec_mtvscr (vector bool int);
8742 void vec_mtvscr (vector signed short);
8743 void vec_mtvscr (vector unsigned short);
8744 void vec_mtvscr (vector bool short);
8745 void vec_mtvscr (vector pixel);
8746 void vec_mtvscr (vector signed char);
8747 void vec_mtvscr (vector unsigned char);
8748 void vec_mtvscr (vector bool char);
8750 vector unsigned short vec_mule (vector unsigned char,
8751 vector unsigned char);
8752 vector signed short vec_mule (vector signed char,
8753 vector signed char);
8754 vector unsigned int vec_mule (vector unsigned short,
8755 vector unsigned short);
8756 vector signed int vec_mule (vector signed short, vector signed short);
8758 vector signed int vec_vmulesh (vector signed short,
8759 vector signed short);
8761 vector unsigned int vec_vmuleuh (vector unsigned short,
8762 vector unsigned short);
8764 vector signed short vec_vmulesb (vector signed char,
8765 vector signed char);
8767 vector unsigned short vec_vmuleub (vector unsigned char,
8768 vector unsigned char);
8770 vector unsigned short vec_mulo (vector unsigned char,
8771 vector unsigned char);
8772 vector signed short vec_mulo (vector signed char, vector signed char);
8773 vector unsigned int vec_mulo (vector unsigned short,
8774 vector unsigned short);
8775 vector signed int vec_mulo (vector signed short, vector signed short);
8777 vector signed int vec_vmulosh (vector signed short,
8778 vector signed short);
8780 vector unsigned int vec_vmulouh (vector unsigned short,
8781 vector unsigned short);
8783 vector signed short vec_vmulosb (vector signed char,
8784 vector signed char);
8786 vector unsigned short vec_vmuloub (vector unsigned char,
8787 vector unsigned char);
8789 vector float vec_nmsub (vector float, vector float, vector float);
8791 vector float vec_nor (vector float, vector float);
8792 vector signed int vec_nor (vector signed int, vector signed int);
8793 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
8794 vector bool int vec_nor (vector bool int, vector bool int);
8795 vector signed short vec_nor (vector signed short, vector signed short);
8796 vector unsigned short vec_nor (vector unsigned short,
8797 vector unsigned short);
8798 vector bool short vec_nor (vector bool short, vector bool short);
8799 vector signed char vec_nor (vector signed char, vector signed char);
8800 vector unsigned char vec_nor (vector unsigned char,
8801 vector unsigned char);
8802 vector bool char vec_nor (vector bool char, vector bool char);
8804 vector float vec_or (vector float, vector float);
8805 vector float vec_or (vector float, vector bool int);
8806 vector float vec_or (vector bool int, vector float);
8807 vector bool int vec_or (vector bool int, vector bool int);
8808 vector signed int vec_or (vector bool int, vector signed int);
8809 vector signed int vec_or (vector signed int, vector bool int);
8810 vector signed int vec_or (vector signed int, vector signed int);
8811 vector unsigned int vec_or (vector bool int, vector unsigned int);
8812 vector unsigned int vec_or (vector unsigned int, vector bool int);
8813 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
8814 vector bool short vec_or (vector bool short, vector bool short);
8815 vector signed short vec_or (vector bool short, vector signed short);
8816 vector signed short vec_or (vector signed short, vector bool short);
8817 vector signed short vec_or (vector signed short, vector signed short);
8818 vector unsigned short vec_or (vector bool short, vector unsigned short);
8819 vector unsigned short vec_or (vector unsigned short, vector bool short);
8820 vector unsigned short vec_or (vector unsigned short,
8821 vector unsigned short);
8822 vector signed char vec_or (vector bool char, vector signed char);
8823 vector bool char vec_or (vector bool char, vector bool char);
8824 vector signed char vec_or (vector signed char, vector bool char);
8825 vector signed char vec_or (vector signed char, vector signed char);
8826 vector unsigned char vec_or (vector bool char, vector unsigned char);
8827 vector unsigned char vec_or (vector unsigned char, vector bool char);
8828 vector unsigned char vec_or (vector unsigned char,
8829 vector unsigned char);
8831 vector signed char vec_pack (vector signed short, vector signed short);
8832 vector unsigned char vec_pack (vector unsigned short,
8833 vector unsigned short);
8834 vector bool char vec_pack (vector bool short, vector bool short);
8835 vector signed short vec_pack (vector signed int, vector signed int);
8836 vector unsigned short vec_pack (vector unsigned int,
8837 vector unsigned int);
8838 vector bool short vec_pack (vector bool int, vector bool int);
8840 vector bool short vec_vpkuwum (vector bool int, vector bool int);
8841 vector signed short vec_vpkuwum (vector signed int, vector signed int);
8842 vector unsigned short vec_vpkuwum (vector unsigned int,
8843 vector unsigned int);
8845 vector bool char vec_vpkuhum (vector bool short, vector bool short);
8846 vector signed char vec_vpkuhum (vector signed short,
8847 vector signed short);
8848 vector unsigned char vec_vpkuhum (vector unsigned short,
8849 vector unsigned short);
8851 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
8853 vector unsigned char vec_packs (vector unsigned short,
8854 vector unsigned short);
8855 vector signed char vec_packs (vector signed short, vector signed short);
8856 vector unsigned short vec_packs (vector unsigned int,
8857 vector unsigned int);
8858 vector signed short vec_packs (vector signed int, vector signed int);
8860 vector signed short vec_vpkswss (vector signed int, vector signed int);
8862 vector unsigned short vec_vpkuwus (vector unsigned int,
8863 vector unsigned int);
8865 vector signed char vec_vpkshss (vector signed short,
8866 vector signed short);
8868 vector unsigned char vec_vpkuhus (vector unsigned short,
8869 vector unsigned short);
8871 vector unsigned char vec_packsu (vector unsigned short,
8872 vector unsigned short);
8873 vector unsigned char vec_packsu (vector signed short,
8874 vector signed short);
8875 vector unsigned short vec_packsu (vector unsigned int,
8876 vector unsigned int);
8877 vector unsigned short vec_packsu (vector signed int, vector signed int);
8879 vector unsigned short vec_vpkswus (vector signed int,
8882 vector unsigned char vec_vpkshus (vector signed short,
8883 vector signed short);
8885 vector float vec_perm (vector float,
8887 vector unsigned char);
8888 vector signed int vec_perm (vector signed int,
8890 vector unsigned char);
8891 vector unsigned int vec_perm (vector unsigned int,
8892 vector unsigned int,
8893 vector unsigned char);
8894 vector bool int vec_perm (vector bool int,
8896 vector unsigned char);
8897 vector signed short vec_perm (vector signed short,
8898 vector signed short,
8899 vector unsigned char);
8900 vector unsigned short vec_perm (vector unsigned short,
8901 vector unsigned short,
8902 vector unsigned char);
8903 vector bool short vec_perm (vector bool short,
8905 vector unsigned char);
8906 vector pixel vec_perm (vector pixel,
8908 vector unsigned char);
8909 vector signed char vec_perm (vector signed char,
8911 vector unsigned char);
8912 vector unsigned char vec_perm (vector unsigned char,
8913 vector unsigned char,
8914 vector unsigned char);
8915 vector bool char vec_perm (vector bool char,
8917 vector unsigned char);
8919 vector float vec_re (vector float);
8921 vector signed char vec_rl (vector signed char,
8922 vector unsigned char);
8923 vector unsigned char vec_rl (vector unsigned char,
8924 vector unsigned char);
8925 vector signed short vec_rl (vector signed short, vector unsigned short);
8926 vector unsigned short vec_rl (vector unsigned short,
8927 vector unsigned short);
8928 vector signed int vec_rl (vector signed int, vector unsigned int);
8929 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
8931 vector signed int vec_vrlw (vector signed int, vector unsigned int);
8932 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
8934 vector signed short vec_vrlh (vector signed short,
8935 vector unsigned short);
8936 vector unsigned short vec_vrlh (vector unsigned short,
8937 vector unsigned short);
8939 vector signed char vec_vrlb (vector signed char, vector unsigned char);
8940 vector unsigned char vec_vrlb (vector unsigned char,
8941 vector unsigned char);
8943 vector float vec_round (vector float);
8945 vector float vec_rsqrte (vector float);
8947 vector float vec_sel (vector float, vector float, vector bool int);
8948 vector float vec_sel (vector float, vector float, vector unsigned int);
8949 vector signed int vec_sel (vector signed int,
8952 vector signed int vec_sel (vector signed int,
8954 vector unsigned int);
8955 vector unsigned int vec_sel (vector unsigned int,
8956 vector unsigned int,
8958 vector unsigned int vec_sel (vector unsigned int,
8959 vector unsigned int,
8960 vector unsigned int);
8961 vector bool int vec_sel (vector bool int,
8964 vector bool int vec_sel (vector bool int,
8966 vector unsigned int);
8967 vector signed short vec_sel (vector signed short,
8968 vector signed short,
8970 vector signed short vec_sel (vector signed short,
8971 vector signed short,
8972 vector unsigned short);
8973 vector unsigned short vec_sel (vector unsigned short,
8974 vector unsigned short,
8976 vector unsigned short vec_sel (vector unsigned short,
8977 vector unsigned short,
8978 vector unsigned short);
8979 vector bool short vec_sel (vector bool short,
8982 vector bool short vec_sel (vector bool short,
8984 vector unsigned short);
8985 vector signed char vec_sel (vector signed char,
8988 vector signed char vec_sel (vector signed char,
8990 vector unsigned char);
8991 vector unsigned char vec_sel (vector unsigned char,
8992 vector unsigned char,
8994 vector unsigned char vec_sel (vector unsigned char,
8995 vector unsigned char,
8996 vector unsigned char);
8997 vector bool char vec_sel (vector bool char,
9000 vector bool char vec_sel (vector bool char,
9002 vector unsigned char);
9004 vector signed char vec_sl (vector signed char,
9005 vector unsigned char);
9006 vector unsigned char vec_sl (vector unsigned char,
9007 vector unsigned char);
9008 vector signed short vec_sl (vector signed short, vector unsigned short);
9009 vector unsigned short vec_sl (vector unsigned short,
9010 vector unsigned short);
9011 vector signed int vec_sl (vector signed int, vector unsigned int);
9012 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
9014 vector signed int vec_vslw (vector signed int, vector unsigned int);
9015 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
9017 vector signed short vec_vslh (vector signed short,
9018 vector unsigned short);
9019 vector unsigned short vec_vslh (vector unsigned short,
9020 vector unsigned short);
9022 vector signed char vec_vslb (vector signed char, vector unsigned char);
9023 vector unsigned char vec_vslb (vector unsigned char,
9024 vector unsigned char);
9026 vector float vec_sld (vector float, vector float, const int);
9027 vector signed int vec_sld (vector signed int,
9030 vector unsigned int vec_sld (vector unsigned int,
9031 vector unsigned int,
9033 vector bool int vec_sld (vector bool int,
9036 vector signed short vec_sld (vector signed short,
9037 vector signed short,
9039 vector unsigned short vec_sld (vector unsigned short,
9040 vector unsigned short,
9042 vector bool short vec_sld (vector bool short,
9045 vector pixel vec_sld (vector pixel,
9048 vector signed char vec_sld (vector signed char,
9051 vector unsigned char vec_sld (vector unsigned char,
9052 vector unsigned char,
9054 vector bool char vec_sld (vector bool char,
9058 vector signed int vec_sll (vector signed int,
9059 vector unsigned int);
9060 vector signed int vec_sll (vector signed int,
9061 vector unsigned short);
9062 vector signed int vec_sll (vector signed int,
9063 vector unsigned char);
9064 vector unsigned int vec_sll (vector unsigned int,
9065 vector unsigned int);
9066 vector unsigned int vec_sll (vector unsigned int,
9067 vector unsigned short);
9068 vector unsigned int vec_sll (vector unsigned int,
9069 vector unsigned char);
9070 vector bool int vec_sll (vector bool int,
9071 vector unsigned int);
9072 vector bool int vec_sll (vector bool int,
9073 vector unsigned short);
9074 vector bool int vec_sll (vector bool int,
9075 vector unsigned char);
9076 vector signed short vec_sll (vector signed short,
9077 vector unsigned int);
9078 vector signed short vec_sll (vector signed short,
9079 vector unsigned short);
9080 vector signed short vec_sll (vector signed short,
9081 vector unsigned char);
9082 vector unsigned short vec_sll (vector unsigned short,
9083 vector unsigned int);
9084 vector unsigned short vec_sll (vector unsigned short,
9085 vector unsigned short);
9086 vector unsigned short vec_sll (vector unsigned short,
9087 vector unsigned char);
9088 vector bool short vec_sll (vector bool short, vector unsigned int);
9089 vector bool short vec_sll (vector bool short, vector unsigned short);
9090 vector bool short vec_sll (vector bool short, vector unsigned char);
9091 vector pixel vec_sll (vector pixel, vector unsigned int);
9092 vector pixel vec_sll (vector pixel, vector unsigned short);
9093 vector pixel vec_sll (vector pixel, vector unsigned char);
9094 vector signed char vec_sll (vector signed char, vector unsigned int);
9095 vector signed char vec_sll (vector signed char, vector unsigned short);
9096 vector signed char vec_sll (vector signed char, vector unsigned char);
9097 vector unsigned char vec_sll (vector unsigned char,
9098 vector unsigned int);
9099 vector unsigned char vec_sll (vector unsigned char,
9100 vector unsigned short);
9101 vector unsigned char vec_sll (vector unsigned char,
9102 vector unsigned char);
9103 vector bool char vec_sll (vector bool char, vector unsigned int);
9104 vector bool char vec_sll (vector bool char, vector unsigned short);
9105 vector bool char vec_sll (vector bool char, vector unsigned char);
9107 vector float vec_slo (vector float, vector signed char);
9108 vector float vec_slo (vector float, vector unsigned char);
9109 vector signed int vec_slo (vector signed int, vector signed char);
9110 vector signed int vec_slo (vector signed int, vector unsigned char);
9111 vector unsigned int vec_slo (vector unsigned int, vector signed char);
9112 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
9113 vector signed short vec_slo (vector signed short, vector signed char);
9114 vector signed short vec_slo (vector signed short, vector unsigned char);
9115 vector unsigned short vec_slo (vector unsigned short,
9116 vector signed char);
9117 vector unsigned short vec_slo (vector unsigned short,
9118 vector unsigned char);
9119 vector pixel vec_slo (vector pixel, vector signed char);
9120 vector pixel vec_slo (vector pixel, vector unsigned char);
9121 vector signed char vec_slo (vector signed char, vector signed char);
9122 vector signed char vec_slo (vector signed char, vector unsigned char);
9123 vector unsigned char vec_slo (vector unsigned char, vector signed char);
9124 vector unsigned char vec_slo (vector unsigned char,
9125 vector unsigned char);
9127 vector signed char vec_splat (vector signed char, const int);
9128 vector unsigned char vec_splat (vector unsigned char, const int);
9129 vector bool char vec_splat (vector bool char, const int);
9130 vector signed short vec_splat (vector signed short, const int);
9131 vector unsigned short vec_splat (vector unsigned short, const int);
9132 vector bool short vec_splat (vector bool short, const int);
9133 vector pixel vec_splat (vector pixel, const int);
9134 vector float vec_splat (vector float, const int);
9135 vector signed int vec_splat (vector signed int, const int);
9136 vector unsigned int vec_splat (vector unsigned int, const int);
9137 vector bool int vec_splat (vector bool int, const int);
9139 vector float vec_vspltw (vector float, const int);
9140 vector signed int vec_vspltw (vector signed int, const int);
9141 vector unsigned int vec_vspltw (vector unsigned int, const int);
9142 vector bool int vec_vspltw (vector bool int, const int);
9144 vector bool short vec_vsplth (vector bool short, const int);
9145 vector signed short vec_vsplth (vector signed short, const int);
9146 vector unsigned short vec_vsplth (vector unsigned short, const int);
9147 vector pixel vec_vsplth (vector pixel, const int);
9149 vector signed char vec_vspltb (vector signed char, const int);
9150 vector unsigned char vec_vspltb (vector unsigned char, const int);
9151 vector bool char vec_vspltb (vector bool char, const int);
9153 vector signed char vec_splat_s8 (const int);
9155 vector signed short vec_splat_s16 (const int);
9157 vector signed int vec_splat_s32 (const int);
9159 vector unsigned char vec_splat_u8 (const int);
9161 vector unsigned short vec_splat_u16 (const int);
9163 vector unsigned int vec_splat_u32 (const int);
9165 vector signed char vec_sr (vector signed char, vector unsigned char);
9166 vector unsigned char vec_sr (vector unsigned char,
9167 vector unsigned char);
9168 vector signed short vec_sr (vector signed short,
9169 vector unsigned short);
9170 vector unsigned short vec_sr (vector unsigned short,
9171 vector unsigned short);
9172 vector signed int vec_sr (vector signed int, vector unsigned int);
9173 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
9175 vector signed int vec_vsrw (vector signed int, vector unsigned int);
9176 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
9178 vector signed short vec_vsrh (vector signed short,
9179 vector unsigned short);
9180 vector unsigned short vec_vsrh (vector unsigned short,
9181 vector unsigned short);
9183 vector signed char vec_vsrb (vector signed char, vector unsigned char);
9184 vector unsigned char vec_vsrb (vector unsigned char,
9185 vector unsigned char);
9187 vector signed char vec_sra (vector signed char, vector unsigned char);
9188 vector unsigned char vec_sra (vector unsigned char,
9189 vector unsigned char);
9190 vector signed short vec_sra (vector signed short,
9191 vector unsigned short);
9192 vector unsigned short vec_sra (vector unsigned short,
9193 vector unsigned short);
9194 vector signed int vec_sra (vector signed int, vector unsigned int);
9195 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
9197 vector signed int vec_vsraw (vector signed int, vector unsigned int);
9198 vector unsigned int vec_vsraw (vector unsigned int,
9199 vector unsigned int);
9201 vector signed short vec_vsrah (vector signed short,
9202 vector unsigned short);
9203 vector unsigned short vec_vsrah (vector unsigned short,
9204 vector unsigned short);
9206 vector signed char vec_vsrab (vector signed char, vector unsigned char);
9207 vector unsigned char vec_vsrab (vector unsigned char,
9208 vector unsigned char);
9210 vector signed int vec_srl (vector signed int, vector unsigned int);
9211 vector signed int vec_srl (vector signed int, vector unsigned short);
9212 vector signed int vec_srl (vector signed int, vector unsigned char);
9213 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
9214 vector unsigned int vec_srl (vector unsigned int,
9215 vector unsigned short);
9216 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
9217 vector bool int vec_srl (vector bool int, vector unsigned int);
9218 vector bool int vec_srl (vector bool int, vector unsigned short);
9219 vector bool int vec_srl (vector bool int, vector unsigned char);
9220 vector signed short vec_srl (vector signed short, vector unsigned int);
9221 vector signed short vec_srl (vector signed short,
9222 vector unsigned short);
9223 vector signed short vec_srl (vector signed short, vector unsigned char);
9224 vector unsigned short vec_srl (vector unsigned short,
9225 vector unsigned int);
9226 vector unsigned short vec_srl (vector unsigned short,
9227 vector unsigned short);
9228 vector unsigned short vec_srl (vector unsigned short,
9229 vector unsigned char);
9230 vector bool short vec_srl (vector bool short, vector unsigned int);
9231 vector bool short vec_srl (vector bool short, vector unsigned short);
9232 vector bool short vec_srl (vector bool short, vector unsigned char);
9233 vector pixel vec_srl (vector pixel, vector unsigned int);
9234 vector pixel vec_srl (vector pixel, vector unsigned short);
9235 vector pixel vec_srl (vector pixel, vector unsigned char);
9236 vector signed char vec_srl (vector signed char, vector unsigned int);
9237 vector signed char vec_srl (vector signed char, vector unsigned short);
9238 vector signed char vec_srl (vector signed char, vector unsigned char);
9239 vector unsigned char vec_srl (vector unsigned char,
9240 vector unsigned int);
9241 vector unsigned char vec_srl (vector unsigned char,
9242 vector unsigned short);
9243 vector unsigned char vec_srl (vector unsigned char,
9244 vector unsigned char);
9245 vector bool char vec_srl (vector bool char, vector unsigned int);
9246 vector bool char vec_srl (vector bool char, vector unsigned short);
9247 vector bool char vec_srl (vector bool char, vector unsigned char);
9249 vector float vec_sro (vector float, vector signed char);
9250 vector float vec_sro (vector float, vector unsigned char);
9251 vector signed int vec_sro (vector signed int, vector signed char);
9252 vector signed int vec_sro (vector signed int, vector unsigned char);
9253 vector unsigned int vec_sro (vector unsigned int, vector signed char);
9254 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
9255 vector signed short vec_sro (vector signed short, vector signed char);
9256 vector signed short vec_sro (vector signed short, vector unsigned char);
9257 vector unsigned short vec_sro (vector unsigned short,
9258 vector signed char);
9259 vector unsigned short vec_sro (vector unsigned short,
9260 vector unsigned char);
9261 vector pixel vec_sro (vector pixel, vector signed char);
9262 vector pixel vec_sro (vector pixel, vector unsigned char);
9263 vector signed char vec_sro (vector signed char, vector signed char);
9264 vector signed char vec_sro (vector signed char, vector unsigned char);
9265 vector unsigned char vec_sro (vector unsigned char, vector signed char);
9266 vector unsigned char vec_sro (vector unsigned char,
9267 vector unsigned char);
9269 void vec_st (vector float, int, vector float *);
9270 void vec_st (vector float, int, float *);
9271 void vec_st (vector signed int, int, vector signed int *);
9272 void vec_st (vector signed int, int, int *);
9273 void vec_st (vector unsigned int, int, vector unsigned int *);
9274 void vec_st (vector unsigned int, int, unsigned int *);
9275 void vec_st (vector bool int, int, vector bool int *);
9276 void vec_st (vector bool int, int, unsigned int *);
9277 void vec_st (vector bool int, int, int *);
9278 void vec_st (vector signed short, int, vector signed short *);
9279 void vec_st (vector signed short, int, short *);
9280 void vec_st (vector unsigned short, int, vector unsigned short *);
9281 void vec_st (vector unsigned short, int, unsigned short *);
9282 void vec_st (vector bool short, int, vector bool short *);
9283 void vec_st (vector bool short, int, unsigned short *);
9284 void vec_st (vector pixel, int, vector pixel *);
9285 void vec_st (vector pixel, int, unsigned short *);
9286 void vec_st (vector pixel, int, short *);
9287 void vec_st (vector bool short, int, short *);
9288 void vec_st (vector signed char, int, vector signed char *);
9289 void vec_st (vector signed char, int, signed char *);
9290 void vec_st (vector unsigned char, int, vector unsigned char *);
9291 void vec_st (vector unsigned char, int, unsigned char *);
9292 void vec_st (vector bool char, int, vector bool char *);
9293 void vec_st (vector bool char, int, unsigned char *);
9294 void vec_st (vector bool char, int, signed char *);
9296 void vec_ste (vector signed char, int, signed char *);
9297 void vec_ste (vector unsigned char, int, unsigned char *);
9298 void vec_ste (vector bool char, int, signed char *);
9299 void vec_ste (vector bool char, int, unsigned char *);
9300 void vec_ste (vector signed short, int, short *);
9301 void vec_ste (vector unsigned short, int, unsigned short *);
9302 void vec_ste (vector bool short, int, short *);
9303 void vec_ste (vector bool short, int, unsigned short *);
9304 void vec_ste (vector pixel, int, short *);
9305 void vec_ste (vector pixel, int, unsigned short *);
9306 void vec_ste (vector float, int, float *);
9307 void vec_ste (vector signed int, int, int *);
9308 void vec_ste (vector unsigned int, int, unsigned int *);
9309 void vec_ste (vector bool int, int, int *);
9310 void vec_ste (vector bool int, int, unsigned int *);
9312 void vec_stvewx (vector float, int, float *);
9313 void vec_stvewx (vector signed int, int, int *);
9314 void vec_stvewx (vector unsigned int, int, unsigned int *);
9315 void vec_stvewx (vector bool int, int, int *);
9316 void vec_stvewx (vector bool int, int, unsigned int *);
9318 void vec_stvehx (vector signed short, int, short *);
9319 void vec_stvehx (vector unsigned short, int, unsigned short *);
9320 void vec_stvehx (vector bool short, int, short *);
9321 void vec_stvehx (vector bool short, int, unsigned short *);
9322 void vec_stvehx (vector pixel, int, short *);
9323 void vec_stvehx (vector pixel, int, unsigned short *);
9325 void vec_stvebx (vector signed char, int, signed char *);
9326 void vec_stvebx (vector unsigned char, int, unsigned char *);
9327 void vec_stvebx (vector bool char, int, signed char *);
9328 void vec_stvebx (vector bool char, int, unsigned char *);
9330 void vec_stl (vector float, int, vector float *);
9331 void vec_stl (vector float, int, float *);
9332 void vec_stl (vector signed int, int, vector signed int *);
9333 void vec_stl (vector signed int, int, int *);
9334 void vec_stl (vector unsigned int, int, vector unsigned int *);
9335 void vec_stl (vector unsigned int, int, unsigned int *);
9336 void vec_stl (vector bool int, int, vector bool int *);
9337 void vec_stl (vector bool int, int, unsigned int *);
9338 void vec_stl (vector bool int, int, int *);
9339 void vec_stl (vector signed short, int, vector signed short *);
9340 void vec_stl (vector signed short, int, short *);
9341 void vec_stl (vector unsigned short, int, vector unsigned short *);
9342 void vec_stl (vector unsigned short, int, unsigned short *);
9343 void vec_stl (vector bool short, int, vector bool short *);
9344 void vec_stl (vector bool short, int, unsigned short *);
9345 void vec_stl (vector bool short, int, short *);
9346 void vec_stl (vector pixel, int, vector pixel *);
9347 void vec_stl (vector pixel, int, unsigned short *);
9348 void vec_stl (vector pixel, int, short *);
9349 void vec_stl (vector signed char, int, vector signed char *);
9350 void vec_stl (vector signed char, int, signed char *);
9351 void vec_stl (vector unsigned char, int, vector unsigned char *);
9352 void vec_stl (vector unsigned char, int, unsigned char *);
9353 void vec_stl (vector bool char, int, vector bool char *);
9354 void vec_stl (vector bool char, int, unsigned char *);
9355 void vec_stl (vector bool char, int, signed char *);
9357 vector signed char vec_sub (vector bool char, vector signed char);
9358 vector signed char vec_sub (vector signed char, vector bool char);
9359 vector signed char vec_sub (vector signed char, vector signed char);
9360 vector unsigned char vec_sub (vector bool char, vector unsigned char);
9361 vector unsigned char vec_sub (vector unsigned char, vector bool char);
9362 vector unsigned char vec_sub (vector unsigned char,
9363 vector unsigned char);
9364 vector signed short vec_sub (vector bool short, vector signed short);
9365 vector signed short vec_sub (vector signed short, vector bool short);
9366 vector signed short vec_sub (vector signed short, vector signed short);
9367 vector unsigned short vec_sub (vector bool short,
9368 vector unsigned short);
9369 vector unsigned short vec_sub (vector unsigned short,
9371 vector unsigned short vec_sub (vector unsigned short,
9372 vector unsigned short);
9373 vector signed int vec_sub (vector bool int, vector signed int);
9374 vector signed int vec_sub (vector signed int, vector bool int);
9375 vector signed int vec_sub (vector signed int, vector signed int);
9376 vector unsigned int vec_sub (vector bool int, vector unsigned int);
9377 vector unsigned int vec_sub (vector unsigned int, vector bool int);
9378 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
9379 vector float vec_sub (vector float, vector float);
9381 vector float vec_vsubfp (vector float, vector float);
9383 vector signed int vec_vsubuwm (vector bool int, vector signed int);
9384 vector signed int vec_vsubuwm (vector signed int, vector bool int);
9385 vector signed int vec_vsubuwm (vector signed int, vector signed int);
9386 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
9387 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
9388 vector unsigned int vec_vsubuwm (vector unsigned int,
9389 vector unsigned int);
9391 vector signed short vec_vsubuhm (vector bool short,
9392 vector signed short);
9393 vector signed short vec_vsubuhm (vector signed short,
9395 vector signed short vec_vsubuhm (vector signed short,
9396 vector signed short);
9397 vector unsigned short vec_vsubuhm (vector bool short,
9398 vector unsigned short);
9399 vector unsigned short vec_vsubuhm (vector unsigned short,
9401 vector unsigned short vec_vsubuhm (vector unsigned short,
9402 vector unsigned short);
9404 vector signed char vec_vsububm (vector bool char, vector signed char);
9405 vector signed char vec_vsububm (vector signed char, vector bool char);
9406 vector signed char vec_vsububm (vector signed char, vector signed char);
9407 vector unsigned char vec_vsububm (vector bool char,
9408 vector unsigned char);
9409 vector unsigned char vec_vsububm (vector unsigned char,
9411 vector unsigned char vec_vsububm (vector unsigned char,
9412 vector unsigned char);
9414 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
9416 vector unsigned char vec_subs (vector bool char, vector unsigned char);
9417 vector unsigned char vec_subs (vector unsigned char, vector bool char);
9418 vector unsigned char vec_subs (vector unsigned char,
9419 vector unsigned char);
9420 vector signed char vec_subs (vector bool char, vector signed char);
9421 vector signed char vec_subs (vector signed char, vector bool char);
9422 vector signed char vec_subs (vector signed char, vector signed char);
9423 vector unsigned short vec_subs (vector bool short,
9424 vector unsigned short);
9425 vector unsigned short vec_subs (vector unsigned short,
9427 vector unsigned short vec_subs (vector unsigned short,
9428 vector unsigned short);
9429 vector signed short vec_subs (vector bool short, vector signed short);
9430 vector signed short vec_subs (vector signed short, vector bool short);
9431 vector signed short vec_subs (vector signed short, vector signed short);
9432 vector unsigned int vec_subs (vector bool int, vector unsigned int);
9433 vector unsigned int vec_subs (vector unsigned int, vector bool int);
9434 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
9435 vector signed int vec_subs (vector bool int, vector signed int);
9436 vector signed int vec_subs (vector signed int, vector bool int);
9437 vector signed int vec_subs (vector signed int, vector signed int);
9439 vector signed int vec_vsubsws (vector bool int, vector signed int);
9440 vector signed int vec_vsubsws (vector signed int, vector bool int);
9441 vector signed int vec_vsubsws (vector signed int, vector signed int);
9443 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
9444 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
9445 vector unsigned int vec_vsubuws (vector unsigned int,
9446 vector unsigned int);
9448 vector signed short vec_vsubshs (vector bool short,
9449 vector signed short);
9450 vector signed short vec_vsubshs (vector signed short,
9452 vector signed short vec_vsubshs (vector signed short,
9453 vector signed short);
9455 vector unsigned short vec_vsubuhs (vector bool short,
9456 vector unsigned short);
9457 vector unsigned short vec_vsubuhs (vector unsigned short,
9459 vector unsigned short vec_vsubuhs (vector unsigned short,
9460 vector unsigned short);
9462 vector signed char vec_vsubsbs (vector bool char, vector signed char);
9463 vector signed char vec_vsubsbs (vector signed char, vector bool char);
9464 vector signed char vec_vsubsbs (vector signed char, vector signed char);
9466 vector unsigned char vec_vsububs (vector bool char,
9467 vector unsigned char);
9468 vector unsigned char vec_vsububs (vector unsigned char,
9470 vector unsigned char vec_vsububs (vector unsigned char,
9471 vector unsigned char);
9473 vector unsigned int vec_sum4s (vector unsigned char,
9474 vector unsigned int);
9475 vector signed int vec_sum4s (vector signed char, vector signed int);
9476 vector signed int vec_sum4s (vector signed short, vector signed int);
9478 vector signed int vec_vsum4shs (vector signed short, vector signed int);
9480 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
9482 vector unsigned int vec_vsum4ubs (vector unsigned char,
9483 vector unsigned int);
9485 vector signed int vec_sum2s (vector signed int, vector signed int);
9487 vector signed int vec_sums (vector signed int, vector signed int);
9489 vector float vec_trunc (vector float);
9491 vector signed short vec_unpackh (vector signed char);
9492 vector bool short vec_unpackh (vector bool char);
9493 vector signed int vec_unpackh (vector signed short);
9494 vector bool int vec_unpackh (vector bool short);
9495 vector unsigned int vec_unpackh (vector pixel);
9497 vector bool int vec_vupkhsh (vector bool short);
9498 vector signed int vec_vupkhsh (vector signed short);
9500 vector unsigned int vec_vupkhpx (vector pixel);
9502 vector bool short vec_vupkhsb (vector bool char);
9503 vector signed short vec_vupkhsb (vector signed char);
9505 vector signed short vec_unpackl (vector signed char);
9506 vector bool short vec_unpackl (vector bool char);
9507 vector unsigned int vec_unpackl (vector pixel);
9508 vector signed int vec_unpackl (vector signed short);
9509 vector bool int vec_unpackl (vector bool short);
9511 vector unsigned int vec_vupklpx (vector pixel);
9513 vector bool int vec_vupklsh (vector bool short);
9514 vector signed int vec_vupklsh (vector signed short);
9516 vector bool short vec_vupklsb (vector bool char);
9517 vector signed short vec_vupklsb (vector signed char);
9519 vector float vec_xor (vector float, vector float);
9520 vector float vec_xor (vector float, vector bool int);
9521 vector float vec_xor (vector bool int, vector float);
9522 vector bool int vec_xor (vector bool int, vector bool int);
9523 vector signed int vec_xor (vector bool int, vector signed int);
9524 vector signed int vec_xor (vector signed int, vector bool int);
9525 vector signed int vec_xor (vector signed int, vector signed int);
9526 vector unsigned int vec_xor (vector bool int, vector unsigned int);
9527 vector unsigned int vec_xor (vector unsigned int, vector bool int);
9528 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
9529 vector bool short vec_xor (vector bool short, vector bool short);
9530 vector signed short vec_xor (vector bool short, vector signed short);
9531 vector signed short vec_xor (vector signed short, vector bool short);
9532 vector signed short vec_xor (vector signed short, vector signed short);
9533 vector unsigned short vec_xor (vector bool short,
9534 vector unsigned short);
9535 vector unsigned short vec_xor (vector unsigned short,
9537 vector unsigned short vec_xor (vector unsigned short,
9538 vector unsigned short);
9539 vector signed char vec_xor (vector bool char, vector signed char);
9540 vector bool char vec_xor (vector bool char, vector bool char);
9541 vector signed char vec_xor (vector signed char, vector bool char);
9542 vector signed char vec_xor (vector signed char, vector signed char);
9543 vector unsigned char vec_xor (vector bool char, vector unsigned char);
9544 vector unsigned char vec_xor (vector unsigned char, vector bool char);
9545 vector unsigned char vec_xor (vector unsigned char,
9546 vector unsigned char);
9548 int vec_all_eq (vector signed char, vector bool char);
9549 int vec_all_eq (vector signed char, vector signed char);
9550 int vec_all_eq (vector unsigned char, vector bool char);
9551 int vec_all_eq (vector unsigned char, vector unsigned char);
9552 int vec_all_eq (vector bool char, vector bool char);
9553 int vec_all_eq (vector bool char, vector unsigned char);
9554 int vec_all_eq (vector bool char, vector signed char);
9555 int vec_all_eq (vector signed short, vector bool short);
9556 int vec_all_eq (vector signed short, vector signed short);
9557 int vec_all_eq (vector unsigned short, vector bool short);
9558 int vec_all_eq (vector unsigned short, vector unsigned short);
9559 int vec_all_eq (vector bool short, vector bool short);
9560 int vec_all_eq (vector bool short, vector unsigned short);
9561 int vec_all_eq (vector bool short, vector signed short);
9562 int vec_all_eq (vector pixel, vector pixel);
9563 int vec_all_eq (vector signed int, vector bool int);
9564 int vec_all_eq (vector signed int, vector signed int);
9565 int vec_all_eq (vector unsigned int, vector bool int);
9566 int vec_all_eq (vector unsigned int, vector unsigned int);
9567 int vec_all_eq (vector bool int, vector bool int);
9568 int vec_all_eq (vector bool int, vector unsigned int);
9569 int vec_all_eq (vector bool int, vector signed int);
9570 int vec_all_eq (vector float, vector float);
9572 int vec_all_ge (vector bool char, vector unsigned char);
9573 int vec_all_ge (vector unsigned char, vector bool char);
9574 int vec_all_ge (vector unsigned char, vector unsigned char);
9575 int vec_all_ge (vector bool char, vector signed char);
9576 int vec_all_ge (vector signed char, vector bool char);
9577 int vec_all_ge (vector signed char, vector signed char);
9578 int vec_all_ge (vector bool short, vector unsigned short);
9579 int vec_all_ge (vector unsigned short, vector bool short);
9580 int vec_all_ge (vector unsigned short, vector unsigned short);
9581 int vec_all_ge (vector signed short, vector signed short);
9582 int vec_all_ge (vector bool short, vector signed short);
9583 int vec_all_ge (vector signed short, vector bool short);
9584 int vec_all_ge (vector bool int, vector unsigned int);
9585 int vec_all_ge (vector unsigned int, vector bool int);
9586 int vec_all_ge (vector unsigned int, vector unsigned int);
9587 int vec_all_ge (vector bool int, vector signed int);
9588 int vec_all_ge (vector signed int, vector bool int);
9589 int vec_all_ge (vector signed int, vector signed int);
9590 int vec_all_ge (vector float, vector float);
9592 int vec_all_gt (vector bool char, vector unsigned char);
9593 int vec_all_gt (vector unsigned char, vector bool char);
9594 int vec_all_gt (vector unsigned char, vector unsigned char);
9595 int vec_all_gt (vector bool char, vector signed char);
9596 int vec_all_gt (vector signed char, vector bool char);
9597 int vec_all_gt (vector signed char, vector signed char);
9598 int vec_all_gt (vector bool short, vector unsigned short);
9599 int vec_all_gt (vector unsigned short, vector bool short);
9600 int vec_all_gt (vector unsigned short, vector unsigned short);
9601 int vec_all_gt (vector bool short, vector signed short);
9602 int vec_all_gt (vector signed short, vector bool short);
9603 int vec_all_gt (vector signed short, vector signed short);
9604 int vec_all_gt (vector bool int, vector unsigned int);
9605 int vec_all_gt (vector unsigned int, vector bool int);
9606 int vec_all_gt (vector unsigned int, vector unsigned int);
9607 int vec_all_gt (vector bool int, vector signed int);
9608 int vec_all_gt (vector signed int, vector bool int);
9609 int vec_all_gt (vector signed int, vector signed int);
9610 int vec_all_gt (vector float, vector float);
9612 int vec_all_in (vector float, vector float);
9614 int vec_all_le (vector bool char, vector unsigned char);
9615 int vec_all_le (vector unsigned char, vector bool char);
9616 int vec_all_le (vector unsigned char, vector unsigned char);
9617 int vec_all_le (vector bool char, vector signed char);
9618 int vec_all_le (vector signed char, vector bool char);
9619 int vec_all_le (vector signed char, vector signed char);
9620 int vec_all_le (vector bool short, vector unsigned short);
9621 int vec_all_le (vector unsigned short, vector bool short);
9622 int vec_all_le (vector unsigned short, vector unsigned short);
9623 int vec_all_le (vector bool short, vector signed short);
9624 int vec_all_le (vector signed short, vector bool short);
9625 int vec_all_le (vector signed short, vector signed short);
9626 int vec_all_le (vector bool int, vector unsigned int);
9627 int vec_all_le (vector unsigned int, vector bool int);
9628 int vec_all_le (vector unsigned int, vector unsigned int);
9629 int vec_all_le (vector bool int, vector signed int);
9630 int vec_all_le (vector signed int, vector bool int);
9631 int vec_all_le (vector signed int, vector signed int);
9632 int vec_all_le (vector float, vector float);
9634 int vec_all_lt (vector bool char, vector unsigned char);
9635 int vec_all_lt (vector unsigned char, vector bool char);
9636 int vec_all_lt (vector unsigned char, vector unsigned char);
9637 int vec_all_lt (vector bool char, vector signed char);
9638 int vec_all_lt (vector signed char, vector bool char);
9639 int vec_all_lt (vector signed char, vector signed char);
9640 int vec_all_lt (vector bool short, vector unsigned short);
9641 int vec_all_lt (vector unsigned short, vector bool short);
9642 int vec_all_lt (vector unsigned short, vector unsigned short);
9643 int vec_all_lt (vector bool short, vector signed short);
9644 int vec_all_lt (vector signed short, vector bool short);
9645 int vec_all_lt (vector signed short, vector signed short);
9646 int vec_all_lt (vector bool int, vector unsigned int);
9647 int vec_all_lt (vector unsigned int, vector bool int);
9648 int vec_all_lt (vector unsigned int, vector unsigned int);
9649 int vec_all_lt (vector bool int, vector signed int);
9650 int vec_all_lt (vector signed int, vector bool int);
9651 int vec_all_lt (vector signed int, vector signed int);
9652 int vec_all_lt (vector float, vector float);
9654 int vec_all_nan (vector float);
9656 int vec_all_ne (vector signed char, vector bool char);
9657 int vec_all_ne (vector signed char, vector signed char);
9658 int vec_all_ne (vector unsigned char, vector bool char);
9659 int vec_all_ne (vector unsigned char, vector unsigned char);
9660 int vec_all_ne (vector bool char, vector bool char);
9661 int vec_all_ne (vector bool char, vector unsigned char);
9662 int vec_all_ne (vector bool char, vector signed char);
9663 int vec_all_ne (vector signed short, vector bool short);
9664 int vec_all_ne (vector signed short, vector signed short);
9665 int vec_all_ne (vector unsigned short, vector bool short);
9666 int vec_all_ne (vector unsigned short, vector unsigned short);
9667 int vec_all_ne (vector bool short, vector bool short);
9668 int vec_all_ne (vector bool short, vector unsigned short);
9669 int vec_all_ne (vector bool short, vector signed short);
9670 int vec_all_ne (vector pixel, vector pixel);
9671 int vec_all_ne (vector signed int, vector bool int);
9672 int vec_all_ne (vector signed int, vector signed int);
9673 int vec_all_ne (vector unsigned int, vector bool int);
9674 int vec_all_ne (vector unsigned int, vector unsigned int);
9675 int vec_all_ne (vector bool int, vector bool int);
9676 int vec_all_ne (vector bool int, vector unsigned int);
9677 int vec_all_ne (vector bool int, vector signed int);
9678 int vec_all_ne (vector float, vector float);
9680 int vec_all_nge (vector float, vector float);
9682 int vec_all_ngt (vector float, vector float);
9684 int vec_all_nle (vector float, vector float);
9686 int vec_all_nlt (vector float, vector float);
9688 int vec_all_numeric (vector float);
9690 int vec_any_eq (vector signed char, vector bool char);
9691 int vec_any_eq (vector signed char, vector signed char);
9692 int vec_any_eq (vector unsigned char, vector bool char);
9693 int vec_any_eq (vector unsigned char, vector unsigned char);
9694 int vec_any_eq (vector bool char, vector bool char);
9695 int vec_any_eq (vector bool char, vector unsigned char);
9696 int vec_any_eq (vector bool char, vector signed char);
9697 int vec_any_eq (vector signed short, vector bool short);
9698 int vec_any_eq (vector signed short, vector signed short);
9699 int vec_any_eq (vector unsigned short, vector bool short);
9700 int vec_any_eq (vector unsigned short, vector unsigned short);
9701 int vec_any_eq (vector bool short, vector bool short);
9702 int vec_any_eq (vector bool short, vector unsigned short);
9703 int vec_any_eq (vector bool short, vector signed short);
9704 int vec_any_eq (vector pixel, vector pixel);
9705 int vec_any_eq (vector signed int, vector bool int);
9706 int vec_any_eq (vector signed int, vector signed int);
9707 int vec_any_eq (vector unsigned int, vector bool int);
9708 int vec_any_eq (vector unsigned int, vector unsigned int);
9709 int vec_any_eq (vector bool int, vector bool int);
9710 int vec_any_eq (vector bool int, vector unsigned int);
9711 int vec_any_eq (vector bool int, vector signed int);
9712 int vec_any_eq (vector float, vector float);
9714 int vec_any_ge (vector signed char, vector bool char);
9715 int vec_any_ge (vector unsigned char, vector bool char);
9716 int vec_any_ge (vector unsigned char, vector unsigned char);
9717 int vec_any_ge (vector signed char, vector signed char);
9718 int vec_any_ge (vector bool char, vector unsigned char);
9719 int vec_any_ge (vector bool char, vector signed char);
9720 int vec_any_ge (vector unsigned short, vector bool short);
9721 int vec_any_ge (vector unsigned short, vector unsigned short);
9722 int vec_any_ge (vector signed short, vector signed short);
9723 int vec_any_ge (vector signed short, vector bool short);
9724 int vec_any_ge (vector bool short, vector unsigned short);
9725 int vec_any_ge (vector bool short, vector signed short);
9726 int vec_any_ge (vector signed int, vector bool int);
9727 int vec_any_ge (vector unsigned int, vector bool int);
9728 int vec_any_ge (vector unsigned int, vector unsigned int);
9729 int vec_any_ge (vector signed int, vector signed int);
9730 int vec_any_ge (vector bool int, vector unsigned int);
9731 int vec_any_ge (vector bool int, vector signed int);
9732 int vec_any_ge (vector float, vector float);
9734 int vec_any_gt (vector bool char, vector unsigned char);
9735 int vec_any_gt (vector unsigned char, vector bool char);
9736 int vec_any_gt (vector unsigned char, vector unsigned char);
9737 int vec_any_gt (vector bool char, vector signed char);
9738 int vec_any_gt (vector signed char, vector bool char);
9739 int vec_any_gt (vector signed char, vector signed char);
9740 int vec_any_gt (vector bool short, vector unsigned short);
9741 int vec_any_gt (vector unsigned short, vector bool short);
9742 int vec_any_gt (vector unsigned short, vector unsigned short);
9743 int vec_any_gt (vector bool short, vector signed short);
9744 int vec_any_gt (vector signed short, vector bool short);
9745 int vec_any_gt (vector signed short, vector signed short);
9746 int vec_any_gt (vector bool int, vector unsigned int);
9747 int vec_any_gt (vector unsigned int, vector bool int);
9748 int vec_any_gt (vector unsigned int, vector unsigned int);
9749 int vec_any_gt (vector bool int, vector signed int);
9750 int vec_any_gt (vector signed int, vector bool int);
9751 int vec_any_gt (vector signed int, vector signed int);
9752 int vec_any_gt (vector float, vector float);
9754 int vec_any_le (vector bool char, vector unsigned char);
9755 int vec_any_le (vector unsigned char, vector bool char);
9756 int vec_any_le (vector unsigned char, vector unsigned char);
9757 int vec_any_le (vector bool char, vector signed char);
9758 int vec_any_le (vector signed char, vector bool char);
9759 int vec_any_le (vector signed char, vector signed char);
9760 int vec_any_le (vector bool short, vector unsigned short);
9761 int vec_any_le (vector unsigned short, vector bool short);
9762 int vec_any_le (vector unsigned short, vector unsigned short);
9763 int vec_any_le (vector bool short, vector signed short);
9764 int vec_any_le (vector signed short, vector bool short);
9765 int vec_any_le (vector signed short, vector signed short);
9766 int vec_any_le (vector bool int, vector unsigned int);
9767 int vec_any_le (vector unsigned int, vector bool int);
9768 int vec_any_le (vector unsigned int, vector unsigned int);
9769 int vec_any_le (vector bool int, vector signed int);
9770 int vec_any_le (vector signed int, vector bool int);
9771 int vec_any_le (vector signed int, vector signed int);
9772 int vec_any_le (vector float, vector float);
9774 int vec_any_lt (vector bool char, vector unsigned char);
9775 int vec_any_lt (vector unsigned char, vector bool char);
9776 int vec_any_lt (vector unsigned char, vector unsigned char);
9777 int vec_any_lt (vector bool char, vector signed char);
9778 int vec_any_lt (vector signed char, vector bool char);
9779 int vec_any_lt (vector signed char, vector signed char);
9780 int vec_any_lt (vector bool short, vector unsigned short);
9781 int vec_any_lt (vector unsigned short, vector bool short);
9782 int vec_any_lt (vector unsigned short, vector unsigned short);
9783 int vec_any_lt (vector bool short, vector signed short);
9784 int vec_any_lt (vector signed short, vector bool short);
9785 int vec_any_lt (vector signed short, vector signed short);
9786 int vec_any_lt (vector bool int, vector unsigned int);
9787 int vec_any_lt (vector unsigned int, vector bool int);
9788 int vec_any_lt (vector unsigned int, vector unsigned int);
9789 int vec_any_lt (vector bool int, vector signed int);
9790 int vec_any_lt (vector signed int, vector bool int);
9791 int vec_any_lt (vector signed int, vector signed int);
9792 int vec_any_lt (vector float, vector float);
9794 int vec_any_nan (vector float);
9796 int vec_any_ne (vector signed char, vector bool char);
9797 int vec_any_ne (vector signed char, vector signed char);
9798 int vec_any_ne (vector unsigned char, vector bool char);
9799 int vec_any_ne (vector unsigned char, vector unsigned char);
9800 int vec_any_ne (vector bool char, vector bool char);
9801 int vec_any_ne (vector bool char, vector unsigned char);
9802 int vec_any_ne (vector bool char, vector signed char);
9803 int vec_any_ne (vector signed short, vector bool short);
9804 int vec_any_ne (vector signed short, vector signed short);
9805 int vec_any_ne (vector unsigned short, vector bool short);
9806 int vec_any_ne (vector unsigned short, vector unsigned short);
9807 int vec_any_ne (vector bool short, vector bool short);
9808 int vec_any_ne (vector bool short, vector unsigned short);
9809 int vec_any_ne (vector bool short, vector signed short);
9810 int vec_any_ne (vector pixel, vector pixel);
9811 int vec_any_ne (vector signed int, vector bool int);
9812 int vec_any_ne (vector signed int, vector signed int);
9813 int vec_any_ne (vector unsigned int, vector bool int);
9814 int vec_any_ne (vector unsigned int, vector unsigned int);
9815 int vec_any_ne (vector bool int, vector bool int);
9816 int vec_any_ne (vector bool int, vector unsigned int);
9817 int vec_any_ne (vector bool int, vector signed int);
9818 int vec_any_ne (vector float, vector float);
9820 int vec_any_nge (vector float, vector float);
9822 int vec_any_ngt (vector float, vector float);
9824 int vec_any_nle (vector float, vector float);
9826 int vec_any_nlt (vector float, vector float);
9828 int vec_any_numeric (vector float);
9830 int vec_any_out (vector float, vector float);
9833 @node SPARC VIS Built-in Functions
9834 @subsection SPARC VIS Built-in Functions
9836 GCC supports SIMD operations on the SPARC using both the generic vector
9837 extensions (@pxref{Vector Extensions}) as well as built-in functions for
9838 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
9839 switch, the VIS extension is exposed as the following built-in functions:
9842 typedef int v2si __attribute__ ((vector_size (8)));
9843 typedef short v4hi __attribute__ ((vector_size (8)));
9844 typedef short v2hi __attribute__ ((vector_size (4)));
9845 typedef char v8qi __attribute__ ((vector_size (8)));
9846 typedef char v4qi __attribute__ ((vector_size (4)));
9848 void * __builtin_vis_alignaddr (void *, long);
9849 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
9850 v2si __builtin_vis_faligndatav2si (v2si, v2si);
9851 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
9852 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
9854 v4hi __builtin_vis_fexpand (v4qi);
9856 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
9857 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
9858 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
9859 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
9860 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
9861 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
9862 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
9864 v4qi __builtin_vis_fpack16 (v4hi);
9865 v8qi __builtin_vis_fpack32 (v2si, v2si);
9866 v2hi __builtin_vis_fpackfix (v2si);
9867 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
9869 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
9872 @node Target Format Checks
9873 @section Format Checks Specific to Particular Target Machines
9875 For some target machines, GCC supports additional options to the
9877 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
9880 * Solaris Format Checks::
9883 @node Solaris Format Checks
9884 @subsection Solaris Format Checks
9886 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
9887 check. @code{cmn_err} accepts a subset of the standard @code{printf}
9888 conversions, and the two-argument @code{%b} conversion for displaying
9889 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
9892 @section Pragmas Accepted by GCC
9896 GCC supports several types of pragmas, primarily in order to compile
9897 code originally written for other compilers. Note that in general
9898 we do not recommend the use of pragmas; @xref{Function Attributes},
9899 for further explanation.
9904 * RS/6000 and PowerPC Pragmas::
9907 * Symbol-Renaming Pragmas::
9908 * Structure-Packing Pragmas::
9910 * Diagnostic Pragmas::
9911 * Visibility Pragmas::
9915 @subsection ARM Pragmas
9917 The ARM target defines pragmas for controlling the default addition of
9918 @code{long_call} and @code{short_call} attributes to functions.
9919 @xref{Function Attributes}, for information about the effects of these
9924 @cindex pragma, long_calls
9925 Set all subsequent functions to have the @code{long_call} attribute.
9928 @cindex pragma, no_long_calls
9929 Set all subsequent functions to have the @code{short_call} attribute.
9931 @item long_calls_off
9932 @cindex pragma, long_calls_off
9933 Do not affect the @code{long_call} or @code{short_call} attributes of
9934 subsequent functions.
9938 @subsection M32C Pragmas
9941 @item memregs @var{number}
9942 @cindex pragma, memregs
9943 Overrides the command line option @code{-memregs=} for the current
9944 file. Use with care! This pragma must be before any function in the
9945 file, and mixing different memregs values in different objects may
9946 make them incompatible. This pragma is useful when a
9947 performance-critical function uses a memreg for temporary values,
9948 as it may allow you to reduce the number of memregs used.
9952 @node RS/6000 and PowerPC Pragmas
9953 @subsection RS/6000 and PowerPC Pragmas
9955 The RS/6000 and PowerPC targets define one pragma for controlling
9956 whether or not the @code{longcall} attribute is added to function
9957 declarations by default. This pragma overrides the @option{-mlongcall}
9958 option, but not the @code{longcall} and @code{shortcall} attributes.
9959 @xref{RS/6000 and PowerPC Options}, for more information about when long
9960 calls are and are not necessary.
9964 @cindex pragma, longcall
9965 Apply the @code{longcall} attribute to all subsequent function
9969 Do not apply the @code{longcall} attribute to subsequent function
9973 @c Describe c4x pragmas here.
9974 @c Describe h8300 pragmas here.
9975 @c Describe sh pragmas here.
9976 @c Describe v850 pragmas here.
9978 @node Darwin Pragmas
9979 @subsection Darwin Pragmas
9981 The following pragmas are available for all architectures running the
9982 Darwin operating system. These are useful for compatibility with other
9986 @item mark @var{tokens}@dots{}
9987 @cindex pragma, mark
9988 This pragma is accepted, but has no effect.
9990 @item options align=@var{alignment}
9991 @cindex pragma, options align
9992 This pragma sets the alignment of fields in structures. The values of
9993 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
9994 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
9995 properly; to restore the previous setting, use @code{reset} for the
9998 @item segment @var{tokens}@dots{}
9999 @cindex pragma, segment
10000 This pragma is accepted, but has no effect.
10002 @item unused (@var{var} [, @var{var}]@dots{})
10003 @cindex pragma, unused
10004 This pragma declares variables to be possibly unused. GCC will not
10005 produce warnings for the listed variables. The effect is similar to
10006 that of the @code{unused} attribute, except that this pragma may appear
10007 anywhere within the variables' scopes.
10010 @node Solaris Pragmas
10011 @subsection Solaris Pragmas
10013 The Solaris target supports @code{#pragma redefine_extname}
10014 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
10015 @code{#pragma} directives for compatibility with the system compiler.
10018 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
10019 @cindex pragma, align
10021 Increase the minimum alignment of each @var{variable} to @var{alignment}.
10022 This is the same as GCC's @code{aligned} attribute @pxref{Variable
10023 Attributes}). Macro expansion occurs on the arguments to this pragma
10024 when compiling C. It does not currently occur when compiling C++, but
10025 this is a bug which may be fixed in a future release.
10027 @item fini (@var{function} [, @var{function}]...)
10028 @cindex pragma, fini
10030 This pragma causes each listed @var{function} to be called after
10031 main, or during shared module unloading, by adding a call to the
10032 @code{.fini} section.
10034 @item init (@var{function} [, @var{function}]...)
10035 @cindex pragma, init
10037 This pragma causes each listed @var{function} to be called during
10038 initialization (before @code{main}) or during shared module loading, by
10039 adding a call to the @code{.init} section.
10043 @node Symbol-Renaming Pragmas
10044 @subsection Symbol-Renaming Pragmas
10046 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
10047 supports two @code{#pragma} directives which change the name used in
10048 assembly for a given declaration. These pragmas are only available on
10049 platforms whose system headers need them. To get this effect on all
10050 platforms supported by GCC, use the asm labels extension (@pxref{Asm
10054 @item redefine_extname @var{oldname} @var{newname}
10055 @cindex pragma, redefine_extname
10057 This pragma gives the C function @var{oldname} the assembly symbol
10058 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
10059 will be defined if this pragma is available (currently only on
10062 @item extern_prefix @var{string}
10063 @cindex pragma, extern_prefix
10065 This pragma causes all subsequent external function and variable
10066 declarations to have @var{string} prepended to their assembly symbols.
10067 This effect may be terminated with another @code{extern_prefix} pragma
10068 whose argument is an empty string. The preprocessor macro
10069 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
10070 available (currently only on Tru64 UNIX)@.
10073 These pragmas and the asm labels extension interact in a complicated
10074 manner. Here are some corner cases you may want to be aware of.
10077 @item Both pragmas silently apply only to declarations with external
10078 linkage. Asm labels do not have this restriction.
10080 @item In C++, both pragmas silently apply only to declarations with
10081 ``C'' linkage. Again, asm labels do not have this restriction.
10083 @item If any of the three ways of changing the assembly name of a
10084 declaration is applied to a declaration whose assembly name has
10085 already been determined (either by a previous use of one of these
10086 features, or because the compiler needed the assembly name in order to
10087 generate code), and the new name is different, a warning issues and
10088 the name does not change.
10090 @item The @var{oldname} used by @code{#pragma redefine_extname} is
10091 always the C-language name.
10093 @item If @code{#pragma extern_prefix} is in effect, and a declaration
10094 occurs with an asm label attached, the prefix is silently ignored for
10097 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
10098 apply to the same declaration, whichever triggered first wins, and a
10099 warning issues if they contradict each other. (We would like to have
10100 @code{#pragma redefine_extname} always win, for consistency with asm
10101 labels, but if @code{#pragma extern_prefix} triggers first we have no
10102 way of knowing that that happened.)
10105 @node Structure-Packing Pragmas
10106 @subsection Structure-Packing Pragmas
10108 For compatibility with Win32, GCC supports a set of @code{#pragma}
10109 directives which change the maximum alignment of members of structures
10110 (other than zero-width bitfields), unions, and classes subsequently
10111 defined. The @var{n} value below always is required to be a small power
10112 of two and specifies the new alignment in bytes.
10115 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
10116 @item @code{#pragma pack()} sets the alignment to the one that was in
10117 effect when compilation started (see also command line option
10118 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
10119 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
10120 setting on an internal stack and then optionally sets the new alignment.
10121 @item @code{#pragma pack(pop)} restores the alignment setting to the one
10122 saved at the top of the internal stack (and removes that stack entry).
10123 Note that @code{#pragma pack([@var{n}])} does not influence this internal
10124 stack; thus it is possible to have @code{#pragma pack(push)} followed by
10125 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
10126 @code{#pragma pack(pop)}.
10129 Some targets, e.g. i386 and powerpc, support the @code{ms_struct}
10130 @code{#pragma} which lays out a structure as the documented
10131 @code{__attribute__ ((ms_struct))}.
10133 @item @code{#pragma ms_struct on} turns on the layout for structures
10135 @item @code{#pragma ms_struct off} turns off the layout for structures
10137 @item @code{#pragma ms_struct reset} goes back to the default layout.
10141 @subsection Weak Pragmas
10143 For compatibility with SVR4, GCC supports a set of @code{#pragma}
10144 directives for declaring symbols to be weak, and defining weak
10148 @item #pragma weak @var{symbol}
10149 @cindex pragma, weak
10150 This pragma declares @var{symbol} to be weak, as if the declaration
10151 had the attribute of the same name. The pragma may appear before
10152 or after the declaration of @var{symbol}, but must appear before
10153 either its first use or its definition. It is not an error for
10154 @var{symbol} to never be defined at all.
10156 @item #pragma weak @var{symbol1} = @var{symbol2}
10157 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
10158 It is an error if @var{symbol2} is not defined in the current
10162 @node Diagnostic Pragmas
10163 @subsection Diagnostic Pragmas
10165 GCC allows the user to selectively enable or disable certain types of
10166 diagnostics, and change the kind of the diagnostic. For example, a
10167 project's policy might require that all sources compile with
10168 @option{-Werror} but certain files might have exceptions allowing
10169 specific types of warnings. Or, a project might selectively enable
10170 diagnostics and treat them as errors depending on which preprocessor
10171 macros are defined.
10174 @item #pragma GCC diagnostic @var{kind} @var{option}
10175 @cindex pragma, diagnostic
10177 Modifies the disposition of a diagnostic. Note that not all
10178 diagnostics are modifiable; at the moment only warnings (normally
10179 controlled by @samp{-W...}) can be controlled, and not all of them.
10180 Use @option{-fdiagnostics-show-option} to determine which diagnostics
10181 are controllable and which option controls them.
10183 @var{kind} is @samp{error} to treat this diagnostic as an error,
10184 @samp{warning} to treat it like a warning (even if @option{-Werror} is
10185 in effect), or @samp{ignored} if the diagnostic is to be ignored.
10186 @var{option} is a double quoted string which matches the command line
10190 #pragma GCC diagnostic warning "-Wformat"
10191 #pragma GCC diagnostic error "-Wformat"
10192 #pragma GCC diagnostic ignored "-Wformat"
10195 Note that these pragmas override any command line options. Also,
10196 while it is syntactically valid to put these pragmas anywhere in your
10197 sources, the only supported location for them is before any data or
10198 functions are defined. Doing otherwise may result in unpredictable
10199 results depending on how the optimizer manages your sources. If the
10200 same option is listed multiple times, the last one specified is the
10201 one that is in effect. This pragma is not intended to be a general
10202 purpose replacement for command line options, but for implementing
10203 strict control over project policies.
10207 @node Visibility Pragmas
10208 @subsection Visibility Pragmas
10211 @item #pragma GCC visibility push(@var{visibility})
10212 @itemx #pragma GCC visibility pop
10213 @cindex pragma, visibility
10215 This pragma allows the user to set the visibility for multiple
10216 declarations without having to give each a visibility attribute
10217 @xref{Function Attributes}, for more information about visibility and
10218 the attribute syntax.
10220 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
10221 declarations. Class members and template specializations are not
10222 affected; if you want to override the visibility for a particular
10223 member or instantiation, you must use an attribute.
10227 @node Unnamed Fields
10228 @section Unnamed struct/union fields within structs/unions
10232 For compatibility with other compilers, GCC allows you to define
10233 a structure or union that contains, as fields, structures and unions
10234 without names. For example:
10247 In this example, the user would be able to access members of the unnamed
10248 union with code like @samp{foo.b}. Note that only unnamed structs and
10249 unions are allowed, you may not have, for example, an unnamed
10252 You must never create such structures that cause ambiguous field definitions.
10253 For example, this structure:
10264 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
10265 Such constructs are not supported and must be avoided. In the future,
10266 such constructs may be detected and treated as compilation errors.
10268 @opindex fms-extensions
10269 Unless @option{-fms-extensions} is used, the unnamed field must be a
10270 structure or union definition without a tag (for example, @samp{struct
10271 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
10272 also be a definition with a tag such as @samp{struct foo @{ int a;
10273 @};}, a reference to a previously defined structure or union such as
10274 @samp{struct foo;}, or a reference to a @code{typedef} name for a
10275 previously defined structure or union type.
10278 @section Thread-Local Storage
10279 @cindex Thread-Local Storage
10280 @cindex @acronym{TLS}
10283 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
10284 are allocated such that there is one instance of the variable per extant
10285 thread. The run-time model GCC uses to implement this originates
10286 in the IA-64 processor-specific ABI, but has since been migrated
10287 to other processors as well. It requires significant support from
10288 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
10289 system libraries (@file{libc.so} and @file{libpthread.so}), so it
10290 is not available everywhere.
10292 At the user level, the extension is visible with a new storage
10293 class keyword: @code{__thread}. For example:
10297 extern __thread struct state s;
10298 static __thread char *p;
10301 The @code{__thread} specifier may be used alone, with the @code{extern}
10302 or @code{static} specifiers, but with no other storage class specifier.
10303 When used with @code{extern} or @code{static}, @code{__thread} must appear
10304 immediately after the other storage class specifier.
10306 The @code{__thread} specifier may be applied to any global, file-scoped
10307 static, function-scoped static, or static data member of a class. It may
10308 not be applied to block-scoped automatic or non-static data member.
10310 When the address-of operator is applied to a thread-local variable, it is
10311 evaluated at run-time and returns the address of the current thread's
10312 instance of that variable. An address so obtained may be used by any
10313 thread. When a thread terminates, any pointers to thread-local variables
10314 in that thread become invalid.
10316 No static initialization may refer to the address of a thread-local variable.
10318 In C++, if an initializer is present for a thread-local variable, it must
10319 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
10322 See @uref{http://people.redhat.com/drepper/tls.pdf,
10323 ELF Handling For Thread-Local Storage} for a detailed explanation of
10324 the four thread-local storage addressing models, and how the run-time
10325 is expected to function.
10328 * C99 Thread-Local Edits::
10329 * C++98 Thread-Local Edits::
10332 @node C99 Thread-Local Edits
10333 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
10335 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
10336 that document the exact semantics of the language extension.
10340 @cite{5.1.2 Execution environments}
10342 Add new text after paragraph 1
10345 Within either execution environment, a @dfn{thread} is a flow of
10346 control within a program. It is implementation defined whether
10347 or not there may be more than one thread associated with a program.
10348 It is implementation defined how threads beyond the first are
10349 created, the name and type of the function called at thread
10350 startup, and how threads may be terminated. However, objects
10351 with thread storage duration shall be initialized before thread
10356 @cite{6.2.4 Storage durations of objects}
10358 Add new text before paragraph 3
10361 An object whose identifier is declared with the storage-class
10362 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
10363 Its lifetime is the entire execution of the thread, and its
10364 stored value is initialized only once, prior to thread startup.
10368 @cite{6.4.1 Keywords}
10370 Add @code{__thread}.
10373 @cite{6.7.1 Storage-class specifiers}
10375 Add @code{__thread} to the list of storage class specifiers in
10378 Change paragraph 2 to
10381 With the exception of @code{__thread}, at most one storage-class
10382 specifier may be given [@dots{}]. The @code{__thread} specifier may
10383 be used alone, or immediately following @code{extern} or
10387 Add new text after paragraph 6
10390 The declaration of an identifier for a variable that has
10391 block scope that specifies @code{__thread} shall also
10392 specify either @code{extern} or @code{static}.
10394 The @code{__thread} specifier shall be used only with
10399 @node C++98 Thread-Local Edits
10400 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
10402 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
10403 that document the exact semantics of the language extension.
10407 @b{[intro.execution]}
10409 New text after paragraph 4
10412 A @dfn{thread} is a flow of control within the abstract machine.
10413 It is implementation defined whether or not there may be more than
10417 New text after paragraph 7
10420 It is unspecified whether additional action must be taken to
10421 ensure when and whether side effects are visible to other threads.
10427 Add @code{__thread}.
10430 @b{[basic.start.main]}
10432 Add after paragraph 5
10435 The thread that begins execution at the @code{main} function is called
10436 the @dfn{main thread}. It is implementation defined how functions
10437 beginning threads other than the main thread are designated or typed.
10438 A function so designated, as well as the @code{main} function, is called
10439 a @dfn{thread startup function}. It is implementation defined what
10440 happens if a thread startup function returns. It is implementation
10441 defined what happens to other threads when any thread calls @code{exit}.
10445 @b{[basic.start.init]}
10447 Add after paragraph 4
10450 The storage for an object of thread storage duration shall be
10451 statically initialized before the first statement of the thread startup
10452 function. An object of thread storage duration shall not require
10453 dynamic initialization.
10457 @b{[basic.start.term]}
10459 Add after paragraph 3
10462 The type of an object with thread storage duration shall not have a
10463 non-trivial destructor, nor shall it be an array type whose elements
10464 (directly or indirectly) have non-trivial destructors.
10470 Add ``thread storage duration'' to the list in paragraph 1.
10475 Thread, static, and automatic storage durations are associated with
10476 objects introduced by declarations [@dots{}].
10479 Add @code{__thread} to the list of specifiers in paragraph 3.
10482 @b{[basic.stc.thread]}
10484 New section before @b{[basic.stc.static]}
10487 The keyword @code{__thread} applied to a non-local object gives the
10488 object thread storage duration.
10490 A local variable or class data member declared both @code{static}
10491 and @code{__thread} gives the variable or member thread storage
10496 @b{[basic.stc.static]}
10501 All objects which have neither thread storage duration, dynamic
10502 storage duration nor are local [@dots{}].
10508 Add @code{__thread} to the list in paragraph 1.
10513 With the exception of @code{__thread}, at most one
10514 @var{storage-class-specifier} shall appear in a given
10515 @var{decl-specifier-seq}. The @code{__thread} specifier may
10516 be used alone, or immediately following the @code{extern} or
10517 @code{static} specifiers. [@dots{}]
10520 Add after paragraph 5
10523 The @code{__thread} specifier can be applied only to the names of objects
10524 and to anonymous unions.
10530 Add after paragraph 6
10533 Non-@code{static} members shall not be @code{__thread}.
10537 @node Binary constants
10538 @section Binary constants using the @samp{0b} prefix
10539 @cindex Binary constants using the @samp{0b} prefix
10541 Integer constants can be written as binary constants, consisting of a
10542 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
10543 @samp{0B}. This is particularly useful in environments that operate a
10544 lot on the bit-level (like microcontrollers).
10546 The following statements are identical:
10555 The type of these constants follows the same rules as for octal or
10556 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
10559 @node C++ Extensions
10560 @chapter Extensions to the C++ Language
10561 @cindex extensions, C++ language
10562 @cindex C++ language extensions
10564 The GNU compiler provides these extensions to the C++ language (and you
10565 can also use most of the C language extensions in your C++ programs). If you
10566 want to write code that checks whether these features are available, you can
10567 test for the GNU compiler the same way as for C programs: check for a
10568 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
10569 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
10570 Predefined Macros,cpp,The GNU C Preprocessor}).
10573 * Volatiles:: What constitutes an access to a volatile object.
10574 * Restricted Pointers:: C99 restricted pointers and references.
10575 * Vague Linkage:: Where G++ puts inlines, vtables and such.
10576 * C++ Interface:: You can use a single C++ header file for both
10577 declarations and definitions.
10578 * Template Instantiation:: Methods for ensuring that exactly one copy of
10579 each needed template instantiation is emitted.
10580 * Bound member functions:: You can extract a function pointer to the
10581 method denoted by a @samp{->*} or @samp{.*} expression.
10582 * C++ Attributes:: Variable, function, and type attributes for C++ only.
10583 * Namespace Association:: Strong using-directives for namespace association.
10584 * Java Exceptions:: Tweaking exception handling to work with Java.
10585 * Deprecated Features:: Things will disappear from g++.
10586 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
10590 @section When is a Volatile Object Accessed?
10591 @cindex accessing volatiles
10592 @cindex volatile read
10593 @cindex volatile write
10594 @cindex volatile access
10596 Both the C and C++ standard have the concept of volatile objects. These
10597 are normally accessed by pointers and used for accessing hardware. The
10598 standards encourage compilers to refrain from optimizations concerning
10599 accesses to volatile objects. The C standard leaves it implementation
10600 defined as to what constitutes a volatile access. The C++ standard omits
10601 to specify this, except to say that C++ should behave in a similar manner
10602 to C with respect to volatiles, where possible. The minimum either
10603 standard specifies is that at a sequence point all previous accesses to
10604 volatile objects have stabilized and no subsequent accesses have
10605 occurred. Thus an implementation is free to reorder and combine
10606 volatile accesses which occur between sequence points, but cannot do so
10607 for accesses across a sequence point. The use of volatiles does not
10608 allow you to violate the restriction on updating objects multiple times
10609 within a sequence point.
10611 @xref{Qualifiers implementation, , Volatile qualifier and the C compiler}.
10613 The behavior differs slightly between C and C++ in the non-obvious cases:
10616 volatile int *src = @var{somevalue};
10620 With C, such expressions are rvalues, and GCC interprets this either as a
10621 read of the volatile object being pointed to or only as request to evaluate
10622 the side-effects. The C++ standard specifies that such expressions do not
10623 undergo lvalue to rvalue conversion, and that the type of the dereferenced
10624 object may be incomplete. The C++ standard does not specify explicitly
10625 that it is this lvalue to rvalue conversion which may be responsible for
10626 causing an access. However, there is reason to believe that it is,
10627 because otherwise certain simple expressions become undefined. However,
10628 because it would surprise most programmers, G++ treats dereferencing a
10629 pointer to volatile object of complete type when the value is unused as
10630 GCC would do for an equivalent type in C. When the object has incomplete
10631 type, G++ issues a warning; if you wish to force an error, you must
10632 force a conversion to rvalue with, for instance, a static cast.
10634 When using a reference to volatile, G++ does not treat equivalent
10635 expressions as accesses to volatiles, but instead issues a warning that
10636 no volatile is accessed. The rationale for this is that otherwise it
10637 becomes difficult to determine where volatile access occur, and not
10638 possible to ignore the return value from functions returning volatile
10639 references. Again, if you wish to force a read, cast the reference to
10642 @node Restricted Pointers
10643 @section Restricting Pointer Aliasing
10644 @cindex restricted pointers
10645 @cindex restricted references
10646 @cindex restricted this pointer
10648 As with the C front end, G++ understands the C99 feature of restricted pointers,
10649 specified with the @code{__restrict__}, or @code{__restrict} type
10650 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
10651 language flag, @code{restrict} is not a keyword in C++.
10653 In addition to allowing restricted pointers, you can specify restricted
10654 references, which indicate that the reference is not aliased in the local
10658 void fn (int *__restrict__ rptr, int &__restrict__ rref)
10665 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
10666 @var{rref} refers to a (different) unaliased integer.
10668 You may also specify whether a member function's @var{this} pointer is
10669 unaliased by using @code{__restrict__} as a member function qualifier.
10672 void T::fn () __restrict__
10679 Within the body of @code{T::fn}, @var{this} will have the effective
10680 definition @code{T *__restrict__ const this}. Notice that the
10681 interpretation of a @code{__restrict__} member function qualifier is
10682 different to that of @code{const} or @code{volatile} qualifier, in that it
10683 is applied to the pointer rather than the object. This is consistent with
10684 other compilers which implement restricted pointers.
10686 As with all outermost parameter qualifiers, @code{__restrict__} is
10687 ignored in function definition matching. This means you only need to
10688 specify @code{__restrict__} in a function definition, rather than
10689 in a function prototype as well.
10691 @node Vague Linkage
10692 @section Vague Linkage
10693 @cindex vague linkage
10695 There are several constructs in C++ which require space in the object
10696 file but are not clearly tied to a single translation unit. We say that
10697 these constructs have ``vague linkage''. Typically such constructs are
10698 emitted wherever they are needed, though sometimes we can be more
10702 @item Inline Functions
10703 Inline functions are typically defined in a header file which can be
10704 included in many different compilations. Hopefully they can usually be
10705 inlined, but sometimes an out-of-line copy is necessary, if the address
10706 of the function is taken or if inlining fails. In general, we emit an
10707 out-of-line copy in all translation units where one is needed. As an
10708 exception, we only emit inline virtual functions with the vtable, since
10709 it will always require a copy.
10711 Local static variables and string constants used in an inline function
10712 are also considered to have vague linkage, since they must be shared
10713 between all inlined and out-of-line instances of the function.
10717 C++ virtual functions are implemented in most compilers using a lookup
10718 table, known as a vtable. The vtable contains pointers to the virtual
10719 functions provided by a class, and each object of the class contains a
10720 pointer to its vtable (or vtables, in some multiple-inheritance
10721 situations). If the class declares any non-inline, non-pure virtual
10722 functions, the first one is chosen as the ``key method'' for the class,
10723 and the vtable is only emitted in the translation unit where the key
10726 @emph{Note:} If the chosen key method is later defined as inline, the
10727 vtable will still be emitted in every translation unit which defines it.
10728 Make sure that any inline virtuals are declared inline in the class
10729 body, even if they are not defined there.
10731 @item type_info objects
10734 C++ requires information about types to be written out in order to
10735 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
10736 For polymorphic classes (classes with virtual functions), the type_info
10737 object is written out along with the vtable so that @samp{dynamic_cast}
10738 can determine the dynamic type of a class object at runtime. For all
10739 other types, we write out the type_info object when it is used: when
10740 applying @samp{typeid} to an expression, throwing an object, or
10741 referring to a type in a catch clause or exception specification.
10743 @item Template Instantiations
10744 Most everything in this section also applies to template instantiations,
10745 but there are other options as well.
10746 @xref{Template Instantiation,,Where's the Template?}.
10750 When used with GNU ld version 2.8 or later on an ELF system such as
10751 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
10752 these constructs will be discarded at link time. This is known as
10755 On targets that don't support COMDAT, but do support weak symbols, GCC
10756 will use them. This way one copy will override all the others, but
10757 the unused copies will still take up space in the executable.
10759 For targets which do not support either COMDAT or weak symbols,
10760 most entities with vague linkage will be emitted as local symbols to
10761 avoid duplicate definition errors from the linker. This will not happen
10762 for local statics in inlines, however, as having multiple copies will
10763 almost certainly break things.
10765 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
10766 another way to control placement of these constructs.
10768 @node C++ Interface
10769 @section #pragma interface and implementation
10771 @cindex interface and implementation headers, C++
10772 @cindex C++ interface and implementation headers
10773 @cindex pragmas, interface and implementation
10775 @code{#pragma interface} and @code{#pragma implementation} provide the
10776 user with a way of explicitly directing the compiler to emit entities
10777 with vague linkage (and debugging information) in a particular
10780 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
10781 most cases, because of COMDAT support and the ``key method'' heuristic
10782 mentioned in @ref{Vague Linkage}. Using them can actually cause your
10783 program to grow due to unnecessary out-of-line copies of inline
10784 functions. Currently (3.4) the only benefit of these
10785 @code{#pragma}s is reduced duplication of debugging information, and
10786 that should be addressed soon on DWARF 2 targets with the use of
10790 @item #pragma interface
10791 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
10792 @kindex #pragma interface
10793 Use this directive in @emph{header files} that define object classes, to save
10794 space in most of the object files that use those classes. Normally,
10795 local copies of certain information (backup copies of inline member
10796 functions, debugging information, and the internal tables that implement
10797 virtual functions) must be kept in each object file that includes class
10798 definitions. You can use this pragma to avoid such duplication. When a
10799 header file containing @samp{#pragma interface} is included in a
10800 compilation, this auxiliary information will not be generated (unless
10801 the main input source file itself uses @samp{#pragma implementation}).
10802 Instead, the object files will contain references to be resolved at link
10805 The second form of this directive is useful for the case where you have
10806 multiple headers with the same name in different directories. If you
10807 use this form, you must specify the same string to @samp{#pragma
10810 @item #pragma implementation
10811 @itemx #pragma implementation "@var{objects}.h"
10812 @kindex #pragma implementation
10813 Use this pragma in a @emph{main input file}, when you want full output from
10814 included header files to be generated (and made globally visible). The
10815 included header file, in turn, should use @samp{#pragma interface}.
10816 Backup copies of inline member functions, debugging information, and the
10817 internal tables used to implement virtual functions are all generated in
10818 implementation files.
10820 @cindex implied @code{#pragma implementation}
10821 @cindex @code{#pragma implementation}, implied
10822 @cindex naming convention, implementation headers
10823 If you use @samp{#pragma implementation} with no argument, it applies to
10824 an include file with the same basename@footnote{A file's @dfn{basename}
10825 was the name stripped of all leading path information and of trailing
10826 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
10827 file. For example, in @file{allclass.cc}, giving just
10828 @samp{#pragma implementation}
10829 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
10831 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
10832 an implementation file whenever you would include it from
10833 @file{allclass.cc} even if you never specified @samp{#pragma
10834 implementation}. This was deemed to be more trouble than it was worth,
10835 however, and disabled.
10837 Use the string argument if you want a single implementation file to
10838 include code from multiple header files. (You must also use
10839 @samp{#include} to include the header file; @samp{#pragma
10840 implementation} only specifies how to use the file---it doesn't actually
10843 There is no way to split up the contents of a single header file into
10844 multiple implementation files.
10847 @cindex inlining and C++ pragmas
10848 @cindex C++ pragmas, effect on inlining
10849 @cindex pragmas in C++, effect on inlining
10850 @samp{#pragma implementation} and @samp{#pragma interface} also have an
10851 effect on function inlining.
10853 If you define a class in a header file marked with @samp{#pragma
10854 interface}, the effect on an inline function defined in that class is
10855 similar to an explicit @code{extern} declaration---the compiler emits
10856 no code at all to define an independent version of the function. Its
10857 definition is used only for inlining with its callers.
10859 @opindex fno-implement-inlines
10860 Conversely, when you include the same header file in a main source file
10861 that declares it as @samp{#pragma implementation}, the compiler emits
10862 code for the function itself; this defines a version of the function
10863 that can be found via pointers (or by callers compiled without
10864 inlining). If all calls to the function can be inlined, you can avoid
10865 emitting the function by compiling with @option{-fno-implement-inlines}.
10866 If any calls were not inlined, you will get linker errors.
10868 @node Template Instantiation
10869 @section Where's the Template?
10870 @cindex template instantiation
10872 C++ templates are the first language feature to require more
10873 intelligence from the environment than one usually finds on a UNIX
10874 system. Somehow the compiler and linker have to make sure that each
10875 template instance occurs exactly once in the executable if it is needed,
10876 and not at all otherwise. There are two basic approaches to this
10877 problem, which are referred to as the Borland model and the Cfront model.
10880 @item Borland model
10881 Borland C++ solved the template instantiation problem by adding the code
10882 equivalent of common blocks to their linker; the compiler emits template
10883 instances in each translation unit that uses them, and the linker
10884 collapses them together. The advantage of this model is that the linker
10885 only has to consider the object files themselves; there is no external
10886 complexity to worry about. This disadvantage is that compilation time
10887 is increased because the template code is being compiled repeatedly.
10888 Code written for this model tends to include definitions of all
10889 templates in the header file, since they must be seen to be
10893 The AT&T C++ translator, Cfront, solved the template instantiation
10894 problem by creating the notion of a template repository, an
10895 automatically maintained place where template instances are stored. A
10896 more modern version of the repository works as follows: As individual
10897 object files are built, the compiler places any template definitions and
10898 instantiations encountered in the repository. At link time, the link
10899 wrapper adds in the objects in the repository and compiles any needed
10900 instances that were not previously emitted. The advantages of this
10901 model are more optimal compilation speed and the ability to use the
10902 system linker; to implement the Borland model a compiler vendor also
10903 needs to replace the linker. The disadvantages are vastly increased
10904 complexity, and thus potential for error; for some code this can be
10905 just as transparent, but in practice it can been very difficult to build
10906 multiple programs in one directory and one program in multiple
10907 directories. Code written for this model tends to separate definitions
10908 of non-inline member templates into a separate file, which should be
10909 compiled separately.
10912 When used with GNU ld version 2.8 or later on an ELF system such as
10913 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
10914 Borland model. On other systems, G++ implements neither automatic
10917 A future version of G++ will support a hybrid model whereby the compiler
10918 will emit any instantiations for which the template definition is
10919 included in the compile, and store template definitions and
10920 instantiation context information into the object file for the rest.
10921 The link wrapper will extract that information as necessary and invoke
10922 the compiler to produce the remaining instantiations. The linker will
10923 then combine duplicate instantiations.
10925 In the mean time, you have the following options for dealing with
10926 template instantiations:
10931 Compile your template-using code with @option{-frepo}. The compiler will
10932 generate files with the extension @samp{.rpo} listing all of the
10933 template instantiations used in the corresponding object files which
10934 could be instantiated there; the link wrapper, @samp{collect2}, will
10935 then update the @samp{.rpo} files to tell the compiler where to place
10936 those instantiations and rebuild any affected object files. The
10937 link-time overhead is negligible after the first pass, as the compiler
10938 will continue to place the instantiations in the same files.
10940 This is your best option for application code written for the Borland
10941 model, as it will just work. Code written for the Cfront model will
10942 need to be modified so that the template definitions are available at
10943 one or more points of instantiation; usually this is as simple as adding
10944 @code{#include <tmethods.cc>} to the end of each template header.
10946 For library code, if you want the library to provide all of the template
10947 instantiations it needs, just try to link all of its object files
10948 together; the link will fail, but cause the instantiations to be
10949 generated as a side effect. Be warned, however, that this may cause
10950 conflicts if multiple libraries try to provide the same instantiations.
10951 For greater control, use explicit instantiation as described in the next
10955 @opindex fno-implicit-templates
10956 Compile your code with @option{-fno-implicit-templates} to disable the
10957 implicit generation of template instances, and explicitly instantiate
10958 all the ones you use. This approach requires more knowledge of exactly
10959 which instances you need than do the others, but it's less
10960 mysterious and allows greater control. You can scatter the explicit
10961 instantiations throughout your program, perhaps putting them in the
10962 translation units where the instances are used or the translation units
10963 that define the templates themselves; you can put all of the explicit
10964 instantiations you need into one big file; or you can create small files
10971 template class Foo<int>;
10972 template ostream& operator <<
10973 (ostream&, const Foo<int>&);
10976 for each of the instances you need, and create a template instantiation
10977 library from those.
10979 If you are using Cfront-model code, you can probably get away with not
10980 using @option{-fno-implicit-templates} when compiling files that don't
10981 @samp{#include} the member template definitions.
10983 If you use one big file to do the instantiations, you may want to
10984 compile it without @option{-fno-implicit-templates} so you get all of the
10985 instances required by your explicit instantiations (but not by any
10986 other files) without having to specify them as well.
10988 G++ has extended the template instantiation syntax given in the ISO
10989 standard to allow forward declaration of explicit instantiations
10990 (with @code{extern}), instantiation of the compiler support data for a
10991 template class (i.e.@: the vtable) without instantiating any of its
10992 members (with @code{inline}), and instantiation of only the static data
10993 members of a template class, without the support data or member
10994 functions (with (@code{static}):
10997 extern template int max (int, int);
10998 inline template class Foo<int>;
10999 static template class Foo<int>;
11003 Do nothing. Pretend G++ does implement automatic instantiation
11004 management. Code written for the Borland model will work fine, but
11005 each translation unit will contain instances of each of the templates it
11006 uses. In a large program, this can lead to an unacceptable amount of code
11010 @node Bound member functions
11011 @section Extracting the function pointer from a bound pointer to member function
11013 @cindex pointer to member function
11014 @cindex bound pointer to member function
11016 In C++, pointer to member functions (PMFs) are implemented using a wide
11017 pointer of sorts to handle all the possible call mechanisms; the PMF
11018 needs to store information about how to adjust the @samp{this} pointer,
11019 and if the function pointed to is virtual, where to find the vtable, and
11020 where in the vtable to look for the member function. If you are using
11021 PMFs in an inner loop, you should really reconsider that decision. If
11022 that is not an option, you can extract the pointer to the function that
11023 would be called for a given object/PMF pair and call it directly inside
11024 the inner loop, to save a bit of time.
11026 Note that you will still be paying the penalty for the call through a
11027 function pointer; on most modern architectures, such a call defeats the
11028 branch prediction features of the CPU@. This is also true of normal
11029 virtual function calls.
11031 The syntax for this extension is
11035 extern int (A::*fp)();
11036 typedef int (*fptr)(A *);
11038 fptr p = (fptr)(a.*fp);
11041 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
11042 no object is needed to obtain the address of the function. They can be
11043 converted to function pointers directly:
11046 fptr p1 = (fptr)(&A::foo);
11049 @opindex Wno-pmf-conversions
11050 You must specify @option{-Wno-pmf-conversions} to use this extension.
11052 @node C++ Attributes
11053 @section C++-Specific Variable, Function, and Type Attributes
11055 Some attributes only make sense for C++ programs.
11058 @item init_priority (@var{priority})
11059 @cindex init_priority attribute
11062 In Standard C++, objects defined at namespace scope are guaranteed to be
11063 initialized in an order in strict accordance with that of their definitions
11064 @emph{in a given translation unit}. No guarantee is made for initializations
11065 across translation units. However, GNU C++ allows users to control the
11066 order of initialization of objects defined at namespace scope with the
11067 @code{init_priority} attribute by specifying a relative @var{priority},
11068 a constant integral expression currently bounded between 101 and 65535
11069 inclusive. Lower numbers indicate a higher priority.
11071 In the following example, @code{A} would normally be created before
11072 @code{B}, but the @code{init_priority} attribute has reversed that order:
11075 Some_Class A __attribute__ ((init_priority (2000)));
11076 Some_Class B __attribute__ ((init_priority (543)));
11080 Note that the particular values of @var{priority} do not matter; only their
11083 @item java_interface
11084 @cindex java_interface attribute
11086 This type attribute informs C++ that the class is a Java interface. It may
11087 only be applied to classes declared within an @code{extern "Java"} block.
11088 Calls to methods declared in this interface will be dispatched using GCJ's
11089 interface table mechanism, instead of regular virtual table dispatch.
11093 See also @xref{Namespace Association}.
11095 @node Namespace Association
11096 @section Namespace Association
11098 @strong{Caution:} The semantics of this extension are not fully
11099 defined. Users should refrain from using this extension as its
11100 semantics may change subtly over time. It is possible that this
11101 extension will be removed in future versions of G++.
11103 A using-directive with @code{__attribute ((strong))} is stronger
11104 than a normal using-directive in two ways:
11108 Templates from the used namespace can be specialized and explicitly
11109 instantiated as though they were members of the using namespace.
11112 The using namespace is considered an associated namespace of all
11113 templates in the used namespace for purposes of argument-dependent
11117 The used namespace must be nested within the using namespace so that
11118 normal unqualified lookup works properly.
11120 This is useful for composing a namespace transparently from
11121 implementation namespaces. For example:
11126 template <class T> struct A @{ @};
11128 using namespace debug __attribute ((__strong__));
11129 template <> struct A<int> @{ @}; // @r{ok to specialize}
11131 template <class T> void f (A<T>);
11136 f (std::A<float>()); // @r{lookup finds} std::f
11141 @node Java Exceptions
11142 @section Java Exceptions
11144 The Java language uses a slightly different exception handling model
11145 from C++. Normally, GNU C++ will automatically detect when you are
11146 writing C++ code that uses Java exceptions, and handle them
11147 appropriately. However, if C++ code only needs to execute destructors
11148 when Java exceptions are thrown through it, GCC will guess incorrectly.
11149 Sample problematic code is:
11152 struct S @{ ~S(); @};
11153 extern void bar(); // @r{is written in Java, and may throw exceptions}
11162 The usual effect of an incorrect guess is a link failure, complaining of
11163 a missing routine called @samp{__gxx_personality_v0}.
11165 You can inform the compiler that Java exceptions are to be used in a
11166 translation unit, irrespective of what it might think, by writing
11167 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
11168 @samp{#pragma} must appear before any functions that throw or catch
11169 exceptions, or run destructors when exceptions are thrown through them.
11171 You cannot mix Java and C++ exceptions in the same translation unit. It
11172 is believed to be safe to throw a C++ exception from one file through
11173 another file compiled for the Java exception model, or vice versa, but
11174 there may be bugs in this area.
11176 @node Deprecated Features
11177 @section Deprecated Features
11179 In the past, the GNU C++ compiler was extended to experiment with new
11180 features, at a time when the C++ language was still evolving. Now that
11181 the C++ standard is complete, some of those features are superseded by
11182 superior alternatives. Using the old features might cause a warning in
11183 some cases that the feature will be dropped in the future. In other
11184 cases, the feature might be gone already.
11186 While the list below is not exhaustive, it documents some of the options
11187 that are now deprecated:
11190 @item -fexternal-templates
11191 @itemx -falt-external-templates
11192 These are two of the many ways for G++ to implement template
11193 instantiation. @xref{Template Instantiation}. The C++ standard clearly
11194 defines how template definitions have to be organized across
11195 implementation units. G++ has an implicit instantiation mechanism that
11196 should work just fine for standard-conforming code.
11198 @item -fstrict-prototype
11199 @itemx -fno-strict-prototype
11200 Previously it was possible to use an empty prototype parameter list to
11201 indicate an unspecified number of parameters (like C), rather than no
11202 parameters, as C++ demands. This feature has been removed, except where
11203 it is required for backwards compatibility @xref{Backwards Compatibility}.
11206 G++ allows a virtual function returning @samp{void *} to be overridden
11207 by one returning a different pointer type. This extension to the
11208 covariant return type rules is now deprecated and will be removed from a
11211 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
11212 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
11213 and will be removed in a future version. Code using these operators
11214 should be modified to use @code{std::min} and @code{std::max} instead.
11216 The named return value extension has been deprecated, and is now
11219 The use of initializer lists with new expressions has been deprecated,
11220 and is now removed from G++.
11222 Floating and complex non-type template parameters have been deprecated,
11223 and are now removed from G++.
11225 The implicit typename extension has been deprecated and is now
11228 The use of default arguments in function pointers, function typedefs
11229 and other places where they are not permitted by the standard is
11230 deprecated and will be removed from a future version of G++.
11232 G++ allows floating-point literals to appear in integral constant expressions,
11233 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
11234 This extension is deprecated and will be removed from a future version.
11236 G++ allows static data members of const floating-point type to be declared
11237 with an initializer in a class definition. The standard only allows
11238 initializers for static members of const integral types and const
11239 enumeration types so this extension has been deprecated and will be removed
11240 from a future version.
11242 @node Backwards Compatibility
11243 @section Backwards Compatibility
11244 @cindex Backwards Compatibility
11245 @cindex ARM [Annotated C++ Reference Manual]
11247 Now that there is a definitive ISO standard C++, G++ has a specification
11248 to adhere to. The C++ language evolved over time, and features that
11249 used to be acceptable in previous drafts of the standard, such as the ARM
11250 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
11251 compilation of C++ written to such drafts, G++ contains some backwards
11252 compatibilities. @emph{All such backwards compatibility features are
11253 liable to disappear in future versions of G++.} They should be considered
11254 deprecated @xref{Deprecated Features}.
11258 If a variable is declared at for scope, it used to remain in scope until
11259 the end of the scope which contained the for statement (rather than just
11260 within the for scope). G++ retains this, but issues a warning, if such a
11261 variable is accessed outside the for scope.
11263 @item Implicit C language
11264 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
11265 scope to set the language. On such systems, all header files are
11266 implicitly scoped inside a C language scope. Also, an empty prototype
11267 @code{()} will be treated as an unspecified number of arguments, rather
11268 than no arguments, as C++ demands.