1 @c Copyright (C) 1988,1989,1992,1993,1994,1996,1998,1999,2000,2001,2002,2003,2004
2 @c Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
7 @chapter Extensions to the C Language Family
8 @cindex extensions, C language
9 @cindex C language extensions
12 GNU C provides several language features not found in ISO standard C@.
13 (The @option{-pedantic} option directs GCC to print a warning message if
14 any of these features is used.) To test for the availability of these
15 features in conditional compilation, check for a predefined macro
16 @code{__GNUC__}, which is always defined under GCC@.
18 These extensions are available in C and Objective-C@. Most of them are
19 also available in C++. @xref{C++ Extensions,,Extensions to the
20 C++ Language}, for extensions that apply @emph{only} to C++.
22 Some features that are in ISO C99 but not C89 or C++ are also, as
23 extensions, accepted by GCC in C89 mode and in C++.
26 * Statement Exprs:: Putting statements and declarations inside expressions.
27 * Local Labels:: Labels local to a block.
28 * Labels as Values:: Getting pointers to labels, and computed gotos.
29 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
30 * Constructing Calls:: Dispatching a call to another function.
31 * Typeof:: @code{typeof}: referring to the type of an expression.
32 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
33 * Long Long:: Double-word integers---@code{long long int}.
34 * Complex:: Data types for complex numbers.
35 * Hex Floats:: Hexadecimal floating-point constants.
36 * Zero Length:: Zero-length arrays.
37 * Variable Length:: Arrays whose length is computed at run time.
38 * Empty Structures:: Structures with no members.
39 * Variadic Macros:: Macros with a variable number of arguments.
40 * Escaped Newlines:: Slightly looser rules for escaped newlines.
41 * Subscripting:: Any array can be subscripted, even if not an lvalue.
42 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
43 * Initializers:: Non-constant initializers.
44 * Compound Literals:: Compound literals give structures, unions
46 * Designated Inits:: Labeling elements of initializers.
47 * Cast to Union:: Casting to union type from any member of the union.
48 * Case Ranges:: `case 1 ... 9' and such.
49 * Mixed Declarations:: Mixing declarations and code.
50 * Function Attributes:: Declaring that functions have no side effects,
51 or that they can never return.
52 * Attribute Syntax:: Formal syntax for attributes.
53 * Function Prototypes:: Prototype declarations and old-style definitions.
54 * C++ Comments:: C++ comments are recognized.
55 * Dollar Signs:: Dollar sign is allowed in identifiers.
56 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
57 * Variable Attributes:: Specifying attributes of variables.
58 * Type Attributes:: Specifying attributes of types.
59 * Alignment:: Inquiring about the alignment of a type or variable.
60 * Inline:: Defining inline functions (as fast as macros).
61 * Extended Asm:: Assembler instructions with C expressions as operands.
62 (With them you can define ``built-in'' functions.)
63 * Constraints:: Constraints for asm operands
64 * Asm Labels:: Specifying the assembler name to use for a C symbol.
65 * Explicit Reg Vars:: Defining variables residing in specified registers.
66 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
67 * Incomplete Enums:: @code{enum foo;}, with details to follow.
68 * Function Names:: Printable strings which are the name of the current
70 * Return Address:: Getting the return or frame address of a function.
71 * Vector Extensions:: Using vector instructions through built-in functions.
72 * Offsetof:: Special syntax for implementing @code{offsetof}.
73 * Other Builtins:: Other built-in functions.
74 * Target Builtins:: Built-in functions specific to particular targets.
75 * Target Format Checks:: Format checks specific to particular targets.
76 * Pragmas:: Pragmas accepted by GCC.
77 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
78 * Thread-Local:: Per-thread variables.
82 @section Statements and Declarations in Expressions
83 @cindex statements inside expressions
84 @cindex declarations inside expressions
85 @cindex expressions containing statements
86 @cindex macros, statements in expressions
88 @c the above section title wrapped and causes an underfull hbox.. i
89 @c changed it from "within" to "in". --mew 4feb93
90 A compound statement enclosed in parentheses may appear as an expression
91 in GNU C@. This allows you to use loops, switches, and local variables
94 Recall that a compound statement is a sequence of statements surrounded
95 by braces; in this construct, parentheses go around the braces. For
99 (@{ int y = foo (); int z;
106 is a valid (though slightly more complex than necessary) expression
107 for the absolute value of @code{foo ()}.
109 The last thing in the compound statement should be an expression
110 followed by a semicolon; the value of this subexpression serves as the
111 value of the entire construct. (If you use some other kind of statement
112 last within the braces, the construct has type @code{void}, and thus
113 effectively no value.)
115 This feature is especially useful in making macro definitions ``safe'' (so
116 that they evaluate each operand exactly once). For example, the
117 ``maximum'' function is commonly defined as a macro in standard C as
121 #define max(a,b) ((a) > (b) ? (a) : (b))
125 @cindex side effects, macro argument
126 But this definition computes either @var{a} or @var{b} twice, with bad
127 results if the operand has side effects. In GNU C, if you know the
128 type of the operands (here taken as @code{int}), you can define
129 the macro safely as follows:
132 #define maxint(a,b) \
133 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
136 Embedded statements are not allowed in constant expressions, such as
137 the value of an enumeration constant, the width of a bit-field, or
138 the initial value of a static variable.
140 If you don't know the type of the operand, you can still do this, but you
141 must use @code{typeof} (@pxref{Typeof}).
143 In G++, the result value of a statement expression undergoes array and
144 function pointer decay, and is returned by value to the enclosing
145 expression. For instance, if @code{A} is a class, then
154 will construct a temporary @code{A} object to hold the result of the
155 statement expression, and that will be used to invoke @code{Foo}.
156 Therefore the @code{this} pointer observed by @code{Foo} will not be the
159 Any temporaries created within a statement within a statement expression
160 will be destroyed at the statement's end. This makes statement
161 expressions inside macros slightly different from function calls. In
162 the latter case temporaries introduced during argument evaluation will
163 be destroyed at the end of the statement that includes the function
164 call. In the statement expression case they will be destroyed during
165 the statement expression. For instance,
168 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
169 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
179 will have different places where temporaries are destroyed. For the
180 @code{macro} case, the temporary @code{X} will be destroyed just after
181 the initialization of @code{b}. In the @code{function} case that
182 temporary will be destroyed when the function returns.
184 These considerations mean that it is probably a bad idea to use
185 statement-expressions of this form in header files that are designed to
186 work with C++. (Note that some versions of the GNU C Library contained
187 header files using statement-expression that lead to precisely this
191 @section Locally Declared Labels
193 @cindex macros, local labels
195 GCC allows you to declare @dfn{local labels} in any nested block
196 scope. A local label is just like an ordinary label, but you can
197 only reference it (with a @code{goto} statement, or by taking its
198 address) within the block in which it was declared.
200 A local label declaration looks like this:
203 __label__ @var{label};
210 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
213 Local label declarations must come at the beginning of the block,
214 before any ordinary declarations or statements.
216 The label declaration defines the label @emph{name}, but does not define
217 the label itself. You must do this in the usual way, with
218 @code{@var{label}:}, within the statements of the statement expression.
220 The local label feature is useful for complex macros. If a macro
221 contains nested loops, a @code{goto} can be useful for breaking out of
222 them. However, an ordinary label whose scope is the whole function
223 cannot be used: if the macro can be expanded several times in one
224 function, the label will be multiply defined in that function. A
225 local label avoids this problem. For example:
228 #define SEARCH(value, array, target) \
231 typeof (target) _SEARCH_target = (target); \
232 typeof (*(array)) *_SEARCH_array = (array); \
235 for (i = 0; i < max; i++) \
236 for (j = 0; j < max; j++) \
237 if (_SEARCH_array[i][j] == _SEARCH_target) \
238 @{ (value) = i; goto found; @} \
244 This could also be written using a statement-expression:
247 #define SEARCH(array, target) \
250 typeof (target) _SEARCH_target = (target); \
251 typeof (*(array)) *_SEARCH_array = (array); \
254 for (i = 0; i < max; i++) \
255 for (j = 0; j < max; j++) \
256 if (_SEARCH_array[i][j] == _SEARCH_target) \
257 @{ value = i; goto found; @} \
264 Local label declarations also make the labels they declare visible to
265 nested functions, if there are any. @xref{Nested Functions}, for details.
267 @node Labels as Values
268 @section Labels as Values
269 @cindex labels as values
270 @cindex computed gotos
271 @cindex goto with computed label
272 @cindex address of a label
274 You can get the address of a label defined in the current function
275 (or a containing function) with the unary operator @samp{&&}. The
276 value has type @code{void *}. This value is a constant and can be used
277 wherever a constant of that type is valid. For example:
285 To use these values, you need to be able to jump to one. This is done
286 with the computed goto statement@footnote{The analogous feature in
287 Fortran is called an assigned goto, but that name seems inappropriate in
288 C, where one can do more than simply store label addresses in label
289 variables.}, @code{goto *@var{exp};}. For example,
296 Any expression of type @code{void *} is allowed.
298 One way of using these constants is in initializing a static array that
299 will serve as a jump table:
302 static void *array[] = @{ &&foo, &&bar, &&hack @};
305 Then you can select a label with indexing, like this:
312 Note that this does not check whether the subscript is in bounds---array
313 indexing in C never does that.
315 Such an array of label values serves a purpose much like that of the
316 @code{switch} statement. The @code{switch} statement is cleaner, so
317 use that rather than an array unless the problem does not fit a
318 @code{switch} statement very well.
320 Another use of label values is in an interpreter for threaded code.
321 The labels within the interpreter function can be stored in the
322 threaded code for super-fast dispatching.
324 You may not use this mechanism to jump to code in a different function.
325 If you do that, totally unpredictable things will happen. The best way to
326 avoid this is to store the label address only in automatic variables and
327 never pass it as an argument.
329 An alternate way to write the above example is
332 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
334 goto *(&&foo + array[i]);
338 This is more friendly to code living in shared libraries, as it reduces
339 the number of dynamic relocations that are needed, and by consequence,
340 allows the data to be read-only.
342 @node Nested Functions
343 @section Nested Functions
344 @cindex nested functions
345 @cindex downward funargs
348 A @dfn{nested function} is a function defined inside another function.
349 (Nested functions are not supported for GNU C++.) The nested function's
350 name is local to the block where it is defined. For example, here we
351 define a nested function named @code{square}, and call it twice:
355 foo (double a, double b)
357 double square (double z) @{ return z * z; @}
359 return square (a) + square (b);
364 The nested function can access all the variables of the containing
365 function that are visible at the point of its definition. This is
366 called @dfn{lexical scoping}. For example, here we show a nested
367 function which uses an inherited variable named @code{offset}:
371 bar (int *array, int offset, int size)
373 int access (int *array, int index)
374 @{ return array[index + offset]; @}
377 for (i = 0; i < size; i++)
378 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
383 Nested function definitions are permitted within functions in the places
384 where variable definitions are allowed; that is, in any block, before
385 the first statement in the block.
387 It is possible to call the nested function from outside the scope of its
388 name by storing its address or passing the address to another function:
391 hack (int *array, int size)
393 void store (int index, int value)
394 @{ array[index] = value; @}
396 intermediate (store, size);
400 Here, the function @code{intermediate} receives the address of
401 @code{store} as an argument. If @code{intermediate} calls @code{store},
402 the arguments given to @code{store} are used to store into @code{array}.
403 But this technique works only so long as the containing function
404 (@code{hack}, in this example) does not exit.
406 If you try to call the nested function through its address after the
407 containing function has exited, all hell will break loose. If you try
408 to call it after a containing scope level has exited, and if it refers
409 to some of the variables that are no longer in scope, you may be lucky,
410 but it's not wise to take the risk. If, however, the nested function
411 does not refer to anything that has gone out of scope, you should be
414 GCC implements taking the address of a nested function using a technique
415 called @dfn{trampolines}. A paper describing them is available as
418 @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}.
420 A nested function can jump to a label inherited from a containing
421 function, provided the label was explicitly declared in the containing
422 function (@pxref{Local Labels}). Such a jump returns instantly to the
423 containing function, exiting the nested function which did the
424 @code{goto} and any intermediate functions as well. Here is an example:
428 bar (int *array, int offset, int size)
431 int access (int *array, int index)
435 return array[index + offset];
439 for (i = 0; i < size; i++)
440 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
444 /* @r{Control comes here from @code{access}
445 if it detects an error.} */
452 A nested function always has internal linkage. Declaring one with
453 @code{extern} is erroneous. If you need to declare the nested function
454 before its definition, use @code{auto} (which is otherwise meaningless
455 for function declarations).
458 bar (int *array, int offset, int size)
461 auto int access (int *, int);
463 int access (int *array, int index)
467 return array[index + offset];
473 @node Constructing Calls
474 @section Constructing Function Calls
475 @cindex constructing calls
476 @cindex forwarding calls
478 Using the built-in functions described below, you can record
479 the arguments a function received, and call another function
480 with the same arguments, without knowing the number or types
483 You can also record the return value of that function call,
484 and later return that value, without knowing what data type
485 the function tried to return (as long as your caller expects
488 However, these built-in functions may interact badly with some
489 sophisticated features or other extensions of the language. It
490 is, therefore, not recommended to use them outside very simple
491 functions acting as mere forwarders for their arguments.
493 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
494 This built-in function returns a pointer to data
495 describing how to perform a call with the same arguments as were passed
496 to the current function.
498 The function saves the arg pointer register, structure value address,
499 and all registers that might be used to pass arguments to a function
500 into a block of memory allocated on the stack. Then it returns the
501 address of that block.
504 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
505 This built-in function invokes @var{function}
506 with a copy of the parameters described by @var{arguments}
509 The value of @var{arguments} should be the value returned by
510 @code{__builtin_apply_args}. The argument @var{size} specifies the size
511 of the stack argument data, in bytes.
513 This function returns a pointer to data describing
514 how to return whatever value was returned by @var{function}. The data
515 is saved in a block of memory allocated on the stack.
517 It is not always simple to compute the proper value for @var{size}. The
518 value is used by @code{__builtin_apply} to compute the amount of data
519 that should be pushed on the stack and copied from the incoming argument
523 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
524 This built-in function returns the value described by @var{result} from
525 the containing function. You should specify, for @var{result}, a value
526 returned by @code{__builtin_apply}.
530 @section Referring to a Type with @code{typeof}
533 @cindex macros, types of arguments
535 Another way to refer to the type of an expression is with @code{typeof}.
536 The syntax of using of this keyword looks like @code{sizeof}, but the
537 construct acts semantically like a type name defined with @code{typedef}.
539 There are two ways of writing the argument to @code{typeof}: with an
540 expression or with a type. Here is an example with an expression:
547 This assumes that @code{x} is an array of pointers to functions;
548 the type described is that of the values of the functions.
550 Here is an example with a typename as the argument:
557 Here the type described is that of pointers to @code{int}.
559 If you are writing a header file that must work when included in ISO C
560 programs, write @code{__typeof__} instead of @code{typeof}.
561 @xref{Alternate Keywords}.
563 A @code{typeof}-construct can be used anywhere a typedef name could be
564 used. For example, you can use it in a declaration, in a cast, or inside
565 of @code{sizeof} or @code{typeof}.
567 @code{typeof} is often useful in conjunction with the
568 statements-within-expressions feature. Here is how the two together can
569 be used to define a safe ``maximum'' macro that operates on any
570 arithmetic type and evaluates each of its arguments exactly once:
574 (@{ typeof (a) _a = (a); \
575 typeof (b) _b = (b); \
576 _a > _b ? _a : _b; @})
579 @cindex underscores in variables in macros
580 @cindex @samp{_} in variables in macros
581 @cindex local variables in macros
582 @cindex variables, local, in macros
583 @cindex macros, local variables in
585 The reason for using names that start with underscores for the local
586 variables is to avoid conflicts with variable names that occur within the
587 expressions that are substituted for @code{a} and @code{b}. Eventually we
588 hope to design a new form of declaration syntax that allows you to declare
589 variables whose scopes start only after their initializers; this will be a
590 more reliable way to prevent such conflicts.
593 Some more examples of the use of @code{typeof}:
597 This declares @code{y} with the type of what @code{x} points to.
604 This declares @code{y} as an array of such values.
611 This declares @code{y} as an array of pointers to characters:
614 typeof (typeof (char *)[4]) y;
618 It is equivalent to the following traditional C declaration:
624 To see the meaning of the declaration using @code{typeof}, and why it
625 might be a useful way to write, rewrite it with these macros:
628 #define pointer(T) typeof(T *)
629 #define array(T, N) typeof(T [N])
633 Now the declaration can be rewritten this way:
636 array (pointer (char), 4) y;
640 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
641 pointers to @code{char}.
644 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
645 a more limited extension which permitted one to write
648 typedef @var{T} = @var{expr};
652 with the effect of declaring @var{T} to have the type of the expression
653 @var{expr}. This extension does not work with GCC 3 (versions between
654 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
655 relies on it should be rewritten to use @code{typeof}:
658 typedef typeof(@var{expr}) @var{T};
662 This will work with all versions of GCC@.
665 @section Conditionals with Omitted Operands
666 @cindex conditional expressions, extensions
667 @cindex omitted middle-operands
668 @cindex middle-operands, omitted
669 @cindex extensions, @code{?:}
670 @cindex @code{?:} extensions
672 The middle operand in a conditional expression may be omitted. Then
673 if the first operand is nonzero, its value is the value of the conditional
676 Therefore, the expression
683 has the value of @code{x} if that is nonzero; otherwise, the value of
686 This example is perfectly equivalent to
692 @cindex side effect in ?:
693 @cindex ?: side effect
695 In this simple case, the ability to omit the middle operand is not
696 especially useful. When it becomes useful is when the first operand does,
697 or may (if it is a macro argument), contain a side effect. Then repeating
698 the operand in the middle would perform the side effect twice. Omitting
699 the middle operand uses the value already computed without the undesirable
700 effects of recomputing it.
703 @section Double-Word Integers
704 @cindex @code{long long} data types
705 @cindex double-word arithmetic
706 @cindex multiprecision arithmetic
707 @cindex @code{LL} integer suffix
708 @cindex @code{ULL} integer suffix
710 ISO C99 supports data types for integers that are at least 64 bits wide,
711 and as an extension GCC supports them in C89 mode and in C++.
712 Simply write @code{long long int} for a signed integer, or
713 @code{unsigned long long int} for an unsigned integer. To make an
714 integer constant of type @code{long long int}, add the suffix @samp{LL}
715 to the integer. To make an integer constant of type @code{unsigned long
716 long int}, add the suffix @samp{ULL} to the integer.
718 You can use these types in arithmetic like any other integer types.
719 Addition, subtraction, and bitwise boolean operations on these types
720 are open-coded on all types of machines. Multiplication is open-coded
721 if the machine supports fullword-to-doubleword a widening multiply
722 instruction. Division and shifts are open-coded only on machines that
723 provide special support. The operations that are not open-coded use
724 special library routines that come with GCC@.
726 There may be pitfalls when you use @code{long long} types for function
727 arguments, unless you declare function prototypes. If a function
728 expects type @code{int} for its argument, and you pass a value of type
729 @code{long long int}, confusion will result because the caller and the
730 subroutine will disagree about the number of bytes for the argument.
731 Likewise, if the function expects @code{long long int} and you pass
732 @code{int}. The best way to avoid such problems is to use prototypes.
735 @section Complex Numbers
736 @cindex complex numbers
737 @cindex @code{_Complex} keyword
738 @cindex @code{__complex__} keyword
740 ISO C99 supports complex floating data types, and as an extension GCC
741 supports them in C89 mode and in C++, and supports complex integer data
742 types which are not part of ISO C99. You can declare complex types
743 using the keyword @code{_Complex}. As an extension, the older GNU
744 keyword @code{__complex__} is also supported.
746 For example, @samp{_Complex double x;} declares @code{x} as a
747 variable whose real part and imaginary part are both of type
748 @code{double}. @samp{_Complex short int y;} declares @code{y} to
749 have real and imaginary parts of type @code{short int}; this is not
750 likely to be useful, but it shows that the set of complex types is
753 To write a constant with a complex data type, use the suffix @samp{i} or
754 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
755 has type @code{_Complex float} and @code{3i} has type
756 @code{_Complex int}. Such a constant always has a pure imaginary
757 value, but you can form any complex value you like by adding one to a
758 real constant. This is a GNU extension; if you have an ISO C99
759 conforming C library (such as GNU libc), and want to construct complex
760 constants of floating type, you should include @code{<complex.h>} and
761 use the macros @code{I} or @code{_Complex_I} instead.
763 @cindex @code{__real__} keyword
764 @cindex @code{__imag__} keyword
765 To extract the real part of a complex-valued expression @var{exp}, write
766 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
767 extract the imaginary part. This is a GNU extension; for values of
768 floating type, you should use the ISO C99 functions @code{crealf},
769 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
770 @code{cimagl}, declared in @code{<complex.h>} and also provided as
771 built-in functions by GCC@.
773 @cindex complex conjugation
774 The operator @samp{~} performs complex conjugation when used on a value
775 with a complex type. This is a GNU extension; for values of
776 floating type, you should use the ISO C99 functions @code{conjf},
777 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
778 provided as built-in functions by GCC@.
780 GCC can allocate complex automatic variables in a noncontiguous
781 fashion; it's even possible for the real part to be in a register while
782 the imaginary part is on the stack (or vice-versa). Only the DWARF2
783 debug info format can represent this, so use of DWARF2 is recommended.
784 If you are using the stabs debug info format, GCC describes a noncontiguous
785 complex variable as if it were two separate variables of noncomplex type.
786 If the variable's actual name is @code{foo}, the two fictitious
787 variables are named @code{foo$real} and @code{foo$imag}. You can
788 examine and set these two fictitious variables with your debugger.
794 ISO C99 supports floating-point numbers written not only in the usual
795 decimal notation, such as @code{1.55e1}, but also numbers such as
796 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
797 supports this in C89 mode (except in some cases when strictly
798 conforming) and in C++. In that format the
799 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
800 mandatory. The exponent is a decimal number that indicates the power of
801 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
808 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
809 is the same as @code{1.55e1}.
811 Unlike for floating-point numbers in the decimal notation the exponent
812 is always required in the hexadecimal notation. Otherwise the compiler
813 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
814 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
815 extension for floating-point constants of type @code{float}.
818 @section Arrays of Length Zero
819 @cindex arrays of length zero
820 @cindex zero-length arrays
821 @cindex length-zero arrays
822 @cindex flexible array members
824 Zero-length arrays are allowed in GNU C@. They are very useful as the
825 last element of a structure which is really a header for a variable-length
834 struct line *thisline = (struct line *)
835 malloc (sizeof (struct line) + this_length);
836 thisline->length = this_length;
839 In ISO C90, you would have to give @code{contents} a length of 1, which
840 means either you waste space or complicate the argument to @code{malloc}.
842 In ISO C99, you would use a @dfn{flexible array member}, which is
843 slightly different in syntax and semantics:
847 Flexible array members are written as @code{contents[]} without
851 Flexible array members have incomplete type, and so the @code{sizeof}
852 operator may not be applied. As a quirk of the original implementation
853 of zero-length arrays, @code{sizeof} evaluates to zero.
856 Flexible array members may only appear as the last member of a
857 @code{struct} that is otherwise non-empty.
860 A structure containing a flexible array member, or a union containing
861 such a structure (possibly recursively), may not be a member of a
862 structure or an element of an array. (However, these uses are
863 permitted by GCC as extensions.)
866 GCC versions before 3.0 allowed zero-length arrays to be statically
867 initialized, as if they were flexible arrays. In addition to those
868 cases that were useful, it also allowed initializations in situations
869 that would corrupt later data. Non-empty initialization of zero-length
870 arrays is now treated like any case where there are more initializer
871 elements than the array holds, in that a suitable warning about "excess
872 elements in array" is given, and the excess elements (all of them, in
873 this case) are ignored.
875 Instead GCC allows static initialization of flexible array members.
876 This is equivalent to defining a new structure containing the original
877 structure followed by an array of sufficient size to contain the data.
878 I.e.@: in the following, @code{f1} is constructed as if it were declared
884 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
887 struct f1 f1; int data[3];
888 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
892 The convenience of this extension is that @code{f1} has the desired
893 type, eliminating the need to consistently refer to @code{f2.f1}.
895 This has symmetry with normal static arrays, in that an array of
896 unknown size is also written with @code{[]}.
898 Of course, this extension only makes sense if the extra data comes at
899 the end of a top-level object, as otherwise we would be overwriting
900 data at subsequent offsets. To avoid undue complication and confusion
901 with initialization of deeply nested arrays, we simply disallow any
902 non-empty initialization except when the structure is the top-level
906 struct foo @{ int x; int y[]; @};
907 struct bar @{ struct foo z; @};
909 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
910 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
911 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
912 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
915 @node Empty Structures
916 @section Structures With No Members
917 @cindex empty structures
918 @cindex zero-size structures
920 GCC permits a C structure to have no members:
927 The structure will have size zero. In C++, empty structures are part
928 of the language. G++ treats empty structures as if they had a single
929 member of type @code{char}.
931 @node Variable Length
932 @section Arrays of Variable Length
933 @cindex variable-length arrays
934 @cindex arrays of variable length
937 Variable-length automatic arrays are allowed in ISO C99, and as an
938 extension GCC accepts them in C89 mode and in C++. (However, GCC's
939 implementation of variable-length arrays does not yet conform in detail
940 to the ISO C99 standard.) These arrays are
941 declared like any other automatic arrays, but with a length that is not
942 a constant expression. The storage is allocated at the point of
943 declaration and deallocated when the brace-level is exited. For
948 concat_fopen (char *s1, char *s2, char *mode)
950 char str[strlen (s1) + strlen (s2) + 1];
953 return fopen (str, mode);
957 @cindex scope of a variable length array
958 @cindex variable-length array scope
959 @cindex deallocating variable length arrays
960 Jumping or breaking out of the scope of the array name deallocates the
961 storage. Jumping into the scope is not allowed; you get an error
964 @cindex @code{alloca} vs variable-length arrays
965 You can use the function @code{alloca} to get an effect much like
966 variable-length arrays. The function @code{alloca} is available in
967 many other C implementations (but not in all). On the other hand,
968 variable-length arrays are more elegant.
970 There are other differences between these two methods. Space allocated
971 with @code{alloca} exists until the containing @emph{function} returns.
972 The space for a variable-length array is deallocated as soon as the array
973 name's scope ends. (If you use both variable-length arrays and
974 @code{alloca} in the same function, deallocation of a variable-length array
975 will also deallocate anything more recently allocated with @code{alloca}.)
977 You can also use variable-length arrays as arguments to functions:
981 tester (int len, char data[len][len])
987 The length of an array is computed once when the storage is allocated
988 and is remembered for the scope of the array in case you access it with
991 If you want to pass the array first and the length afterward, you can
992 use a forward declaration in the parameter list---another GNU extension.
996 tester (int len; char data[len][len], int len)
1002 @cindex parameter forward declaration
1003 The @samp{int len} before the semicolon is a @dfn{parameter forward
1004 declaration}, and it serves the purpose of making the name @code{len}
1005 known when the declaration of @code{data} is parsed.
1007 You can write any number of such parameter forward declarations in the
1008 parameter list. They can be separated by commas or semicolons, but the
1009 last one must end with a semicolon, which is followed by the ``real''
1010 parameter declarations. Each forward declaration must match a ``real''
1011 declaration in parameter name and data type. ISO C99 does not support
1012 parameter forward declarations.
1014 @node Variadic Macros
1015 @section Macros with a Variable Number of Arguments.
1016 @cindex variable number of arguments
1017 @cindex macro with variable arguments
1018 @cindex rest argument (in macro)
1019 @cindex variadic macros
1021 In the ISO C standard of 1999, a macro can be declared to accept a
1022 variable number of arguments much as a function can. The syntax for
1023 defining the macro is similar to that of a function. Here is an
1027 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1030 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1031 such a macro, it represents the zero or more tokens until the closing
1032 parenthesis that ends the invocation, including any commas. This set of
1033 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1034 wherever it appears. See the CPP manual for more information.
1036 GCC has long supported variadic macros, and used a different syntax that
1037 allowed you to give a name to the variable arguments just like any other
1038 argument. Here is an example:
1041 #define debug(format, args...) fprintf (stderr, format, args)
1044 This is in all ways equivalent to the ISO C example above, but arguably
1045 more readable and descriptive.
1047 GNU CPP has two further variadic macro extensions, and permits them to
1048 be used with either of the above forms of macro definition.
1050 In standard C, you are not allowed to leave the variable argument out
1051 entirely; but you are allowed to pass an empty argument. For example,
1052 this invocation is invalid in ISO C, because there is no comma after
1059 GNU CPP permits you to completely omit the variable arguments in this
1060 way. In the above examples, the compiler would complain, though since
1061 the expansion of the macro still has the extra comma after the format
1064 To help solve this problem, CPP behaves specially for variable arguments
1065 used with the token paste operator, @samp{##}. If instead you write
1068 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1071 and if the variable arguments are omitted or empty, the @samp{##}
1072 operator causes the preprocessor to remove the comma before it. If you
1073 do provide some variable arguments in your macro invocation, GNU CPP
1074 does not complain about the paste operation and instead places the
1075 variable arguments after the comma. Just like any other pasted macro
1076 argument, these arguments are not macro expanded.
1078 @node Escaped Newlines
1079 @section Slightly Looser Rules for Escaped Newlines
1080 @cindex escaped newlines
1081 @cindex newlines (escaped)
1083 Recently, the preprocessor has relaxed its treatment of escaped
1084 newlines. Previously, the newline had to immediately follow a
1085 backslash. The current implementation allows whitespace in the form
1086 of spaces, horizontal and vertical tabs, and form feeds between the
1087 backslash and the subsequent newline. The preprocessor issues a
1088 warning, but treats it as a valid escaped newline and combines the two
1089 lines to form a single logical line. This works within comments and
1090 tokens, as well as between tokens. Comments are @emph{not} treated as
1091 whitespace for the purposes of this relaxation, since they have not
1092 yet been replaced with spaces.
1095 @section Non-Lvalue Arrays May Have Subscripts
1096 @cindex subscripting
1097 @cindex arrays, non-lvalue
1099 @cindex subscripting and function values
1100 In ISO C99, arrays that are not lvalues still decay to pointers, and
1101 may be subscripted, although they may not be modified or used after
1102 the next sequence point and the unary @samp{&} operator may not be
1103 applied to them. As an extension, GCC allows such arrays to be
1104 subscripted in C89 mode, though otherwise they do not decay to
1105 pointers outside C99 mode. For example,
1106 this is valid in GNU C though not valid in C89:
1110 struct foo @{int a[4];@};
1116 return f().a[index];
1122 @section Arithmetic on @code{void}- and Function-Pointers
1123 @cindex void pointers, arithmetic
1124 @cindex void, size of pointer to
1125 @cindex function pointers, arithmetic
1126 @cindex function, size of pointer to
1128 In GNU C, addition and subtraction operations are supported on pointers to
1129 @code{void} and on pointers to functions. This is done by treating the
1130 size of a @code{void} or of a function as 1.
1132 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1133 and on function types, and returns 1.
1135 @opindex Wpointer-arith
1136 The option @option{-Wpointer-arith} requests a warning if these extensions
1140 @section Non-Constant Initializers
1141 @cindex initializers, non-constant
1142 @cindex non-constant initializers
1144 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1145 automatic variable are not required to be constant expressions in GNU C@.
1146 Here is an example of an initializer with run-time varying elements:
1149 foo (float f, float g)
1151 float beat_freqs[2] = @{ f-g, f+g @};
1156 @node Compound Literals
1157 @section Compound Literals
1158 @cindex constructor expressions
1159 @cindex initializations in expressions
1160 @cindex structures, constructor expression
1161 @cindex expressions, constructor
1162 @cindex compound literals
1163 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1165 ISO C99 supports compound literals. A compound literal looks like
1166 a cast containing an initializer. Its value is an object of the
1167 type specified in the cast, containing the elements specified in
1168 the initializer; it is an lvalue. As an extension, GCC supports
1169 compound literals in C89 mode and in C++.
1171 Usually, the specified type is a structure. Assume that
1172 @code{struct foo} and @code{structure} are declared as shown:
1175 struct foo @{int a; char b[2];@} structure;
1179 Here is an example of constructing a @code{struct foo} with a compound literal:
1182 structure = ((struct foo) @{x + y, 'a', 0@});
1186 This is equivalent to writing the following:
1190 struct foo temp = @{x + y, 'a', 0@};
1195 You can also construct an array. If all the elements of the compound literal
1196 are (made up of) simple constant expressions, suitable for use in
1197 initializers of objects of static storage duration, then the compound
1198 literal can be coerced to a pointer to its first element and used in
1199 such an initializer, as shown here:
1202 char **foo = (char *[]) @{ "x", "y", "z" @};
1205 Compound literals for scalar types and union types are is
1206 also allowed, but then the compound literal is equivalent
1209 As a GNU extension, GCC allows initialization of objects with static storage
1210 duration by compound literals (which is not possible in ISO C99, because
1211 the initializer is not a constant).
1212 It is handled as if the object was initialized only with the bracket
1213 enclosed list if compound literal's and object types match.
1214 The initializer list of the compound literal must be constant.
1215 If the object being initialized has array type of unknown size, the size is
1216 determined by compound literal size.
1219 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1220 static int y[] = (int []) @{1, 2, 3@};
1221 static int z[] = (int [3]) @{1@};
1225 The above lines are equivalent to the following:
1227 static struct foo x = @{1, 'a', 'b'@};
1228 static int y[] = @{1, 2, 3@};
1229 static int z[] = @{1, 0, 0@};
1232 @node Designated Inits
1233 @section Designated Initializers
1234 @cindex initializers with labeled elements
1235 @cindex labeled elements in initializers
1236 @cindex case labels in initializers
1237 @cindex designated initializers
1239 Standard C89 requires the elements of an initializer to appear in a fixed
1240 order, the same as the order of the elements in the array or structure
1243 In ISO C99 you can give the elements in any order, specifying the array
1244 indices or structure field names they apply to, and GNU C allows this as
1245 an extension in C89 mode as well. This extension is not
1246 implemented in GNU C++.
1248 To specify an array index, write
1249 @samp{[@var{index}] =} before the element value. For example,
1252 int a[6] = @{ [4] = 29, [2] = 15 @};
1259 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1263 The index values must be constant expressions, even if the array being
1264 initialized is automatic.
1266 An alternative syntax for this which has been obsolete since GCC 2.5 but
1267 GCC still accepts is to write @samp{[@var{index}]} before the element
1268 value, with no @samp{=}.
1270 To initialize a range of elements to the same value, write
1271 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1272 extension. For example,
1275 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1279 If the value in it has side-effects, the side-effects will happen only once,
1280 not for each initialized field by the range initializer.
1283 Note that the length of the array is the highest value specified
1286 In a structure initializer, specify the name of a field to initialize
1287 with @samp{.@var{fieldname} =} before the element value. For example,
1288 given the following structure,
1291 struct point @{ int x, y; @};
1295 the following initialization
1298 struct point p = @{ .y = yvalue, .x = xvalue @};
1305 struct point p = @{ xvalue, yvalue @};
1308 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1309 @samp{@var{fieldname}:}, as shown here:
1312 struct point p = @{ y: yvalue, x: xvalue @};
1316 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1317 @dfn{designator}. You can also use a designator (or the obsolete colon
1318 syntax) when initializing a union, to specify which element of the union
1319 should be used. For example,
1322 union foo @{ int i; double d; @};
1324 union foo f = @{ .d = 4 @};
1328 will convert 4 to a @code{double} to store it in the union using
1329 the second element. By contrast, casting 4 to type @code{union foo}
1330 would store it into the union as the integer @code{i}, since it is
1331 an integer. (@xref{Cast to Union}.)
1333 You can combine this technique of naming elements with ordinary C
1334 initialization of successive elements. Each initializer element that
1335 does not have a designator applies to the next consecutive element of the
1336 array or structure. For example,
1339 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1346 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1349 Labeling the elements of an array initializer is especially useful
1350 when the indices are characters or belong to an @code{enum} type.
1355 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1356 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1359 @cindex designator lists
1360 You can also write a series of @samp{.@var{fieldname}} and
1361 @samp{[@var{index}]} designators before an @samp{=} to specify a
1362 nested subobject to initialize; the list is taken relative to the
1363 subobject corresponding to the closest surrounding brace pair. For
1364 example, with the @samp{struct point} declaration above:
1367 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1371 If the same field is initialized multiple times, it will have value from
1372 the last initialization. If any such overridden initialization has
1373 side-effect, it is unspecified whether the side-effect happens or not.
1374 Currently, GCC will discard them and issue a warning.
1377 @section Case Ranges
1379 @cindex ranges in case statements
1381 You can specify a range of consecutive values in a single @code{case} label,
1385 case @var{low} ... @var{high}:
1389 This has the same effect as the proper number of individual @code{case}
1390 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1392 This feature is especially useful for ranges of ASCII character codes:
1398 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1399 it may be parsed wrong when you use it with integer values. For example,
1414 @section Cast to a Union Type
1415 @cindex cast to a union
1416 @cindex union, casting to a
1418 A cast to union type is similar to other casts, except that the type
1419 specified is a union type. You can specify the type either with
1420 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1421 a constructor though, not a cast, and hence does not yield an lvalue like
1422 normal casts. (@xref{Compound Literals}.)
1424 The types that may be cast to the union type are those of the members
1425 of the union. Thus, given the following union and variables:
1428 union foo @{ int i; double d; @};
1434 both @code{x} and @code{y} can be cast to type @code{union foo}.
1436 Using the cast as the right-hand side of an assignment to a variable of
1437 union type is equivalent to storing in a member of the union:
1442 u = (union foo) x @equiv{} u.i = x
1443 u = (union foo) y @equiv{} u.d = y
1446 You can also use the union cast as a function argument:
1449 void hack (union foo);
1451 hack ((union foo) x);
1454 @node Mixed Declarations
1455 @section Mixed Declarations and Code
1456 @cindex mixed declarations and code
1457 @cindex declarations, mixed with code
1458 @cindex code, mixed with declarations
1460 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1461 within compound statements. As an extension, GCC also allows this in
1462 C89 mode. For example, you could do:
1471 Each identifier is visible from where it is declared until the end of
1472 the enclosing block.
1474 @node Function Attributes
1475 @section Declaring Attributes of Functions
1476 @cindex function attributes
1477 @cindex declaring attributes of functions
1478 @cindex functions that never return
1479 @cindex functions that have no side effects
1480 @cindex functions in arbitrary sections
1481 @cindex functions that behave like malloc
1482 @cindex @code{volatile} applied to function
1483 @cindex @code{const} applied to function
1484 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1485 @cindex functions with non-null pointer arguments
1486 @cindex functions that are passed arguments in registers on the 386
1487 @cindex functions that pop the argument stack on the 386
1488 @cindex functions that do not pop the argument stack on the 386
1490 In GNU C, you declare certain things about functions called in your program
1491 which help the compiler optimize function calls and check your code more
1494 The keyword @code{__attribute__} allows you to specify special
1495 attributes when making a declaration. This keyword is followed by an
1496 attribute specification inside double parentheses. The following
1497 attributes are currently defined for functions on all targets:
1498 @code{noreturn}, @code{noinline}, @code{always_inline},
1499 @code{pure}, @code{const}, @code{nothrow}, @code{sentinel},
1500 @code{format}, @code{format_arg}, @code{no_instrument_function},
1501 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1502 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
1503 @code{alias}, @code{warn_unused_result} and @code{nonnull}. Several other
1504 attributes are defined for functions on particular target systems. Other
1505 attributes, including @code{section} are supported for variables declarations
1506 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1508 You may also specify attributes with @samp{__} preceding and following
1509 each keyword. This allows you to use them in header files without
1510 being concerned about a possible macro of the same name. For example,
1511 you may use @code{__noreturn__} instead of @code{noreturn}.
1513 @xref{Attribute Syntax}, for details of the exact syntax for using
1517 @c Keep this table alphabetized by attribute name. Treat _ as space.
1519 @item alias ("@var{target}")
1520 @cindex @code{alias} attribute
1521 The @code{alias} attribute causes the declaration to be emitted as an
1522 alias for another symbol, which must be specified. For instance,
1525 void __f () @{ /* @r{Do something.} */; @}
1526 void f () __attribute__ ((weak, alias ("__f")));
1529 declares @samp{f} to be a weak alias for @samp{__f}. In C++, the
1530 mangled name for the target must be used.
1532 Not all target machines support this attribute.
1535 @cindex @code{always_inline} function attribute
1536 Generally, functions are not inlined unless optimization is specified.
1537 For functions declared inline, this attribute inlines the function even
1538 if no optimization level was specified.
1541 @cindex functions that do pop the argument stack on the 386
1543 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1544 assume that the calling function will pop off the stack space used to
1545 pass arguments. This is
1546 useful to override the effects of the @option{-mrtd} switch.
1549 @cindex @code{const} function attribute
1550 Many functions do not examine any values except their arguments, and
1551 have no effects except the return value. Basically this is just slightly
1552 more strict class than the @code{pure} attribute above, since function is not
1553 allowed to read global memory.
1555 @cindex pointer arguments
1556 Note that a function that has pointer arguments and examines the data
1557 pointed to must @emph{not} be declared @code{const}. Likewise, a
1558 function that calls a non-@code{const} function usually must not be
1559 @code{const}. It does not make sense for a @code{const} function to
1562 The attribute @code{const} is not implemented in GCC versions earlier
1563 than 2.5. An alternative way to declare that a function has no side
1564 effects, which works in the current version and in some older versions,
1568 typedef int intfn ();
1570 extern const intfn square;
1573 This approach does not work in GNU C++ from 2.6.0 on, since the language
1574 specifies that the @samp{const} must be attached to the return value.
1578 @cindex @code{constructor} function attribute
1579 @cindex @code{destructor} function attribute
1580 The @code{constructor} attribute causes the function to be called
1581 automatically before execution enters @code{main ()}. Similarly, the
1582 @code{destructor} attribute causes the function to be called
1583 automatically after @code{main ()} has completed or @code{exit ()} has
1584 been called. Functions with these attributes are useful for
1585 initializing data that will be used implicitly during the execution of
1588 These attributes are not currently implemented for Objective-C@.
1591 @cindex @code{deprecated} attribute.
1592 The @code{deprecated} attribute results in a warning if the function
1593 is used anywhere in the source file. This is useful when identifying
1594 functions that are expected to be removed in a future version of a
1595 program. The warning also includes the location of the declaration
1596 of the deprecated function, to enable users to easily find further
1597 information about why the function is deprecated, or what they should
1598 do instead. Note that the warnings only occurs for uses:
1601 int old_fn () __attribute__ ((deprecated));
1603 int (*fn_ptr)() = old_fn;
1606 results in a warning on line 3 but not line 2.
1608 The @code{deprecated} attribute can also be used for variables and
1609 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
1612 @cindex @code{__declspec(dllexport)}
1613 On Microsoft Windows targets and Symbian OS targets the
1614 @code{dllexport} attribute causes the compiler to provide a global
1615 pointer to a pointer in a DLL, so that it can be referenced with the
1616 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
1617 name is formed by combining @code{_imp__} and the function or variable
1620 You can use @code{__declspec(dllexport)} as a synonym for
1621 @code{__attribute__ ((dllexport))} for compatibility with other
1624 On systems that support the @code{visibility} attribute, this
1625 attribute also implies ``default'' visibility, unless a
1626 @code{visibility} attribute is explicitly specified. You should avoid
1627 the use of @code{dllexport} with ``hidden'' or ``internal''
1628 visibility; in the future GCC may issue an error for those cases.
1630 Currently, the @code{dllexport} attribute is ignored for inlined
1631 functions, unless the @option{-fkeep-inline-functions} flag has been
1632 used. The attribute is also ignored for undefined symbols.
1634 When applied to C++ classes. the attribute marks defined non-inlined
1635 member functions and static data members as exports. Static consts
1636 initialized in-class are not marked unless they are also defined
1639 For Microsoft Windows targets there are alternative methods for
1640 including the symbol in the DLL's export table such as using a
1641 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
1642 the @option{--export-all} linker flag.
1645 @cindex @code{__declspec(dllimport)}
1646 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
1647 attribute causes the compiler to reference a function or variable via
1648 a global pointer to a pointer that is set up by the DLL exporting the
1649 symbol. The attribute implies @code{extern} storage. On Microsoft
1650 Windows targets, the pointer name is formed by combining @code{_imp__}
1651 and the function or variable name.
1653 You can use @code{__declspec(dllimport)} as a synonym for
1654 @code{__attribute__ ((dllimport))} for compatibility with other
1657 Currently, the attribute is ignored for inlined functions. If the
1658 attribute is applied to a symbol @emph{definition}, an error is reported.
1659 If a symbol previously declared @code{dllimport} is later defined, the
1660 attribute is ignored in subsequent references, and a warning is emitted.
1661 The attribute is also overridden by a subsequent declaration as
1664 When applied to C++ classes, the attribute marks non-inlined
1665 member functions and static data members as imports. However, the
1666 attribute is ignored for virtual methods to allow creation of vtables
1669 On the SH Symbian OS target the @code{dllimport} attribute also has
1670 another affect - it can cause the vtable and run-time type information
1671 for a class to be exported. This happens when the class has a
1672 dllimport'ed constructor or a non-inline, non-pure virtual function
1673 and, for either of those two conditions, the class also has a inline
1674 constructor or destructor and has a key function that is defined in
1675 the current translation unit.
1677 For Microsoft Windows based targets the use of the @code{dllimport}
1678 attribute on functions is not necessary, but provides a small
1679 performance benefit by eliminating a thunk in the DLL. The use of the
1680 @code{dllimport} attribute on imported variables was required on older
1681 versions of the GNU linker, but can now be avoided by passing the
1682 @option{--enable-auto-import} switch to the GNU linker. As with
1683 functions, using the attribute for a variable eliminates a thunk in
1686 One drawback to using this attribute is that a pointer to a function
1687 or variable marked as @code{dllimport} cannot be used as a constant
1688 address. On Microsoft Windows targets, the attribute can be disabled
1689 for functions by setting the @option{-mnop-fun-dllimport} flag.
1692 @cindex eight bit data on the H8/300, H8/300H, and H8S
1693 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1694 variable should be placed into the eight bit data section.
1695 The compiler will generate more efficient code for certain operations
1696 on data in the eight bit data area. Note the eight bit data area is limited to
1699 You must use GAS and GLD from GNU binutils version 2.7 or later for
1700 this attribute to work correctly.
1703 @cindex functions which handle memory bank switching
1704 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
1705 use a calling convention that takes care of switching memory banks when
1706 entering and leaving a function. This calling convention is also the
1707 default when using the @option{-mlong-calls} option.
1709 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
1710 to call and return from a function.
1712 On 68HC11 the compiler will generate a sequence of instructions
1713 to invoke a board-specific routine to switch the memory bank and call the
1714 real function. The board-specific routine simulates a @code{call}.
1715 At the end of a function, it will jump to a board-specific routine
1716 instead of using @code{rts}. The board-specific return routine simulates
1720 @cindex functions that pop the argument stack on the 386
1721 On the Intel 386, the @code{fastcall} attribute causes the compiler to
1722 pass the first two arguments in the registers ECX and EDX. Subsequent
1723 arguments are passed on the stack. The called function will pop the
1724 arguments off the stack. If the number of arguments is variable all
1725 arguments are pushed on the stack.
1727 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
1728 @cindex @code{format} function attribute
1730 The @code{format} attribute specifies that a function takes @code{printf},
1731 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
1732 should be type-checked against a format string. For example, the
1737 my_printf (void *my_object, const char *my_format, ...)
1738 __attribute__ ((format (printf, 2, 3)));
1742 causes the compiler to check the arguments in calls to @code{my_printf}
1743 for consistency with the @code{printf} style format string argument
1746 The parameter @var{archetype} determines how the format string is
1747 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
1748 or @code{strfmon}. (You can also use @code{__printf__},
1749 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
1750 parameter @var{string-index} specifies which argument is the format
1751 string argument (starting from 1), while @var{first-to-check} is the
1752 number of the first argument to check against the format string. For
1753 functions where the arguments are not available to be checked (such as
1754 @code{vprintf}), specify the third parameter as zero. In this case the
1755 compiler only checks the format string for consistency. For
1756 @code{strftime} formats, the third parameter is required to be zero.
1757 Since non-static C++ methods have an implicit @code{this} argument, the
1758 arguments of such methods should be counted from two, not one, when
1759 giving values for @var{string-index} and @var{first-to-check}.
1761 In the example above, the format string (@code{my_format}) is the second
1762 argument of the function @code{my_print}, and the arguments to check
1763 start with the third argument, so the correct parameters for the format
1764 attribute are 2 and 3.
1766 @opindex ffreestanding
1767 @opindex fno-builtin
1768 The @code{format} attribute allows you to identify your own functions
1769 which take format strings as arguments, so that GCC can check the
1770 calls to these functions for errors. The compiler always (unless
1771 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
1772 for the standard library functions @code{printf}, @code{fprintf},
1773 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
1774 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
1775 warnings are requested (using @option{-Wformat}), so there is no need to
1776 modify the header file @file{stdio.h}. In C99 mode, the functions
1777 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
1778 @code{vsscanf} are also checked. Except in strictly conforming C
1779 standard modes, the X/Open function @code{strfmon} is also checked as
1780 are @code{printf_unlocked} and @code{fprintf_unlocked}.
1781 @xref{C Dialect Options,,Options Controlling C Dialect}.
1783 The target may provide additional types of format checks.
1784 @xref{Target Format Checks,,Format Checks Specific to Particular
1787 @item format_arg (@var{string-index})
1788 @cindex @code{format_arg} function attribute
1789 @opindex Wformat-nonliteral
1790 The @code{format_arg} attribute specifies that a function takes a format
1791 string for a @code{printf}, @code{scanf}, @code{strftime} or
1792 @code{strfmon} style function and modifies it (for example, to translate
1793 it into another language), so the result can be passed to a
1794 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
1795 function (with the remaining arguments to the format function the same
1796 as they would have been for the unmodified string). For example, the
1801 my_dgettext (char *my_domain, const char *my_format)
1802 __attribute__ ((format_arg (2)));
1806 causes the compiler to check the arguments in calls to a @code{printf},
1807 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
1808 format string argument is a call to the @code{my_dgettext} function, for
1809 consistency with the format string argument @code{my_format}. If the
1810 @code{format_arg} attribute had not been specified, all the compiler
1811 could tell in such calls to format functions would be that the format
1812 string argument is not constant; this would generate a warning when
1813 @option{-Wformat-nonliteral} is used, but the calls could not be checked
1814 without the attribute.
1816 The parameter @var{string-index} specifies which argument is the format
1817 string argument (starting from one). Since non-static C++ methods have
1818 an implicit @code{this} argument, the arguments of such methods should
1819 be counted from two.
1821 The @code{format-arg} attribute allows you to identify your own
1822 functions which modify format strings, so that GCC can check the
1823 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
1824 type function whose operands are a call to one of your own function.
1825 The compiler always treats @code{gettext}, @code{dgettext}, and
1826 @code{dcgettext} in this manner except when strict ISO C support is
1827 requested by @option{-ansi} or an appropriate @option{-std} option, or
1828 @option{-ffreestanding} or @option{-fno-builtin}
1829 is used. @xref{C Dialect Options,,Options
1830 Controlling C Dialect}.
1832 @item function_vector
1833 @cindex calling functions through the function vector on the H8/300 processors
1834 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1835 function should be called through the function vector. Calling a
1836 function through the function vector will reduce code size, however;
1837 the function vector has a limited size (maximum 128 entries on the H8/300
1838 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
1840 You must use GAS and GLD from GNU binutils version 2.7 or later for
1841 this attribute to work correctly.
1844 @cindex interrupt handler functions
1845 Use this attribute on the ARM, AVR, C4x, M32R/D and Xstormy16 ports to indicate
1846 that the specified function is an interrupt handler. The compiler will
1847 generate function entry and exit sequences suitable for use in an
1848 interrupt handler when this attribute is present.
1850 Note, interrupt handlers for the m68k, H8/300, H8/300H, H8S, and SH processors
1851 can be specified via the @code{interrupt_handler} attribute.
1853 Note, on the AVR, interrupts will be enabled inside the function.
1855 Note, for the ARM, you can specify the kind of interrupt to be handled by
1856 adding an optional parameter to the interrupt attribute like this:
1859 void f () __attribute__ ((interrupt ("IRQ")));
1862 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
1864 @item interrupt_handler
1865 @cindex interrupt handler functions on the m68k, H8/300 and SH processors
1866 Use this attribute on the m68k, H8/300, H8/300H, H8S, and SH to indicate that
1867 the specified function is an interrupt handler. The compiler will generate
1868 function entry and exit sequences suitable for use in an interrupt
1869 handler when this attribute is present.
1871 @item long_call/short_call
1872 @cindex indirect calls on ARM
1873 This attribute specifies how a particular function is called on
1874 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
1875 command line switch and @code{#pragma long_calls} settings. The
1876 @code{long_call} attribute causes the compiler to always call the
1877 function by first loading its address into a register and then using the
1878 contents of that register. The @code{short_call} attribute always places
1879 the offset to the function from the call site into the @samp{BL}
1880 instruction directly.
1882 @item longcall/shortcall
1883 @cindex functions called via pointer on the RS/6000 and PowerPC
1884 On the RS/6000 and PowerPC, the @code{longcall} attribute causes the
1885 compiler to always call this function via a pointer, just as it would if
1886 the @option{-mlongcall} option had been specified. The @code{shortcall}
1887 attribute causes the compiler not to do this. These attributes override
1888 both the @option{-mlongcall} switch and the @code{#pragma longcall}
1891 @xref{RS/6000 and PowerPC Options}, for more information on whether long
1892 calls are necessary.
1895 @cindex @code{malloc} attribute
1896 The @code{malloc} attribute is used to tell the compiler that a function
1897 may be treated as if any non-@code{NULL} pointer it returns cannot
1898 alias any other pointer valid when the function returns.
1899 This will often improve optimization.
1900 Standard functions with this property include @code{malloc} and
1901 @code{calloc}. @code{realloc}-like functions have this property as
1902 long as the old pointer is never referred to (including comparing it
1903 to the new pointer) after the function returns a non-@code{NULL}
1906 @item model (@var{model-name})
1907 @cindex function addressability on the M32R/D
1908 @cindex variable addressability on the IA-64
1910 On the M32R/D, use this attribute to set the addressability of an
1911 object, and of the code generated for a function. The identifier
1912 @var{model-name} is one of @code{small}, @code{medium}, or
1913 @code{large}, representing each of the code models.
1915 Small model objects live in the lower 16MB of memory (so that their
1916 addresses can be loaded with the @code{ld24} instruction), and are
1917 callable with the @code{bl} instruction.
1919 Medium model objects may live anywhere in the 32-bit address space (the
1920 compiler will generate @code{seth/add3} instructions to load their addresses),
1921 and are callable with the @code{bl} instruction.
1923 Large model objects may live anywhere in the 32-bit address space (the
1924 compiler will generate @code{seth/add3} instructions to load their addresses),
1925 and may not be reachable with the @code{bl} instruction (the compiler will
1926 generate the much slower @code{seth/add3/jl} instruction sequence).
1928 On IA-64, use this attribute to set the addressability of an object.
1929 At present, the only supported identifier for @var{model-name} is
1930 @code{small}, indicating addressability via ``small'' (22-bit)
1931 addresses (so that their addresses can be loaded with the @code{addl}
1932 instruction). Caveat: such addressing is by definition not position
1933 independent and hence this attribute must not be used for objects
1934 defined by shared libraries.
1937 @cindex function without a prologue/epilogue code
1938 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
1939 specified function does not need prologue/epilogue sequences generated by
1940 the compiler. It is up to the programmer to provide these sequences.
1943 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
1944 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
1945 use the normal calling convention based on @code{jsr} and @code{rts}.
1946 This attribute can be used to cancel the effect of the @option{-mlong-calls}
1949 @item no_instrument_function
1950 @cindex @code{no_instrument_function} function attribute
1951 @opindex finstrument-functions
1952 If @option{-finstrument-functions} is given, profiling function calls will
1953 be generated at entry and exit of most user-compiled functions.
1954 Functions with this attribute will not be so instrumented.
1957 @cindex @code{noinline} function attribute
1958 This function attribute prevents a function from being considered for
1961 @item nonnull (@var{arg-index}, @dots{})
1962 @cindex @code{nonnull} function attribute
1963 The @code{nonnull} attribute specifies that some function parameters should
1964 be non-null pointers. For instance, the declaration:
1968 my_memcpy (void *dest, const void *src, size_t len)
1969 __attribute__((nonnull (1, 2)));
1973 causes the compiler to check that, in calls to @code{my_memcpy},
1974 arguments @var{dest} and @var{src} are non-null. If the compiler
1975 determines that a null pointer is passed in an argument slot marked
1976 as non-null, and the @option{-Wnonnull} option is enabled, a warning
1977 is issued. The compiler may also choose to make optimizations based
1978 on the knowledge that certain function arguments will not be null.
1980 If no argument index list is given to the @code{nonnull} attribute,
1981 all pointer arguments are marked as non-null. To illustrate, the
1982 following declaration is equivalent to the previous example:
1986 my_memcpy (void *dest, const void *src, size_t len)
1987 __attribute__((nonnull));
1991 @cindex @code{noreturn} function attribute
1992 A few standard library functions, such as @code{abort} and @code{exit},
1993 cannot return. GCC knows this automatically. Some programs define
1994 their own functions that never return. You can declare them
1995 @code{noreturn} to tell the compiler this fact. For example,
1999 void fatal () __attribute__ ((noreturn));
2002 fatal (/* @r{@dots{}} */)
2004 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2010 The @code{noreturn} keyword tells the compiler to assume that
2011 @code{fatal} cannot return. It can then optimize without regard to what
2012 would happen if @code{fatal} ever did return. This makes slightly
2013 better code. More importantly, it helps avoid spurious warnings of
2014 uninitialized variables.
2016 The @code{noreturn} keyword does not affect the exceptional path when that
2017 applies: a @code{noreturn}-marked function may still return to the caller
2018 by throwing an exception or calling @code{longjmp}.
2020 Do not assume that registers saved by the calling function are
2021 restored before calling the @code{noreturn} function.
2023 It does not make sense for a @code{noreturn} function to have a return
2024 type other than @code{void}.
2026 The attribute @code{noreturn} is not implemented in GCC versions
2027 earlier than 2.5. An alternative way to declare that a function does
2028 not return, which works in the current version and in some older
2029 versions, is as follows:
2032 typedef void voidfn ();
2034 volatile voidfn fatal;
2038 @cindex @code{nothrow} function attribute
2039 The @code{nothrow} attribute is used to inform the compiler that a
2040 function cannot throw an exception. For example, most functions in
2041 the standard C library can be guaranteed not to throw an exception
2042 with the notable exceptions of @code{qsort} and @code{bsearch} that
2043 take function pointer arguments. The @code{nothrow} attribute is not
2044 implemented in GCC versions earlier than 3.3.
2047 @cindex @code{pure} function attribute
2048 Many functions have no effects except the return value and their
2049 return value depends only on the parameters and/or global variables.
2050 Such a function can be subject
2051 to common subexpression elimination and loop optimization just as an
2052 arithmetic operator would be. These functions should be declared
2053 with the attribute @code{pure}. For example,
2056 int square (int) __attribute__ ((pure));
2060 says that the hypothetical function @code{square} is safe to call
2061 fewer times than the program says.
2063 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2064 Interesting non-pure functions are functions with infinite loops or those
2065 depending on volatile memory or other system resource, that may change between
2066 two consecutive calls (such as @code{feof} in a multithreading environment).
2068 The attribute @code{pure} is not implemented in GCC versions earlier
2071 @item regparm (@var{number})
2072 @cindex @code{regparm} attribute
2073 @cindex functions that are passed arguments in registers on the 386
2074 On the Intel 386, the @code{regparm} attribute causes the compiler to
2075 pass up to @var{number} integer arguments in registers EAX,
2076 EDX, and ECX instead of on the stack. Functions that take a
2077 variable number of arguments will continue to be passed all of their
2078 arguments on the stack.
2080 Beware that on some ELF systems this attribute is unsuitable for
2081 global functions in shared libraries with lazy binding (which is the
2082 default). Lazy binding will send the first call via resolving code in
2083 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2084 per the standard calling conventions. Solaris 8 is affected by this.
2085 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2086 safe since the loaders there save all registers. (Lazy binding can be
2087 disabled with the linker or the loader if desired, to avoid the
2091 @cindex save all registers on the H8/300, H8/300H, and H8S
2092 Use this attribute on the H8/300, H8/300H, and H8S to indicate that
2093 all registers except the stack pointer should be saved in the prologue
2094 regardless of whether they are used or not.
2096 @item section ("@var{section-name}")
2097 @cindex @code{section} function attribute
2098 Normally, the compiler places the code it generates in the @code{text} section.
2099 Sometimes, however, you need additional sections, or you need certain
2100 particular functions to appear in special sections. The @code{section}
2101 attribute specifies that a function lives in a particular section.
2102 For example, the declaration:
2105 extern void foobar (void) __attribute__ ((section ("bar")));
2109 puts the function @code{foobar} in the @code{bar} section.
2111 Some file formats do not support arbitrary sections so the @code{section}
2112 attribute is not available on all platforms.
2113 If you need to map the entire contents of a module to a particular
2114 section, consider using the facilities of the linker instead.
2117 @cindex @code{sentinel} function attribute
2118 This function attribute ensures that a parameter in a function call is
2119 an explicit @code{NULL}. The attribute is only valid on variadic
2120 functions. By default, the sentinel is located at position zero, the
2121 last parameter of the function call. If an optional integer position
2122 argument P is supplied to the attribute, the sentinel must be located at
2123 position P counting backwards from the end of the argument list.
2126 __attribute__ ((sentinel))
2128 __attribute__ ((sentinel(0)))
2131 The attribute is automatically set with a position of 0 for the built-in
2132 functions @code{execl} and @code{execlp}. The built-in function
2133 @code{execle} has the attribute set with a position of 1.
2135 A valid @code{NULL} in this context is defined as zero with any pointer
2136 type. If your system defines the @code{NULL} macro with an integer type
2137 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2138 with a copy that redefines NULL appropriately.
2140 The warnings for missing or incorrect sentinels are enabled with
2144 See long_call/short_call.
2147 See longcall/shortcall.
2150 @cindex signal handler functions on the AVR processors
2151 Use this attribute on the AVR to indicate that the specified
2152 function is a signal handler. The compiler will generate function
2153 entry and exit sequences suitable for use in a signal handler when this
2154 attribute is present. Interrupts will be disabled inside the function.
2157 Use this attribute on the SH to indicate an @code{interrupt_handler}
2158 function should switch to an alternate stack. It expects a string
2159 argument that names a global variable holding the address of the
2164 void f () __attribute__ ((interrupt_handler,
2165 sp_switch ("alt_stack")));
2169 @cindex functions that pop the argument stack on the 386
2170 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2171 assume that the called function will pop off the stack space used to
2172 pass arguments, unless it takes a variable number of arguments.
2175 @cindex tiny data section on the H8/300H and H8S
2176 Use this attribute on the H8/300H and H8S to indicate that the specified
2177 variable should be placed into the tiny data section.
2178 The compiler will generate more efficient code for loads and stores
2179 on data in the tiny data section. Note the tiny data area is limited to
2180 slightly under 32kbytes of data.
2183 Use this attribute on the SH for an @code{interrupt_handler} to return using
2184 @code{trapa} instead of @code{rte}. This attribute expects an integer
2185 argument specifying the trap number to be used.
2188 @cindex @code{unused} attribute.
2189 This attribute, attached to a function, means that the function is meant
2190 to be possibly unused. GCC will not produce a warning for this
2194 @cindex @code{used} attribute.
2195 This attribute, attached to a function, means that code must be emitted
2196 for the function even if it appears that the function is not referenced.
2197 This is useful, for example, when the function is referenced only in
2200 @item visibility ("@var{visibility_type}")
2201 @cindex @code{visibility} attribute
2202 The @code{visibility} attribute on ELF targets causes the declaration
2203 to be emitted with default, hidden, protected or internal visibility.
2206 void __attribute__ ((visibility ("protected")))
2207 f () @{ /* @r{Do something.} */; @}
2208 int i __attribute__ ((visibility ("hidden")));
2211 See the ELF gABI for complete details, but the short story is:
2214 @c keep this list of visibilities in alphabetical order.
2217 Default visibility is the normal case for ELF. This value is
2218 available for the visibility attribute to override other options
2219 that may change the assumed visibility of symbols.
2222 Hidden visibility indicates that the symbol will not be placed into
2223 the dynamic symbol table, so no other @dfn{module} (executable or
2224 shared library) can reference it directly.
2227 Internal visibility is like hidden visibility, but with additional
2228 processor specific semantics. Unless otherwise specified by the psABI,
2229 GCC defines internal visibility to mean that the function is @emph{never}
2230 called from another module. Note that hidden symbols, while they cannot
2231 be referenced directly by other modules, can be referenced indirectly via
2232 function pointers. By indicating that a symbol cannot be called from
2233 outside the module, GCC may for instance omit the load of a PIC register
2234 since it is known that the calling function loaded the correct value.
2237 Protected visibility indicates that the symbol will be placed in the
2238 dynamic symbol table, but that references within the defining module
2239 will bind to the local symbol. That is, the symbol cannot be overridden
2244 Not all ELF targets support this attribute.
2246 @item warn_unused_result
2247 @cindex @code{warn_unused_result} attribute
2248 The @code{warn_unused_result} attribute causes a warning to be emitted
2249 if a caller of the function with this attribute does not use its
2250 return value. This is useful for functions where not checking
2251 the result is either a security problem or always a bug, such as
2255 int fn () __attribute__ ((warn_unused_result));
2258 if (fn () < 0) return -1;
2264 results in warning on line 5.
2267 @cindex @code{weak} attribute
2268 The @code{weak} attribute causes the declaration to be emitted as a weak
2269 symbol rather than a global. This is primarily useful in defining
2270 library functions which can be overridden in user code, though it can
2271 also be used with non-function declarations. Weak symbols are supported
2272 for ELF targets, and also for a.out targets when using the GNU assembler
2277 You can specify multiple attributes in a declaration by separating them
2278 by commas within the double parentheses or by immediately following an
2279 attribute declaration with another attribute declaration.
2281 @cindex @code{#pragma}, reason for not using
2282 @cindex pragma, reason for not using
2283 Some people object to the @code{__attribute__} feature, suggesting that
2284 ISO C's @code{#pragma} should be used instead. At the time
2285 @code{__attribute__} was designed, there were two reasons for not doing
2290 It is impossible to generate @code{#pragma} commands from a macro.
2293 There is no telling what the same @code{#pragma} might mean in another
2297 These two reasons applied to almost any application that might have been
2298 proposed for @code{#pragma}. It was basically a mistake to use
2299 @code{#pragma} for @emph{anything}.
2301 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2302 to be generated from macros. In addition, a @code{#pragma GCC}
2303 namespace is now in use for GCC-specific pragmas. However, it has been
2304 found convenient to use @code{__attribute__} to achieve a natural
2305 attachment of attributes to their corresponding declarations, whereas
2306 @code{#pragma GCC} is of use for constructs that do not naturally form
2307 part of the grammar. @xref{Other Directives,,Miscellaneous
2308 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2310 @node Attribute Syntax
2311 @section Attribute Syntax
2312 @cindex attribute syntax
2314 This section describes the syntax with which @code{__attribute__} may be
2315 used, and the constructs to which attribute specifiers bind, for the C
2316 language. Some details may vary for C++ and Objective-C@. Because of
2317 infelicities in the grammar for attributes, some forms described here
2318 may not be successfully parsed in all cases.
2320 There are some problems with the semantics of attributes in C++. For
2321 example, there are no manglings for attributes, although they may affect
2322 code generation, so problems may arise when attributed types are used in
2323 conjunction with templates or overloading. Similarly, @code{typeid}
2324 does not distinguish between types with different attributes. Support
2325 for attributes in C++ may be restricted in future to attributes on
2326 declarations only, but not on nested declarators.
2328 @xref{Function Attributes}, for details of the semantics of attributes
2329 applying to functions. @xref{Variable Attributes}, for details of the
2330 semantics of attributes applying to variables. @xref{Type Attributes},
2331 for details of the semantics of attributes applying to structure, union
2332 and enumerated types.
2334 An @dfn{attribute specifier} is of the form
2335 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2336 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2337 each attribute is one of the following:
2341 Empty. Empty attributes are ignored.
2344 A word (which may be an identifier such as @code{unused}, or a reserved
2345 word such as @code{const}).
2348 A word, followed by, in parentheses, parameters for the attribute.
2349 These parameters take one of the following forms:
2353 An identifier. For example, @code{mode} attributes use this form.
2356 An identifier followed by a comma and a non-empty comma-separated list
2357 of expressions. For example, @code{format} attributes use this form.
2360 A possibly empty comma-separated list of expressions. For example,
2361 @code{format_arg} attributes use this form with the list being a single
2362 integer constant expression, and @code{alias} attributes use this form
2363 with the list being a single string constant.
2367 An @dfn{attribute specifier list} is a sequence of one or more attribute
2368 specifiers, not separated by any other tokens.
2370 In GNU C, an attribute specifier list may appear after the colon following a
2371 label, other than a @code{case} or @code{default} label. The only
2372 attribute it makes sense to use after a label is @code{unused}. This
2373 feature is intended for code generated by programs which contains labels
2374 that may be unused but which is compiled with @option{-Wall}. It would
2375 not normally be appropriate to use in it human-written code, though it
2376 could be useful in cases where the code that jumps to the label is
2377 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2378 such placement of attribute lists, as it is permissible for a
2379 declaration, which could begin with an attribute list, to be labelled in
2380 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2381 does not arise there.
2383 An attribute specifier list may appear as part of a @code{struct},
2384 @code{union} or @code{enum} specifier. It may go either immediately
2385 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2386 the closing brace. It is ignored if the content of the structure, union
2387 or enumerated type is not defined in the specifier in which the
2388 attribute specifier list is used---that is, in usages such as
2389 @code{struct __attribute__((foo)) bar} with no following opening brace.
2390 Where attribute specifiers follow the closing brace, they are considered
2391 to relate to the structure, union or enumerated type defined, not to any
2392 enclosing declaration the type specifier appears in, and the type
2393 defined is not complete until after the attribute specifiers.
2394 @c Otherwise, there would be the following problems: a shift/reduce
2395 @c conflict between attributes binding the struct/union/enum and
2396 @c binding to the list of specifiers/qualifiers; and "aligned"
2397 @c attributes could use sizeof for the structure, but the size could be
2398 @c changed later by "packed" attributes.
2400 Otherwise, an attribute specifier appears as part of a declaration,
2401 counting declarations of unnamed parameters and type names, and relates
2402 to that declaration (which may be nested in another declaration, for
2403 example in the case of a parameter declaration), or to a particular declarator
2404 within a declaration. Where an
2405 attribute specifier is applied to a parameter declared as a function or
2406 an array, it should apply to the function or array rather than the
2407 pointer to which the parameter is implicitly converted, but this is not
2408 yet correctly implemented.
2410 Any list of specifiers and qualifiers at the start of a declaration may
2411 contain attribute specifiers, whether or not such a list may in that
2412 context contain storage class specifiers. (Some attributes, however,
2413 are essentially in the nature of storage class specifiers, and only make
2414 sense where storage class specifiers may be used; for example,
2415 @code{section}.) There is one necessary limitation to this syntax: the
2416 first old-style parameter declaration in a function definition cannot
2417 begin with an attribute specifier, because such an attribute applies to
2418 the function instead by syntax described below (which, however, is not
2419 yet implemented in this case). In some other cases, attribute
2420 specifiers are permitted by this grammar but not yet supported by the
2421 compiler. All attribute specifiers in this place relate to the
2422 declaration as a whole. In the obsolescent usage where a type of
2423 @code{int} is implied by the absence of type specifiers, such a list of
2424 specifiers and qualifiers may be an attribute specifier list with no
2425 other specifiers or qualifiers.
2427 An attribute specifier list may appear immediately before a declarator
2428 (other than the first) in a comma-separated list of declarators in a
2429 declaration of more than one identifier using a single list of
2430 specifiers and qualifiers. Such attribute specifiers apply
2431 only to the identifier before whose declarator they appear. For
2435 __attribute__((noreturn)) void d0 (void),
2436 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2441 the @code{noreturn} attribute applies to all the functions
2442 declared; the @code{format} attribute only applies to @code{d1}.
2444 An attribute specifier list may appear immediately before the comma,
2445 @code{=} or semicolon terminating the declaration of an identifier other
2446 than a function definition. At present, such attribute specifiers apply
2447 to the declared object or function, but in future they may attach to the
2448 outermost adjacent declarator. In simple cases there is no difference,
2449 but, for example, in
2452 void (****f)(void) __attribute__((noreturn));
2456 at present the @code{noreturn} attribute applies to @code{f}, which
2457 causes a warning since @code{f} is not a function, but in future it may
2458 apply to the function @code{****f}. The precise semantics of what
2459 attributes in such cases will apply to are not yet specified. Where an
2460 assembler name for an object or function is specified (@pxref{Asm
2461 Labels}), at present the attribute must follow the @code{asm}
2462 specification; in future, attributes before the @code{asm} specification
2463 may apply to the adjacent declarator, and those after it to the declared
2466 An attribute specifier list may, in future, be permitted to appear after
2467 the declarator in a function definition (before any old-style parameter
2468 declarations or the function body).
2470 Attribute specifiers may be mixed with type qualifiers appearing inside
2471 the @code{[]} of a parameter array declarator, in the C99 construct by
2472 which such qualifiers are applied to the pointer to which the array is
2473 implicitly converted. Such attribute specifiers apply to the pointer,
2474 not to the array, but at present this is not implemented and they are
2477 An attribute specifier list may appear at the start of a nested
2478 declarator. At present, there are some limitations in this usage: the
2479 attributes correctly apply to the declarator, but for most individual
2480 attributes the semantics this implies are not implemented.
2481 When attribute specifiers follow the @code{*} of a pointer
2482 declarator, they may be mixed with any type qualifiers present.
2483 The following describes the formal semantics of this syntax. It will make the
2484 most sense if you are familiar with the formal specification of
2485 declarators in the ISO C standard.
2487 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2488 D1}, where @code{T} contains declaration specifiers that specify a type
2489 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2490 contains an identifier @var{ident}. The type specified for @var{ident}
2491 for derived declarators whose type does not include an attribute
2492 specifier is as in the ISO C standard.
2494 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2495 and the declaration @code{T D} specifies the type
2496 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2497 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2498 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2500 If @code{D1} has the form @code{*
2501 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2502 declaration @code{T D} specifies the type
2503 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2504 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2505 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2511 void (__attribute__((noreturn)) ****f) (void);
2515 specifies the type ``pointer to pointer to pointer to pointer to
2516 non-returning function returning @code{void}''. As another example,
2519 char *__attribute__((aligned(8))) *f;
2523 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2524 Note again that this does not work with most attributes; for example,
2525 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2526 is not yet supported.
2528 For compatibility with existing code written for compiler versions that
2529 did not implement attributes on nested declarators, some laxity is
2530 allowed in the placing of attributes. If an attribute that only applies
2531 to types is applied to a declaration, it will be treated as applying to
2532 the type of that declaration. If an attribute that only applies to
2533 declarations is applied to the type of a declaration, it will be treated
2534 as applying to that declaration; and, for compatibility with code
2535 placing the attributes immediately before the identifier declared, such
2536 an attribute applied to a function return type will be treated as
2537 applying to the function type, and such an attribute applied to an array
2538 element type will be treated as applying to the array type. If an
2539 attribute that only applies to function types is applied to a
2540 pointer-to-function type, it will be treated as applying to the pointer
2541 target type; if such an attribute is applied to a function return type
2542 that is not a pointer-to-function type, it will be treated as applying
2543 to the function type.
2545 @node Function Prototypes
2546 @section Prototypes and Old-Style Function Definitions
2547 @cindex function prototype declarations
2548 @cindex old-style function definitions
2549 @cindex promotion of formal parameters
2551 GNU C extends ISO C to allow a function prototype to override a later
2552 old-style non-prototype definition. Consider the following example:
2555 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2562 /* @r{Prototype function declaration.} */
2563 int isroot P((uid_t));
2565 /* @r{Old-style function definition.} */
2567 isroot (x) /* ??? lossage here ??? */
2574 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2575 not allow this example, because subword arguments in old-style
2576 non-prototype definitions are promoted. Therefore in this example the
2577 function definition's argument is really an @code{int}, which does not
2578 match the prototype argument type of @code{short}.
2580 This restriction of ISO C makes it hard to write code that is portable
2581 to traditional C compilers, because the programmer does not know
2582 whether the @code{uid_t} type is @code{short}, @code{int}, or
2583 @code{long}. Therefore, in cases like these GNU C allows a prototype
2584 to override a later old-style definition. More precisely, in GNU C, a
2585 function prototype argument type overrides the argument type specified
2586 by a later old-style definition if the former type is the same as the
2587 latter type before promotion. Thus in GNU C the above example is
2588 equivalent to the following:
2601 GNU C++ does not support old-style function definitions, so this
2602 extension is irrelevant.
2605 @section C++ Style Comments
2607 @cindex C++ comments
2608 @cindex comments, C++ style
2610 In GNU C, you may use C++ style comments, which start with @samp{//} and
2611 continue until the end of the line. Many other C implementations allow
2612 such comments, and they are included in the 1999 C standard. However,
2613 C++ style comments are not recognized if you specify an @option{-std}
2614 option specifying a version of ISO C before C99, or @option{-ansi}
2615 (equivalent to @option{-std=c89}).
2618 @section Dollar Signs in Identifier Names
2620 @cindex dollar signs in identifier names
2621 @cindex identifier names, dollar signs in
2623 In GNU C, you may normally use dollar signs in identifier names.
2624 This is because many traditional C implementations allow such identifiers.
2625 However, dollar signs in identifiers are not supported on a few target
2626 machines, typically because the target assembler does not allow them.
2628 @node Character Escapes
2629 @section The Character @key{ESC} in Constants
2631 You can use the sequence @samp{\e} in a string or character constant to
2632 stand for the ASCII character @key{ESC}.
2635 @section Inquiring on Alignment of Types or Variables
2637 @cindex type alignment
2638 @cindex variable alignment
2640 The keyword @code{__alignof__} allows you to inquire about how an object
2641 is aligned, or the minimum alignment usually required by a type. Its
2642 syntax is just like @code{sizeof}.
2644 For example, if the target machine requires a @code{double} value to be
2645 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2646 This is true on many RISC machines. On more traditional machine
2647 designs, @code{__alignof__ (double)} is 4 or even 2.
2649 Some machines never actually require alignment; they allow reference to any
2650 data type even at an odd address. For these machines, @code{__alignof__}
2651 reports the @emph{recommended} alignment of a type.
2653 If the operand of @code{__alignof__} is an lvalue rather than a type,
2654 its value is the required alignment for its type, taking into account
2655 any minimum alignment specified with GCC's @code{__attribute__}
2656 extension (@pxref{Variable Attributes}). For example, after this
2660 struct foo @{ int x; char y; @} foo1;
2664 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2665 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2667 It is an error to ask for the alignment of an incomplete type.
2669 @node Variable Attributes
2670 @section Specifying Attributes of Variables
2671 @cindex attribute of variables
2672 @cindex variable attributes
2674 The keyword @code{__attribute__} allows you to specify special
2675 attributes of variables or structure fields. This keyword is followed
2676 by an attribute specification inside double parentheses. Some
2677 attributes are currently defined generically for variables.
2678 Other attributes are defined for variables on particular target
2679 systems. Other attributes are available for functions
2680 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
2681 Other front ends might define more attributes
2682 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2684 You may also specify attributes with @samp{__} preceding and following
2685 each keyword. This allows you to use them in header files without
2686 being concerned about a possible macro of the same name. For example,
2687 you may use @code{__aligned__} instead of @code{aligned}.
2689 @xref{Attribute Syntax}, for details of the exact syntax for using
2693 @cindex @code{aligned} attribute
2694 @item aligned (@var{alignment})
2695 This attribute specifies a minimum alignment for the variable or
2696 structure field, measured in bytes. For example, the declaration:
2699 int x __attribute__ ((aligned (16))) = 0;
2703 causes the compiler to allocate the global variable @code{x} on a
2704 16-byte boundary. On a 68040, this could be used in conjunction with
2705 an @code{asm} expression to access the @code{move16} instruction which
2706 requires 16-byte aligned operands.
2708 You can also specify the alignment of structure fields. For example, to
2709 create a double-word aligned @code{int} pair, you could write:
2712 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2716 This is an alternative to creating a union with a @code{double} member
2717 that forces the union to be double-word aligned.
2719 As in the preceding examples, you can explicitly specify the alignment
2720 (in bytes) that you wish the compiler to use for a given variable or
2721 structure field. Alternatively, you can leave out the alignment factor
2722 and just ask the compiler to align a variable or field to the maximum
2723 useful alignment for the target machine you are compiling for. For
2724 example, you could write:
2727 short array[3] __attribute__ ((aligned));
2730 Whenever you leave out the alignment factor in an @code{aligned} attribute
2731 specification, the compiler automatically sets the alignment for the declared
2732 variable or field to the largest alignment which is ever used for any data
2733 type on the target machine you are compiling for. Doing this can often make
2734 copy operations more efficient, because the compiler can use whatever
2735 instructions copy the biggest chunks of memory when performing copies to
2736 or from the variables or fields that you have aligned this way.
2738 The @code{aligned} attribute can only increase the alignment; but you
2739 can decrease it by specifying @code{packed} as well. See below.
2741 Note that the effectiveness of @code{aligned} attributes may be limited
2742 by inherent limitations in your linker. On many systems, the linker is
2743 only able to arrange for variables to be aligned up to a certain maximum
2744 alignment. (For some linkers, the maximum supported alignment may
2745 be very very small.) If your linker is only able to align variables
2746 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
2747 in an @code{__attribute__} will still only provide you with 8 byte
2748 alignment. See your linker documentation for further information.
2750 @item cleanup (@var{cleanup_function})
2751 @cindex @code{cleanup} attribute
2752 The @code{cleanup} attribute runs a function when the variable goes
2753 out of scope. This attribute can only be applied to auto function
2754 scope variables; it may not be applied to parameters or variables
2755 with static storage duration. The function must take one parameter,
2756 a pointer to a type compatible with the variable. The return value
2757 of the function (if any) is ignored.
2759 If @option{-fexceptions} is enabled, then @var{cleanup_function}
2760 will be run during the stack unwinding that happens during the
2761 processing of the exception. Note that the @code{cleanup} attribute
2762 does not allow the exception to be caught, only to perform an action.
2763 It is undefined what happens if @var{cleanup_function} does not
2768 @cindex @code{common} attribute
2769 @cindex @code{nocommon} attribute
2772 The @code{common} attribute requests GCC to place a variable in
2773 ``common'' storage. The @code{nocommon} attribute requests the
2774 opposite -- to allocate space for it directly.
2776 These attributes override the default chosen by the
2777 @option{-fno-common} and @option{-fcommon} flags respectively.
2780 @cindex @code{deprecated} attribute
2781 The @code{deprecated} attribute results in a warning if the variable
2782 is used anywhere in the source file. This is useful when identifying
2783 variables that are expected to be removed in a future version of a
2784 program. The warning also includes the location of the declaration
2785 of the deprecated variable, to enable users to easily find further
2786 information about why the variable is deprecated, or what they should
2787 do instead. Note that the warning only occurs for uses:
2790 extern int old_var __attribute__ ((deprecated));
2792 int new_fn () @{ return old_var; @}
2795 results in a warning on line 3 but not line 2.
2797 The @code{deprecated} attribute can also be used for functions and
2798 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
2800 @item mode (@var{mode})
2801 @cindex @code{mode} attribute
2802 This attribute specifies the data type for the declaration---whichever
2803 type corresponds to the mode @var{mode}. This in effect lets you
2804 request an integer or floating point type according to its width.
2806 You may also specify a mode of @samp{byte} or @samp{__byte__} to
2807 indicate the mode corresponding to a one-byte integer, @samp{word} or
2808 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
2809 or @samp{__pointer__} for the mode used to represent pointers.
2812 @cindex @code{packed} attribute
2813 The @code{packed} attribute specifies that a variable or structure field
2814 should have the smallest possible alignment---one byte for a variable,
2815 and one bit for a field, unless you specify a larger value with the
2816 @code{aligned} attribute.
2818 Here is a structure in which the field @code{x} is packed, so that it
2819 immediately follows @code{a}:
2825 int x[2] __attribute__ ((packed));
2829 @item section ("@var{section-name}")
2830 @cindex @code{section} variable attribute
2831 Normally, the compiler places the objects it generates in sections like
2832 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
2833 or you need certain particular variables to appear in special sections,
2834 for example to map to special hardware. The @code{section}
2835 attribute specifies that a variable (or function) lives in a particular
2836 section. For example, this small program uses several specific section names:
2839 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
2840 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
2841 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
2842 int init_data __attribute__ ((section ("INITDATA"))) = 0;
2846 /* Initialize stack pointer */
2847 init_sp (stack + sizeof (stack));
2849 /* Initialize initialized data */
2850 memcpy (&init_data, &data, &edata - &data);
2852 /* Turn on the serial ports */
2859 Use the @code{section} attribute with an @emph{initialized} definition
2860 of a @emph{global} variable, as shown in the example. GCC issues
2861 a warning and otherwise ignores the @code{section} attribute in
2862 uninitialized variable declarations.
2864 You may only use the @code{section} attribute with a fully initialized
2865 global definition because of the way linkers work. The linker requires
2866 each object be defined once, with the exception that uninitialized
2867 variables tentatively go in the @code{common} (or @code{bss}) section
2868 and can be multiply ``defined''. You can force a variable to be
2869 initialized with the @option{-fno-common} flag or the @code{nocommon}
2872 Some file formats do not support arbitrary sections so the @code{section}
2873 attribute is not available on all platforms.
2874 If you need to map the entire contents of a module to a particular
2875 section, consider using the facilities of the linker instead.
2878 @cindex @code{shared} variable attribute
2879 On Microsoft Windows, in addition to putting variable definitions in a named
2880 section, the section can also be shared among all running copies of an
2881 executable or DLL@. For example, this small program defines shared data
2882 by putting it in a named section @code{shared} and marking the section
2886 int foo __attribute__((section ("shared"), shared)) = 0;
2891 /* Read and write foo. All running
2892 copies see the same value. */
2898 You may only use the @code{shared} attribute along with @code{section}
2899 attribute with a fully initialized global definition because of the way
2900 linkers work. See @code{section} attribute for more information.
2902 The @code{shared} attribute is only available on Microsoft Windows@.
2904 @item tls_model ("@var{tls_model}")
2905 @cindex @code{tls_model} attribute
2906 The @code{tls_model} attribute sets thread-local storage model
2907 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
2908 overriding @option{-ftls-model=} command line switch on a per-variable
2910 The @var{tls_model} argument should be one of @code{global-dynamic},
2911 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
2913 Not all targets support this attribute.
2915 @item transparent_union
2916 This attribute, attached to a function parameter which is a union, means
2917 that the corresponding argument may have the type of any union member,
2918 but the argument is passed as if its type were that of the first union
2919 member. For more details see @xref{Type Attributes}. You can also use
2920 this attribute on a @code{typedef} for a union data type; then it
2921 applies to all function parameters with that type.
2924 This attribute, attached to a variable, means that the variable is meant
2925 to be possibly unused. GCC will not produce a warning for this
2928 @item vector_size (@var{bytes})
2929 This attribute specifies the vector size for the variable, measured in
2930 bytes. For example, the declaration:
2933 int foo __attribute__ ((vector_size (16)));
2937 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
2938 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
2939 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
2941 This attribute is only applicable to integral and float scalars,
2942 although arrays, pointers, and function return values are allowed in
2943 conjunction with this construct.
2945 Aggregates with this attribute are invalid, even if they are of the same
2946 size as a corresponding scalar. For example, the declaration:
2949 struct S @{ int a; @};
2950 struct S __attribute__ ((vector_size (16))) foo;
2954 is invalid even if the size of the structure is the same as the size of
2958 The @code{weak} attribute is described in @xref{Function Attributes}.
2961 The @code{dllimport} attribute is described in @xref{Function Attributes}.
2964 The @code{dllexport} attribute is described in @xref{Function Attributes}.
2968 @subsection M32R/D Variable Attributes
2970 One attribute is currently defined for the M32R/D.
2973 @item model (@var{model-name})
2974 @cindex variable addressability on the M32R/D
2975 Use this attribute on the M32R/D to set the addressability of an object.
2976 The identifier @var{model-name} is one of @code{small}, @code{medium},
2977 or @code{large}, representing each of the code models.
2979 Small model objects live in the lower 16MB of memory (so that their
2980 addresses can be loaded with the @code{ld24} instruction).
2982 Medium and large model objects may live anywhere in the 32-bit address space
2983 (the compiler will generate @code{seth/add3} instructions to load their
2987 @subsection i386 Variable Attributes
2989 Two attributes are currently defined for i386 configurations:
2990 @code{ms_struct} and @code{gcc_struct}
2995 @cindex @code{ms_struct} attribute
2996 @cindex @code{gcc_struct} attribute
2998 If @code{packed} is used on a structure, or if bit-fields are used
2999 it may be that the Microsoft ABI packs them differently
3000 than GCC would normally pack them. Particularly when moving packed
3001 data between functions compiled with GCC and the native Microsoft compiler
3002 (either via function call or as data in a file), it may be necessary to access
3005 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3006 compilers to match the native Microsoft compiler.
3009 @subsection Xstormy16 Variable Attributes
3011 One attribute is currently defined for xstormy16 configurations:
3016 @cindex @code{below100} attribute
3018 If a variable has the @code{below100} attribute (@code{BELOW100} is
3019 allowed also), GCC will place the variable in the first 0x100 bytes of
3020 memory and use special opcodes to access it. Such variables will be
3021 placed in either the @code{.bss_below100} section or the
3022 @code{.data_below100} section.
3026 @node Type Attributes
3027 @section Specifying Attributes of Types
3028 @cindex attribute of types
3029 @cindex type attributes
3031 The keyword @code{__attribute__} allows you to specify special
3032 attributes of @code{struct} and @code{union} types when you define such
3033 types. This keyword is followed by an attribute specification inside
3034 double parentheses. Six attributes are currently defined for types:
3035 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3036 @code{deprecated} and @code{may_alias}. Other attributes are defined for
3037 functions (@pxref{Function Attributes}) and for variables
3038 (@pxref{Variable Attributes}).
3040 You may also specify any one of these attributes with @samp{__}
3041 preceding and following its keyword. This allows you to use these
3042 attributes in header files without being concerned about a possible
3043 macro of the same name. For example, you may use @code{__aligned__}
3044 instead of @code{aligned}.
3046 You may specify the @code{aligned} and @code{transparent_union}
3047 attributes either in a @code{typedef} declaration or just past the
3048 closing curly brace of a complete enum, struct or union type
3049 @emph{definition} and the @code{packed} attribute only past the closing
3050 brace of a definition.
3052 You may also specify attributes between the enum, struct or union
3053 tag and the name of the type rather than after the closing brace.
3055 @xref{Attribute Syntax}, for details of the exact syntax for using
3059 @cindex @code{aligned} attribute
3060 @item aligned (@var{alignment})
3061 This attribute specifies a minimum alignment (in bytes) for variables
3062 of the specified type. For example, the declarations:
3065 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3066 typedef int more_aligned_int __attribute__ ((aligned (8)));
3070 force the compiler to insure (as far as it can) that each variable whose
3071 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3072 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3073 variables of type @code{struct S} aligned to 8-byte boundaries allows
3074 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3075 store) instructions when copying one variable of type @code{struct S} to
3076 another, thus improving run-time efficiency.
3078 Note that the alignment of any given @code{struct} or @code{union} type
3079 is required by the ISO C standard to be at least a perfect multiple of
3080 the lowest common multiple of the alignments of all of the members of
3081 the @code{struct} or @code{union} in question. This means that you @emph{can}
3082 effectively adjust the alignment of a @code{struct} or @code{union}
3083 type by attaching an @code{aligned} attribute to any one of the members
3084 of such a type, but the notation illustrated in the example above is a
3085 more obvious, intuitive, and readable way to request the compiler to
3086 adjust the alignment of an entire @code{struct} or @code{union} type.
3088 As in the preceding example, you can explicitly specify the alignment
3089 (in bytes) that you wish the compiler to use for a given @code{struct}
3090 or @code{union} type. Alternatively, you can leave out the alignment factor
3091 and just ask the compiler to align a type to the maximum
3092 useful alignment for the target machine you are compiling for. For
3093 example, you could write:
3096 struct S @{ short f[3]; @} __attribute__ ((aligned));
3099 Whenever you leave out the alignment factor in an @code{aligned}
3100 attribute specification, the compiler automatically sets the alignment
3101 for the type to the largest alignment which is ever used for any data
3102 type on the target machine you are compiling for. Doing this can often
3103 make copy operations more efficient, because the compiler can use
3104 whatever instructions copy the biggest chunks of memory when performing
3105 copies to or from the variables which have types that you have aligned
3108 In the example above, if the size of each @code{short} is 2 bytes, then
3109 the size of the entire @code{struct S} type is 6 bytes. The smallest
3110 power of two which is greater than or equal to that is 8, so the
3111 compiler sets the alignment for the entire @code{struct S} type to 8
3114 Note that although you can ask the compiler to select a time-efficient
3115 alignment for a given type and then declare only individual stand-alone
3116 objects of that type, the compiler's ability to select a time-efficient
3117 alignment is primarily useful only when you plan to create arrays of
3118 variables having the relevant (efficiently aligned) type. If you
3119 declare or use arrays of variables of an efficiently-aligned type, then
3120 it is likely that your program will also be doing pointer arithmetic (or
3121 subscripting, which amounts to the same thing) on pointers to the
3122 relevant type, and the code that the compiler generates for these
3123 pointer arithmetic operations will often be more efficient for
3124 efficiently-aligned types than for other types.
3126 The @code{aligned} attribute can only increase the alignment; but you
3127 can decrease it by specifying @code{packed} as well. See below.
3129 Note that the effectiveness of @code{aligned} attributes may be limited
3130 by inherent limitations in your linker. On many systems, the linker is
3131 only able to arrange for variables to be aligned up to a certain maximum
3132 alignment. (For some linkers, the maximum supported alignment may
3133 be very very small.) If your linker is only able to align variables
3134 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3135 in an @code{__attribute__} will still only provide you with 8 byte
3136 alignment. See your linker documentation for further information.
3139 This attribute, attached to @code{struct} or @code{union} type
3140 definition, specifies that each member of the structure or union is
3141 placed to minimize the memory required. When attached to an @code{enum}
3142 definition, it indicates that the smallest integral type should be used.
3144 @opindex fshort-enums
3145 Specifying this attribute for @code{struct} and @code{union} types is
3146 equivalent to specifying the @code{packed} attribute on each of the
3147 structure or union members. Specifying the @option{-fshort-enums}
3148 flag on the line is equivalent to specifying the @code{packed}
3149 attribute on all @code{enum} definitions.
3151 In the following example @code{struct my_packed_struct}'s members are
3152 packed closely together, but the internal layout of its @code{s} member
3153 is not packed -- to do that, @code{struct my_unpacked_struct} would need to
3157 struct my_unpacked_struct
3163 struct my_packed_struct __attribute__ ((__packed__))
3167 struct my_unpacked_struct s;
3171 You may only specify this attribute on the definition of a @code{enum},
3172 @code{struct} or @code{union}, not on a @code{typedef} which does not
3173 also define the enumerated type, structure or union.
3175 @item transparent_union
3176 This attribute, attached to a @code{union} type definition, indicates
3177 that any function parameter having that union type causes calls to that
3178 function to be treated in a special way.
3180 First, the argument corresponding to a transparent union type can be of
3181 any type in the union; no cast is required. Also, if the union contains
3182 a pointer type, the corresponding argument can be a null pointer
3183 constant or a void pointer expression; and if the union contains a void
3184 pointer type, the corresponding argument can be any pointer expression.
3185 If the union member type is a pointer, qualifiers like @code{const} on
3186 the referenced type must be respected, just as with normal pointer
3189 Second, the argument is passed to the function using the calling
3190 conventions of the first member of the transparent union, not the calling
3191 conventions of the union itself. All members of the union must have the
3192 same machine representation; this is necessary for this argument passing
3195 Transparent unions are designed for library functions that have multiple
3196 interfaces for compatibility reasons. For example, suppose the
3197 @code{wait} function must accept either a value of type @code{int *} to
3198 comply with Posix, or a value of type @code{union wait *} to comply with
3199 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3200 @code{wait} would accept both kinds of arguments, but it would also
3201 accept any other pointer type and this would make argument type checking
3202 less useful. Instead, @code{<sys/wait.h>} might define the interface
3210 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3212 pid_t wait (wait_status_ptr_t);
3215 This interface allows either @code{int *} or @code{union wait *}
3216 arguments to be passed, using the @code{int *} calling convention.
3217 The program can call @code{wait} with arguments of either type:
3220 int w1 () @{ int w; return wait (&w); @}
3221 int w2 () @{ union wait w; return wait (&w); @}
3224 With this interface, @code{wait}'s implementation might look like this:
3227 pid_t wait (wait_status_ptr_t p)
3229 return waitpid (-1, p.__ip, 0);
3234 When attached to a type (including a @code{union} or a @code{struct}),
3235 this attribute means that variables of that type are meant to appear
3236 possibly unused. GCC will not produce a warning for any variables of
3237 that type, even if the variable appears to do nothing. This is often
3238 the case with lock or thread classes, which are usually defined and then
3239 not referenced, but contain constructors and destructors that have
3240 nontrivial bookkeeping functions.
3243 The @code{deprecated} attribute results in a warning if the type
3244 is used anywhere in the source file. This is useful when identifying
3245 types that are expected to be removed in a future version of a program.
3246 If possible, the warning also includes the location of the declaration
3247 of the deprecated type, to enable users to easily find further
3248 information about why the type is deprecated, or what they should do
3249 instead. Note that the warnings only occur for uses and then only
3250 if the type is being applied to an identifier that itself is not being
3251 declared as deprecated.
3254 typedef int T1 __attribute__ ((deprecated));
3258 typedef T1 T3 __attribute__ ((deprecated));
3259 T3 z __attribute__ ((deprecated));
3262 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3263 warning is issued for line 4 because T2 is not explicitly
3264 deprecated. Line 5 has no warning because T3 is explicitly
3265 deprecated. Similarly for line 6.
3267 The @code{deprecated} attribute can also be used for functions and
3268 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3271 Accesses to objects with types with this attribute are not subjected to
3272 type-based alias analysis, but are instead assumed to be able to alias
3273 any other type of objects, just like the @code{char} type. See
3274 @option{-fstrict-aliasing} for more information on aliasing issues.
3279 typedef short __attribute__((__may_alias__)) short_a;
3285 short_a *b = (short_a *) &a;
3289 if (a == 0x12345678)
3296 If you replaced @code{short_a} with @code{short} in the variable
3297 declaration, the above program would abort when compiled with
3298 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3299 above in recent GCC versions.
3301 @subsection i386 Type Attributes
3303 Two attributes are currently defined for i386 configurations:
3304 @code{ms_struct} and @code{gcc_struct}
3308 @cindex @code{ms_struct}
3309 @cindex @code{gcc_struct}
3311 If @code{packed} is used on a structure, or if bit-fields are used
3312 it may be that the Microsoft ABI packs them differently
3313 than GCC would normally pack them. Particularly when moving packed
3314 data between functions compiled with GCC and the native Microsoft compiler
3315 (either via function call or as data in a file), it may be necessary to access
3318 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3319 compilers to match the native Microsoft compiler.
3322 To specify multiple attributes, separate them by commas within the
3323 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3327 @section An Inline Function is As Fast As a Macro
3328 @cindex inline functions
3329 @cindex integrating function code
3331 @cindex macros, inline alternative
3333 By declaring a function @code{inline}, you can direct GCC to
3334 integrate that function's code into the code for its callers. This
3335 makes execution faster by eliminating the function-call overhead; in
3336 addition, if any of the actual argument values are constant, their known
3337 values may permit simplifications at compile time so that not all of the
3338 inline function's code needs to be included. The effect on code size is
3339 less predictable; object code may be larger or smaller with function
3340 inlining, depending on the particular case. Inlining of functions is an
3341 optimization and it really ``works'' only in optimizing compilation. If
3342 you don't use @option{-O}, no function is really inline.
3344 Inline functions are included in the ISO C99 standard, but there are
3345 currently substantial differences between what GCC implements and what
3346 the ISO C99 standard requires.
3348 To declare a function inline, use the @code{inline} keyword in its
3349 declaration, like this:
3359 (If you are writing a header file to be included in ISO C programs, write
3360 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3361 You can also make all ``simple enough'' functions inline with the option
3362 @option{-finline-functions}.
3365 Note that certain usages in a function definition can make it unsuitable
3366 for inline substitution. Among these usages are: use of varargs, use of
3367 alloca, use of variable sized data types (@pxref{Variable Length}),
3368 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3369 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3370 will warn when a function marked @code{inline} could not be substituted,
3371 and will give the reason for the failure.
3373 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3374 does not affect the linkage of the function.
3376 @cindex automatic @code{inline} for C++ member fns
3377 @cindex @code{inline} automatic for C++ member fns
3378 @cindex member fns, automatically @code{inline}
3379 @cindex C++ member fns, automatically @code{inline}
3380 @opindex fno-default-inline
3381 GCC automatically inlines member functions defined within the class
3382 body of C++ programs even if they are not explicitly declared
3383 @code{inline}. (You can override this with @option{-fno-default-inline};
3384 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3386 @cindex inline functions, omission of
3387 @opindex fkeep-inline-functions
3388 When a function is both inline and @code{static}, if all calls to the
3389 function are integrated into the caller, and the function's address is
3390 never used, then the function's own assembler code is never referenced.
3391 In this case, GCC does not actually output assembler code for the
3392 function, unless you specify the option @option{-fkeep-inline-functions}.
3393 Some calls cannot be integrated for various reasons (in particular,
3394 calls that precede the function's definition cannot be integrated, and
3395 neither can recursive calls within the definition). If there is a
3396 nonintegrated call, then the function is compiled to assembler code as
3397 usual. The function must also be compiled as usual if the program
3398 refers to its address, because that can't be inlined.
3400 @cindex non-static inline function
3401 When an inline function is not @code{static}, then the compiler must assume
3402 that there may be calls from other source files; since a global symbol can
3403 be defined only once in any program, the function must not be defined in
3404 the other source files, so the calls therein cannot be integrated.
3405 Therefore, a non-@code{static} inline function is always compiled on its
3406 own in the usual fashion.
3408 If you specify both @code{inline} and @code{extern} in the function
3409 definition, then the definition is used only for inlining. In no case
3410 is the function compiled on its own, not even if you refer to its
3411 address explicitly. Such an address becomes an external reference, as
3412 if you had only declared the function, and had not defined it.
3414 This combination of @code{inline} and @code{extern} has almost the
3415 effect of a macro. The way to use it is to put a function definition in
3416 a header file with these keywords, and put another copy of the
3417 definition (lacking @code{inline} and @code{extern}) in a library file.
3418 The definition in the header file will cause most calls to the function
3419 to be inlined. If any uses of the function remain, they will refer to
3420 the single copy in the library.
3422 Since GCC eventually will implement ISO C99 semantics for
3423 inline functions, it is best to use @code{static inline} only
3424 to guarantee compatibility. (The
3425 existing semantics will remain available when @option{-std=gnu89} is
3426 specified, but eventually the default will be @option{-std=gnu99} and
3427 that will implement the C99 semantics, though it does not do so yet.)
3429 GCC does not inline any functions when not optimizing unless you specify
3430 the @samp{always_inline} attribute for the function, like this:
3434 inline void foo (const char) __attribute__((always_inline));
3438 @section Assembler Instructions with C Expression Operands
3439 @cindex extended @code{asm}
3440 @cindex @code{asm} expressions
3441 @cindex assembler instructions
3444 In an assembler instruction using @code{asm}, you can specify the
3445 operands of the instruction using C expressions. This means you need not
3446 guess which registers or memory locations will contain the data you want
3449 You must specify an assembler instruction template much like what
3450 appears in a machine description, plus an operand constraint string for
3453 For example, here is how to use the 68881's @code{fsinx} instruction:
3456 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3460 Here @code{angle} is the C expression for the input operand while
3461 @code{result} is that of the output operand. Each has @samp{"f"} as its
3462 operand constraint, saying that a floating point register is required.
3463 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3464 output operands' constraints must use @samp{=}. The constraints use the
3465 same language used in the machine description (@pxref{Constraints}).
3467 Each operand is described by an operand-constraint string followed by
3468 the C expression in parentheses. A colon separates the assembler
3469 template from the first output operand and another separates the last
3470 output operand from the first input, if any. Commas separate the
3471 operands within each group. The total number of operands is currently
3472 limited to 30; this limitation may be lifted in some future version of
3475 If there are no output operands but there are input operands, you must
3476 place two consecutive colons surrounding the place where the output
3479 As of GCC version 3.1, it is also possible to specify input and output
3480 operands using symbolic names which can be referenced within the
3481 assembler code. These names are specified inside square brackets
3482 preceding the constraint string, and can be referenced inside the
3483 assembler code using @code{%[@var{name}]} instead of a percentage sign
3484 followed by the operand number. Using named operands the above example
3488 asm ("fsinx %[angle],%[output]"
3489 : [output] "=f" (result)
3490 : [angle] "f" (angle));
3494 Note that the symbolic operand names have no relation whatsoever to
3495 other C identifiers. You may use any name you like, even those of
3496 existing C symbols, but you must ensure that no two operands within the same
3497 assembler construct use the same symbolic name.
3499 Output operand expressions must be lvalues; the compiler can check this.
3500 The input operands need not be lvalues. The compiler cannot check
3501 whether the operands have data types that are reasonable for the
3502 instruction being executed. It does not parse the assembler instruction
3503 template and does not know what it means or even whether it is valid
3504 assembler input. The extended @code{asm} feature is most often used for
3505 machine instructions the compiler itself does not know exist. If
3506 the output expression cannot be directly addressed (for example, it is a
3507 bit-field), your constraint must allow a register. In that case, GCC
3508 will use the register as the output of the @code{asm}, and then store
3509 that register into the output.
3511 The ordinary output operands must be write-only; GCC will assume that
3512 the values in these operands before the instruction are dead and need
3513 not be generated. Extended asm supports input-output or read-write
3514 operands. Use the constraint character @samp{+} to indicate such an
3515 operand and list it with the output operands. You should only use
3516 read-write operands when the constraints for the operand (or the
3517 operand in which only some of the bits are to be changed) allow a
3520 You may, as an alternative, logically split its function into two
3521 separate operands, one input operand and one write-only output
3522 operand. The connection between them is expressed by constraints
3523 which say they need to be in the same location when the instruction
3524 executes. You can use the same C expression for both operands, or
3525 different expressions. For example, here we write the (fictitious)
3526 @samp{combine} instruction with @code{bar} as its read-only source
3527 operand and @code{foo} as its read-write destination:
3530 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3534 The constraint @samp{"0"} for operand 1 says that it must occupy the
3535 same location as operand 0. A number in constraint is allowed only in
3536 an input operand and it must refer to an output operand.
3538 Only a number in the constraint can guarantee that one operand will be in
3539 the same place as another. The mere fact that @code{foo} is the value
3540 of both operands is not enough to guarantee that they will be in the
3541 same place in the generated assembler code. The following would not
3545 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3548 Various optimizations or reloading could cause operands 0 and 1 to be in
3549 different registers; GCC knows no reason not to do so. For example, the
3550 compiler might find a copy of the value of @code{foo} in one register and
3551 use it for operand 1, but generate the output operand 0 in a different
3552 register (copying it afterward to @code{foo}'s own address). Of course,
3553 since the register for operand 1 is not even mentioned in the assembler
3554 code, the result will not work, but GCC can't tell that.
3556 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3557 the operand number for a matching constraint. For example:
3560 asm ("cmoveq %1,%2,%[result]"
3561 : [result] "=r"(result)
3562 : "r" (test), "r"(new), "[result]"(old));
3565 Sometimes you need to make an @code{asm} operand be a specific register,
3566 but there's no matching constraint letter for that register @emph{by
3567 itself}. To force the operand into that register, use a local variable
3568 for the operand and specify the register in the variable declaration.
3569 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
3570 register constraint letter that matches the register:
3573 register int *p1 asm ("r0") = @dots{};
3574 register int *p2 asm ("r1") = @dots{};
3575 register int *result asm ("r0");
3576 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3579 Some instructions clobber specific hard registers. To describe this,
3580 write a third colon after the input operands, followed by the names of
3581 the clobbered hard registers (given as strings). Here is a realistic
3582 example for the VAX:
3585 asm volatile ("movc3 %0,%1,%2"
3587 : "g" (from), "g" (to), "g" (count)
3588 : "r0", "r1", "r2", "r3", "r4", "r5");
3591 You may not write a clobber description in a way that overlaps with an
3592 input or output operand. For example, you may not have an operand
3593 describing a register class with one member if you mention that register
3594 in the clobber list. Variables declared to live in specific registers
3595 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
3596 have no part mentioned in the clobber description.
3597 There is no way for you to specify that an input
3598 operand is modified without also specifying it as an output
3599 operand. Note that if all the output operands you specify are for this
3600 purpose (and hence unused), you will then also need to specify
3601 @code{volatile} for the @code{asm} construct, as described below, to
3602 prevent GCC from deleting the @code{asm} statement as unused.
3604 If you refer to a particular hardware register from the assembler code,
3605 you will probably have to list the register after the third colon to
3606 tell the compiler the register's value is modified. In some assemblers,
3607 the register names begin with @samp{%}; to produce one @samp{%} in the
3608 assembler code, you must write @samp{%%} in the input.
3610 If your assembler instruction can alter the condition code register, add
3611 @samp{cc} to the list of clobbered registers. GCC on some machines
3612 represents the condition codes as a specific hardware register;
3613 @samp{cc} serves to name this register. On other machines, the
3614 condition code is handled differently, and specifying @samp{cc} has no
3615 effect. But it is valid no matter what the machine.
3617 If your assembler instructions access memory in an unpredictable
3618 fashion, add @samp{memory} to the list of clobbered registers. This
3619 will cause GCC to not keep memory values cached in registers across the
3620 assembler instruction and not optimize stores or loads to that memory.
3621 You will also want to add the @code{volatile} keyword if the memory
3622 affected is not listed in the inputs or outputs of the @code{asm}, as
3623 the @samp{memory} clobber does not count as a side-effect of the
3624 @code{asm}. If you know how large the accessed memory is, you can add
3625 it as input or output but if this is not known, you should add
3626 @samp{memory}. As an example, if you access ten bytes of a string, you
3627 can use a memory input like:
3630 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
3633 Note that in the following example the memory input is necessary,
3634 otherwise GCC might optimize the store to @code{x} away:
3641 asm ("magic stuff accessing an 'int' pointed to by '%1'"
3642 "=&d" (r) : "a" (y), "m" (*y));
3647 You can put multiple assembler instructions together in a single
3648 @code{asm} template, separated by the characters normally used in assembly
3649 code for the system. A combination that works in most places is a newline
3650 to break the line, plus a tab character to move to the instruction field
3651 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
3652 assembler allows semicolons as a line-breaking character. Note that some
3653 assembler dialects use semicolons to start a comment.
3654 The input operands are guaranteed not to use any of the clobbered
3655 registers, and neither will the output operands' addresses, so you can
3656 read and write the clobbered registers as many times as you like. Here
3657 is an example of multiple instructions in a template; it assumes the
3658 subroutine @code{_foo} accepts arguments in registers 9 and 10:
3661 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
3663 : "g" (from), "g" (to)
3667 Unless an output operand has the @samp{&} constraint modifier, GCC
3668 may allocate it in the same register as an unrelated input operand, on
3669 the assumption the inputs are consumed before the outputs are produced.
3670 This assumption may be false if the assembler code actually consists of
3671 more than one instruction. In such a case, use @samp{&} for each output
3672 operand that may not overlap an input. @xref{Modifiers}.
3674 If you want to test the condition code produced by an assembler
3675 instruction, you must include a branch and a label in the @code{asm}
3676 construct, as follows:
3679 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
3685 This assumes your assembler supports local labels, as the GNU assembler
3686 and most Unix assemblers do.
3688 Speaking of labels, jumps from one @code{asm} to another are not
3689 supported. The compiler's optimizers do not know about these jumps, and
3690 therefore they cannot take account of them when deciding how to
3693 @cindex macros containing @code{asm}
3694 Usually the most convenient way to use these @code{asm} instructions is to
3695 encapsulate them in macros that look like functions. For example,
3699 (@{ double __value, __arg = (x); \
3700 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
3705 Here the variable @code{__arg} is used to make sure that the instruction
3706 operates on a proper @code{double} value, and to accept only those
3707 arguments @code{x} which can convert automatically to a @code{double}.
3709 Another way to make sure the instruction operates on the correct data
3710 type is to use a cast in the @code{asm}. This is different from using a
3711 variable @code{__arg} in that it converts more different types. For
3712 example, if the desired type were @code{int}, casting the argument to
3713 @code{int} would accept a pointer with no complaint, while assigning the
3714 argument to an @code{int} variable named @code{__arg} would warn about
3715 using a pointer unless the caller explicitly casts it.
3717 If an @code{asm} has output operands, GCC assumes for optimization
3718 purposes the instruction has no side effects except to change the output
3719 operands. This does not mean instructions with a side effect cannot be
3720 used, but you must be careful, because the compiler may eliminate them
3721 if the output operands aren't used, or move them out of loops, or
3722 replace two with one if they constitute a common subexpression. Also,
3723 if your instruction does have a side effect on a variable that otherwise
3724 appears not to change, the old value of the variable may be reused later
3725 if it happens to be found in a register.
3727 You can prevent an @code{asm} instruction from being deleted, moved
3728 significantly, or combined, by writing the keyword @code{volatile} after
3729 the @code{asm}. For example:
3732 #define get_and_set_priority(new) \
3734 asm volatile ("get_and_set_priority %0, %1" \
3735 : "=g" (__old) : "g" (new)); \
3740 If you write an @code{asm} instruction with no outputs, GCC will know
3741 the instruction has side-effects and will not delete the instruction or
3742 move it outside of loops.
3744 The @code{volatile} keyword indicates that the instruction has
3745 important side-effects. GCC will not delete a volatile @code{asm} if
3746 it is reachable. (The instruction can still be deleted if GCC can
3747 prove that control-flow will never reach the location of the
3748 instruction.) In addition, GCC will not reschedule instructions
3749 across a volatile @code{asm} instruction. For example:
3752 *(volatile int *)addr = foo;
3753 asm volatile ("eieio" : : );
3757 Assume @code{addr} contains the address of a memory mapped device
3758 register. The PowerPC @code{eieio} instruction (Enforce In-order
3759 Execution of I/O) tells the CPU to make sure that the store to that
3760 device register happens before it issues any other I/O@.
3762 Note that even a volatile @code{asm} instruction can be moved in ways
3763 that appear insignificant to the compiler, such as across jump
3764 instructions. You can't expect a sequence of volatile @code{asm}
3765 instructions to remain perfectly consecutive. If you want consecutive
3766 output, use a single @code{asm}. Also, GCC will perform some
3767 optimizations across a volatile @code{asm} instruction; GCC does not
3768 ``forget everything'' when it encounters a volatile @code{asm}
3769 instruction the way some other compilers do.
3771 An @code{asm} instruction without any operands or clobbers (an ``old
3772 style'' @code{asm}) will be treated identically to a volatile
3773 @code{asm} instruction.
3775 It is a natural idea to look for a way to give access to the condition
3776 code left by the assembler instruction. However, when we attempted to
3777 implement this, we found no way to make it work reliably. The problem
3778 is that output operands might need reloading, which would result in
3779 additional following ``store'' instructions. On most machines, these
3780 instructions would alter the condition code before there was time to
3781 test it. This problem doesn't arise for ordinary ``test'' and
3782 ``compare'' instructions because they don't have any output operands.
3784 For reasons similar to those described above, it is not possible to give
3785 an assembler instruction access to the condition code left by previous
3788 If you are writing a header file that should be includable in ISO C
3789 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
3792 @subsection Size of an @code{asm}
3794 Some targets require that GCC track the size of each instruction used in
3795 order to generate correct code. Because the final length of an
3796 @code{asm} is only known by the assembler, GCC must make an estimate as
3797 to how big it will be. The estimate is formed by counting the number of
3798 statements in the pattern of the @code{asm} and multiplying that by the
3799 length of the longest instruction on that processor. Statements in the
3800 @code{asm} are identified by newline characters and whatever statement
3801 separator characters are supported by the assembler; on most processors
3802 this is the `@code{;}' character.
3804 Normally, GCC's estimate is perfectly adequate to ensure that correct
3805 code is generated, but it is possible to confuse the compiler if you use
3806 pseudo instructions or assembler macros that expand into multiple real
3807 instructions or if you use assembler directives that expand to more
3808 space in the object file than would be needed for a single instruction.
3809 If this happens then the assembler will produce a diagnostic saying that
3810 a label is unreachable.
3812 @subsection i386 floating point asm operands
3814 There are several rules on the usage of stack-like regs in
3815 asm_operands insns. These rules apply only to the operands that are
3820 Given a set of input regs that die in an asm_operands, it is
3821 necessary to know which are implicitly popped by the asm, and
3822 which must be explicitly popped by gcc.
3824 An input reg that is implicitly popped by the asm must be
3825 explicitly clobbered, unless it is constrained to match an
3829 For any input reg that is implicitly popped by an asm, it is
3830 necessary to know how to adjust the stack to compensate for the pop.
3831 If any non-popped input is closer to the top of the reg-stack than
3832 the implicitly popped reg, it would not be possible to know what the
3833 stack looked like---it's not clear how the rest of the stack ``slides
3836 All implicitly popped input regs must be closer to the top of
3837 the reg-stack than any input that is not implicitly popped.
3839 It is possible that if an input dies in an insn, reload might
3840 use the input reg for an output reload. Consider this example:
3843 asm ("foo" : "=t" (a) : "f" (b));
3846 This asm says that input B is not popped by the asm, and that
3847 the asm pushes a result onto the reg-stack, i.e., the stack is one
3848 deeper after the asm than it was before. But, it is possible that
3849 reload will think that it can use the same reg for both the input and
3850 the output, if input B dies in this insn.
3852 If any input operand uses the @code{f} constraint, all output reg
3853 constraints must use the @code{&} earlyclobber.
3855 The asm above would be written as
3858 asm ("foo" : "=&t" (a) : "f" (b));
3862 Some operands need to be in particular places on the stack. All
3863 output operands fall in this category---there is no other way to
3864 know which regs the outputs appear in unless the user indicates
3865 this in the constraints.
3867 Output operands must specifically indicate which reg an output
3868 appears in after an asm. @code{=f} is not allowed: the operand
3869 constraints must select a class with a single reg.
3872 Output operands may not be ``inserted'' between existing stack regs.
3873 Since no 387 opcode uses a read/write operand, all output operands
3874 are dead before the asm_operands, and are pushed by the asm_operands.
3875 It makes no sense to push anywhere but the top of the reg-stack.
3877 Output operands must start at the top of the reg-stack: output
3878 operands may not ``skip'' a reg.
3881 Some asm statements may need extra stack space for internal
3882 calculations. This can be guaranteed by clobbering stack registers
3883 unrelated to the inputs and outputs.
3887 Here are a couple of reasonable asms to want to write. This asm
3888 takes one input, which is internally popped, and produces two outputs.
3891 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
3894 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
3895 and replaces them with one output. The user must code the @code{st(1)}
3896 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
3899 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
3905 @section Controlling Names Used in Assembler Code
3906 @cindex assembler names for identifiers
3907 @cindex names used in assembler code
3908 @cindex identifiers, names in assembler code
3910 You can specify the name to be used in the assembler code for a C
3911 function or variable by writing the @code{asm} (or @code{__asm__})
3912 keyword after the declarator as follows:
3915 int foo asm ("myfoo") = 2;
3919 This specifies that the name to be used for the variable @code{foo} in
3920 the assembler code should be @samp{myfoo} rather than the usual
3923 On systems where an underscore is normally prepended to the name of a C
3924 function or variable, this feature allows you to define names for the
3925 linker that do not start with an underscore.
3927 It does not make sense to use this feature with a non-static local
3928 variable since such variables do not have assembler names. If you are
3929 trying to put the variable in a particular register, see @ref{Explicit
3930 Reg Vars}. GCC presently accepts such code with a warning, but will
3931 probably be changed to issue an error, rather than a warning, in the
3934 You cannot use @code{asm} in this way in a function @emph{definition}; but
3935 you can get the same effect by writing a declaration for the function
3936 before its definition and putting @code{asm} there, like this:
3939 extern func () asm ("FUNC");
3946 It is up to you to make sure that the assembler names you choose do not
3947 conflict with any other assembler symbols. Also, you must not use a
3948 register name; that would produce completely invalid assembler code. GCC
3949 does not as yet have the ability to store static variables in registers.
3950 Perhaps that will be added.
3952 @node Explicit Reg Vars
3953 @section Variables in Specified Registers
3954 @cindex explicit register variables
3955 @cindex variables in specified registers
3956 @cindex specified registers
3957 @cindex registers, global allocation
3959 GNU C allows you to put a few global variables into specified hardware
3960 registers. You can also specify the register in which an ordinary
3961 register variable should be allocated.
3965 Global register variables reserve registers throughout the program.
3966 This may be useful in programs such as programming language
3967 interpreters which have a couple of global variables that are accessed
3971 Local register variables in specific registers do not reserve the
3972 registers, except at the point where they are used as input or output
3973 operands in an @code{asm} statement and the @code{asm} statement itself is
3974 not deleted. The compiler's data flow analysis is capable of determining
3975 where the specified registers contain live values, and where they are
3976 available for other uses. Stores into local register variables may be deleted
3977 when they appear to be dead according to dataflow analysis. References
3978 to local register variables may be deleted or moved or simplified.
3980 These local variables are sometimes convenient for use with the extended
3981 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
3982 output of the assembler instruction directly into a particular register.
3983 (This will work provided the register you specify fits the constraints
3984 specified for that operand in the @code{asm}.)
3992 @node Global Reg Vars
3993 @subsection Defining Global Register Variables
3994 @cindex global register variables
3995 @cindex registers, global variables in
3997 You can define a global register variable in GNU C like this:
4000 register int *foo asm ("a5");
4004 Here @code{a5} is the name of the register which should be used. Choose a
4005 register which is normally saved and restored by function calls on your
4006 machine, so that library routines will not clobber it.
4008 Naturally the register name is cpu-dependent, so you would need to
4009 conditionalize your program according to cpu type. The register
4010 @code{a5} would be a good choice on a 68000 for a variable of pointer
4011 type. On machines with register windows, be sure to choose a ``global''
4012 register that is not affected magically by the function call mechanism.
4014 In addition, operating systems on one type of cpu may differ in how they
4015 name the registers; then you would need additional conditionals. For
4016 example, some 68000 operating systems call this register @code{%a5}.
4018 Eventually there may be a way of asking the compiler to choose a register
4019 automatically, but first we need to figure out how it should choose and
4020 how to enable you to guide the choice. No solution is evident.
4022 Defining a global register variable in a certain register reserves that
4023 register entirely for this use, at least within the current compilation.
4024 The register will not be allocated for any other purpose in the functions
4025 in the current compilation. The register will not be saved and restored by
4026 these functions. Stores into this register are never deleted even if they
4027 would appear to be dead, but references may be deleted or moved or
4030 It is not safe to access the global register variables from signal
4031 handlers, or from more than one thread of control, because the system
4032 library routines may temporarily use the register for other things (unless
4033 you recompile them specially for the task at hand).
4035 @cindex @code{qsort}, and global register variables
4036 It is not safe for one function that uses a global register variable to
4037 call another such function @code{foo} by way of a third function
4038 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4039 different source file in which the variable wasn't declared). This is
4040 because @code{lose} might save the register and put some other value there.
4041 For example, you can't expect a global register variable to be available in
4042 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4043 might have put something else in that register. (If you are prepared to
4044 recompile @code{qsort} with the same global register variable, you can
4045 solve this problem.)
4047 If you want to recompile @code{qsort} or other source files which do not
4048 actually use your global register variable, so that they will not use that
4049 register for any other purpose, then it suffices to specify the compiler
4050 option @option{-ffixed-@var{reg}}. You need not actually add a global
4051 register declaration to their source code.
4053 A function which can alter the value of a global register variable cannot
4054 safely be called from a function compiled without this variable, because it
4055 could clobber the value the caller expects to find there on return.
4056 Therefore, the function which is the entry point into the part of the
4057 program that uses the global register variable must explicitly save and
4058 restore the value which belongs to its caller.
4060 @cindex register variable after @code{longjmp}
4061 @cindex global register after @code{longjmp}
4062 @cindex value after @code{longjmp}
4065 On most machines, @code{longjmp} will restore to each global register
4066 variable the value it had at the time of the @code{setjmp}. On some
4067 machines, however, @code{longjmp} will not change the value of global
4068 register variables. To be portable, the function that called @code{setjmp}
4069 should make other arrangements to save the values of the global register
4070 variables, and to restore them in a @code{longjmp}. This way, the same
4071 thing will happen regardless of what @code{longjmp} does.
4073 All global register variable declarations must precede all function
4074 definitions. If such a declaration could appear after function
4075 definitions, the declaration would be too late to prevent the register from
4076 being used for other purposes in the preceding functions.
4078 Global register variables may not have initial values, because an
4079 executable file has no means to supply initial contents for a register.
4081 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4082 registers, but certain library functions, such as @code{getwd}, as well
4083 as the subroutines for division and remainder, modify g3 and g4. g1 and
4084 g2 are local temporaries.
4086 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4087 Of course, it will not do to use more than a few of those.
4089 @node Local Reg Vars
4090 @subsection Specifying Registers for Local Variables
4091 @cindex local variables, specifying registers
4092 @cindex specifying registers for local variables
4093 @cindex registers for local variables
4095 You can define a local register variable with a specified register
4099 register int *foo asm ("a5");
4103 Here @code{a5} is the name of the register which should be used. Note
4104 that this is the same syntax used for defining global register
4105 variables, but for a local variable it would appear within a function.
4107 Naturally the register name is cpu-dependent, but this is not a
4108 problem, since specific registers are most often useful with explicit
4109 assembler instructions (@pxref{Extended Asm}). Both of these things
4110 generally require that you conditionalize your program according to
4113 In addition, operating systems on one type of cpu may differ in how they
4114 name the registers; then you would need additional conditionals. For
4115 example, some 68000 operating systems call this register @code{%a5}.
4117 Defining such a register variable does not reserve the register; it
4118 remains available for other uses in places where flow control determines
4119 the variable's value is not live.
4121 This option does not guarantee that GCC will generate code that has
4122 this variable in the register you specify at all times. You may not
4123 code an explicit reference to this register in the @emph{assembler
4124 instruction template} part of an @code{asm} statement and assume it will
4125 always refer to this variable. However, using the variable as an
4126 @code{asm} @emph{operand} guarantees that the specified register is used
4129 Stores into local register variables may be deleted when they appear to be dead
4130 according to dataflow analysis. References to local register variables may
4131 be deleted or moved or simplified.
4133 @node Alternate Keywords
4134 @section Alternate Keywords
4135 @cindex alternate keywords
4136 @cindex keywords, alternate
4138 @option{-ansi} and the various @option{-std} options disable certain
4139 keywords. This causes trouble when you want to use GNU C extensions, or
4140 a general-purpose header file that should be usable by all programs,
4141 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4142 @code{inline} are not available in programs compiled with
4143 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4144 program compiled with @option{-std=c99}). The ISO C99 keyword
4145 @code{restrict} is only available when @option{-std=gnu99} (which will
4146 eventually be the default) or @option{-std=c99} (or the equivalent
4147 @option{-std=iso9899:1999}) is used.
4149 The way to solve these problems is to put @samp{__} at the beginning and
4150 end of each problematical keyword. For example, use @code{__asm__}
4151 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4153 Other C compilers won't accept these alternative keywords; if you want to
4154 compile with another compiler, you can define the alternate keywords as
4155 macros to replace them with the customary keywords. It looks like this:
4163 @findex __extension__
4165 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4167 prevent such warnings within one expression by writing
4168 @code{__extension__} before the expression. @code{__extension__} has no
4169 effect aside from this.
4171 @node Incomplete Enums
4172 @section Incomplete @code{enum} Types
4174 You can define an @code{enum} tag without specifying its possible values.
4175 This results in an incomplete type, much like what you get if you write
4176 @code{struct foo} without describing the elements. A later declaration
4177 which does specify the possible values completes the type.
4179 You can't allocate variables or storage using the type while it is
4180 incomplete. However, you can work with pointers to that type.
4182 This extension may not be very useful, but it makes the handling of
4183 @code{enum} more consistent with the way @code{struct} and @code{union}
4186 This extension is not supported by GNU C++.
4188 @node Function Names
4189 @section Function Names as Strings
4190 @cindex @code{__func__} identifier
4191 @cindex @code{__FUNCTION__} identifier
4192 @cindex @code{__PRETTY_FUNCTION__} identifier
4194 GCC provides three magic variables which hold the name of the current
4195 function, as a string. The first of these is @code{__func__}, which
4196 is part of the C99 standard:
4199 The identifier @code{__func__} is implicitly declared by the translator
4200 as if, immediately following the opening brace of each function
4201 definition, the declaration
4204 static const char __func__[] = "function-name";
4207 appeared, where function-name is the name of the lexically-enclosing
4208 function. This name is the unadorned name of the function.
4211 @code{__FUNCTION__} is another name for @code{__func__}. Older
4212 versions of GCC recognize only this name. However, it is not
4213 standardized. For maximum portability, we recommend you use
4214 @code{__func__}, but provide a fallback definition with the
4218 #if __STDC_VERSION__ < 199901L
4220 # define __func__ __FUNCTION__
4222 # define __func__ "<unknown>"
4227 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4228 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4229 the type signature of the function as well as its bare name. For
4230 example, this program:
4234 extern int printf (char *, ...);
4241 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4242 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4260 __PRETTY_FUNCTION__ = void a::sub(int)
4263 These identifiers are not preprocessor macros. In GCC 3.3 and
4264 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4265 were treated as string literals; they could be used to initialize
4266 @code{char} arrays, and they could be concatenated with other string
4267 literals. GCC 3.4 and later treat them as variables, like
4268 @code{__func__}. In C++, @code{__FUNCTION__} and
4269 @code{__PRETTY_FUNCTION__} have always been variables.
4271 @node Return Address
4272 @section Getting the Return or Frame Address of a Function
4274 These functions may be used to get information about the callers of a
4277 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4278 This function returns the return address of the current function, or of
4279 one of its callers. The @var{level} argument is number of frames to
4280 scan up the call stack. A value of @code{0} yields the return address
4281 of the current function, a value of @code{1} yields the return address
4282 of the caller of the current function, and so forth. When inlining
4283 the expected behavior is that the function will return the address of
4284 the function that will be returned to. To work around this behavior use
4285 the @code{noinline} function attribute.
4287 The @var{level} argument must be a constant integer.
4289 On some machines it may be impossible to determine the return address of
4290 any function other than the current one; in such cases, or when the top
4291 of the stack has been reached, this function will return @code{0} or a
4292 random value. In addition, @code{__builtin_frame_address} may be used
4293 to determine if the top of the stack has been reached.
4295 This function should only be used with a nonzero argument for debugging
4299 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4300 This function is similar to @code{__builtin_return_address}, but it
4301 returns the address of the function frame rather than the return address
4302 of the function. Calling @code{__builtin_frame_address} with a value of
4303 @code{0} yields the frame address of the current function, a value of
4304 @code{1} yields the frame address of the caller of the current function,
4307 The frame is the area on the stack which holds local variables and saved
4308 registers. The frame address is normally the address of the first word
4309 pushed on to the stack by the function. However, the exact definition
4310 depends upon the processor and the calling convention. If the processor
4311 has a dedicated frame pointer register, and the function has a frame,
4312 then @code{__builtin_frame_address} will return the value of the frame
4315 On some machines it may be impossible to determine the frame address of
4316 any function other than the current one; in such cases, or when the top
4317 of the stack has been reached, this function will return @code{0} if
4318 the first frame pointer is properly initialized by the startup code.
4320 This function should only be used with a nonzero argument for debugging
4324 @node Vector Extensions
4325 @section Using vector instructions through built-in functions
4327 On some targets, the instruction set contains SIMD vector instructions that
4328 operate on multiple values contained in one large register at the same time.
4329 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4332 The first step in using these extensions is to provide the necessary data
4333 types. This should be done using an appropriate @code{typedef}:
4336 typedef int v4si __attribute__ ((vector_size (16)));
4339 The @code{int} type specifies the base type, while the attribute specifies
4340 the vector size for the variable, measured in bytes. For example, the
4341 declaration above causes the compiler to set the mode for the @code{v4si}
4342 type to be 16 bytes wide and divided into @code{int} sized units. For
4343 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4344 corresponding mode of @code{foo} will be @acronym{V4SI}.
4346 The @code{vector_size} attribute is only applicable to integral and
4347 float scalars, although arrays, pointers, and function return values
4348 are allowed in conjunction with this construct.
4350 All the basic integer types can be used as base types, both as signed
4351 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4352 @code{long long}. In addition, @code{float} and @code{double} can be
4353 used to build floating-point vector types.
4355 Specifying a combination that is not valid for the current architecture
4356 will cause GCC to synthesize the instructions using a narrower mode.
4357 For example, if you specify a variable of type @code{V4SI} and your
4358 architecture does not allow for this specific SIMD type, GCC will
4359 produce code that uses 4 @code{SIs}.
4361 The types defined in this manner can be used with a subset of normal C
4362 operations. Currently, GCC will allow using the following operators
4363 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4365 The operations behave like C++ @code{valarrays}. Addition is defined as
4366 the addition of the corresponding elements of the operands. For
4367 example, in the code below, each of the 4 elements in @var{a} will be
4368 added to the corresponding 4 elements in @var{b} and the resulting
4369 vector will be stored in @var{c}.
4372 typedef int v4si __attribute__ ((vector_size (16)));
4379 Subtraction, multiplication, division, and the logical operations
4380 operate in a similar manner. Likewise, the result of using the unary
4381 minus or complement operators on a vector type is a vector whose
4382 elements are the negative or complemented values of the corresponding
4383 elements in the operand.
4385 You can declare variables and use them in function calls and returns, as
4386 well as in assignments and some casts. You can specify a vector type as
4387 a return type for a function. Vector types can also be used as function
4388 arguments. It is possible to cast from one vector type to another,
4389 provided they are of the same size (in fact, you can also cast vectors
4390 to and from other datatypes of the same size).
4392 You cannot operate between vectors of different lengths or different
4393 signedness without a cast.
4395 A port that supports hardware vector operations, usually provides a set
4396 of built-in functions that can be used to operate on vectors. For
4397 example, a function to add two vectors and multiply the result by a
4398 third could look like this:
4401 v4si f (v4si a, v4si b, v4si c)
4403 v4si tmp = __builtin_addv4si (a, b);
4404 return __builtin_mulv4si (tmp, c);
4411 @findex __builtin_offsetof
4413 GCC implements for both C and C++ a syntactic extension to implement
4414 the @code{offsetof} macro.
4418 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
4420 offsetof_member_designator:
4422 | offsetof_member_designator "." @code{identifier}
4423 | offsetof_member_designator "[" @code{expr} "]"
4426 This extension is sufficient such that
4429 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
4432 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
4433 may be dependent. In either case, @var{member} may consist of a single
4434 identifier, or a sequence of member accesses and array references.
4436 @node Other Builtins
4437 @section Other built-in functions provided by GCC
4438 @cindex built-in functions
4439 @findex __builtin_isgreater
4440 @findex __builtin_isgreaterequal
4441 @findex __builtin_isless
4442 @findex __builtin_islessequal
4443 @findex __builtin_islessgreater
4444 @findex __builtin_isunordered
4599 @findex fprintf_unlocked
4601 @findex fputs_unlocked
4711 @findex printf_unlocked
4740 @findex significandf
4741 @findex significandl
4808 GCC provides a large number of built-in functions other than the ones
4809 mentioned above. Some of these are for internal use in the processing
4810 of exceptions or variable-length argument lists and will not be
4811 documented here because they may change from time to time; we do not
4812 recommend general use of these functions.
4814 The remaining functions are provided for optimization purposes.
4816 @opindex fno-builtin
4817 GCC includes built-in versions of many of the functions in the standard
4818 C library. The versions prefixed with @code{__builtin_} will always be
4819 treated as having the same meaning as the C library function even if you
4820 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
4821 Many of these functions are only optimized in certain cases; if they are
4822 not optimized in a particular case, a call to the library function will
4827 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
4828 @option{-std=c99}), the functions
4829 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
4830 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
4831 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
4832 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
4833 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
4834 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
4835 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
4836 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
4837 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
4838 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
4839 @code{significandf}, @code{significandl}, @code{significand},
4840 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
4841 @code{strdup}, @code{strfmon}, @code{toascii}, @code{y0f}, @code{y0l},
4842 @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
4844 may be handled as built-in functions.
4845 All these functions have corresponding versions
4846 prefixed with @code{__builtin_}, which may be used even in strict C89
4849 The ISO C99 functions
4850 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
4851 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
4852 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
4853 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
4854 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
4855 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
4856 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
4857 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
4858 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
4859 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
4860 @code{cimagl}, @code{cimag}, @code{conjf}, @code{conjl}, @code{conj},
4861 @code{copysignf}, @code{copysignl}, @code{copysign}, @code{cpowf},
4862 @code{cpowl}, @code{cpow}, @code{cprojf}, @code{cprojl}, @code{cproj},
4863 @code{crealf}, @code{creall}, @code{creal}, @code{csinf}, @code{csinhf},
4864 @code{csinhl}, @code{csinh}, @code{csinl}, @code{csin}, @code{csqrtf},
4865 @code{csqrtl}, @code{csqrt}, @code{ctanf}, @code{ctanhf}, @code{ctanhl},
4866 @code{ctanh}, @code{ctanl}, @code{ctan}, @code{erfcf}, @code{erfcl},
4867 @code{erfc}, @code{erff}, @code{erfl}, @code{erf}, @code{exp2f},
4868 @code{exp2l}, @code{exp2}, @code{expm1f}, @code{expm1l}, @code{expm1},
4869 @code{fdimf}, @code{fdiml}, @code{fdim}, @code{fmaf}, @code{fmal},
4870 @code{fmaxf}, @code{fmaxl}, @code{fmax}, @code{fma}, @code{fminf},
4871 @code{fminl}, @code{fmin}, @code{hypotf}, @code{hypotl}, @code{hypot},
4872 @code{ilogbf}, @code{ilogbl}, @code{ilogb}, @code{imaxabs},
4873 @code{isblank}, @code{iswblank}, @code{lgammaf}, @code{lgammal},
4874 @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
4875 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
4876 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
4877 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
4878 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
4879 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
4880 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
4881 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
4882 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
4883 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
4884 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
4885 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
4886 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
4887 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
4888 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
4889 are handled as built-in functions
4890 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
4892 There are also built-in versions of the ISO C99 functions
4893 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
4894 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
4895 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
4896 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
4897 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
4898 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
4899 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
4900 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
4901 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
4902 that are recognized in any mode since ISO C90 reserves these names for
4903 the purpose to which ISO C99 puts them. All these functions have
4904 corresponding versions prefixed with @code{__builtin_}.
4906 The ISO C94 functions
4907 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
4908 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
4909 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
4911 are handled as built-in functions
4912 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
4914 The ISO C90 functions
4915 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
4916 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
4917 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
4918 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
4919 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
4920 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
4921 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
4922 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
4923 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
4924 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
4925 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
4926 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
4927 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
4928 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
4929 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
4930 @code{vprintf} and @code{vsprintf}
4931 are all recognized as built-in functions unless
4932 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
4933 is specified for an individual function). All of these functions have
4934 corresponding versions prefixed with @code{__builtin_}.
4936 GCC provides built-in versions of the ISO C99 floating point comparison
4937 macros that avoid raising exceptions for unordered operands. They have
4938 the same names as the standard macros ( @code{isgreater},
4939 @code{isgreaterequal}, @code{isless}, @code{islessequal},
4940 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
4941 prefixed. We intend for a library implementor to be able to simply
4942 @code{#define} each standard macro to its built-in equivalent.
4944 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
4946 You can use the built-in function @code{__builtin_types_compatible_p} to
4947 determine whether two types are the same.
4949 This built-in function returns 1 if the unqualified versions of the
4950 types @var{type1} and @var{type2} (which are types, not expressions) are
4951 compatible, 0 otherwise. The result of this built-in function can be
4952 used in integer constant expressions.
4954 This built-in function ignores top level qualifiers (e.g., @code{const},
4955 @code{volatile}). For example, @code{int} is equivalent to @code{const
4958 The type @code{int[]} and @code{int[5]} are compatible. On the other
4959 hand, @code{int} and @code{char *} are not compatible, even if the size
4960 of their types, on the particular architecture are the same. Also, the
4961 amount of pointer indirection is taken into account when determining
4962 similarity. Consequently, @code{short *} is not similar to
4963 @code{short **}. Furthermore, two types that are typedefed are
4964 considered compatible if their underlying types are compatible.
4966 An @code{enum} type is not considered to be compatible with another
4967 @code{enum} type even if both are compatible with the same integer
4968 type; this is what the C standard specifies.
4969 For example, @code{enum @{foo, bar@}} is not similar to
4970 @code{enum @{hot, dog@}}.
4972 You would typically use this function in code whose execution varies
4973 depending on the arguments' types. For example:
4979 if (__builtin_types_compatible_p (typeof (x), long double)) \
4980 tmp = foo_long_double (tmp); \
4981 else if (__builtin_types_compatible_p (typeof (x), double)) \
4982 tmp = foo_double (tmp); \
4983 else if (__builtin_types_compatible_p (typeof (x), float)) \
4984 tmp = foo_float (tmp); \
4991 @emph{Note:} This construct is only available for C.
4995 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
4997 You can use the built-in function @code{__builtin_choose_expr} to
4998 evaluate code depending on the value of a constant expression. This
4999 built-in function returns @var{exp1} if @var{const_exp}, which is a
5000 constant expression that must be able to be determined at compile time,
5001 is nonzero. Otherwise it returns 0.
5003 This built-in function is analogous to the @samp{? :} operator in C,
5004 except that the expression returned has its type unaltered by promotion
5005 rules. Also, the built-in function does not evaluate the expression
5006 that was not chosen. For example, if @var{const_exp} evaluates to true,
5007 @var{exp2} is not evaluated even if it has side-effects.
5009 This built-in function can return an lvalue if the chosen argument is an
5012 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5013 type. Similarly, if @var{exp2} is returned, its return type is the same
5020 __builtin_choose_expr ( \
5021 __builtin_types_compatible_p (typeof (x), double), \
5023 __builtin_choose_expr ( \
5024 __builtin_types_compatible_p (typeof (x), float), \
5026 /* @r{The void expression results in a compile-time error} \
5027 @r{when assigning the result to something.} */ \
5031 @emph{Note:} This construct is only available for C. Furthermore, the
5032 unused expression (@var{exp1} or @var{exp2} depending on the value of
5033 @var{const_exp}) may still generate syntax errors. This may change in
5038 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5039 You can use the built-in function @code{__builtin_constant_p} to
5040 determine if a value is known to be constant at compile-time and hence
5041 that GCC can perform constant-folding on expressions involving that
5042 value. The argument of the function is the value to test. The function
5043 returns the integer 1 if the argument is known to be a compile-time
5044 constant and 0 if it is not known to be a compile-time constant. A
5045 return of 0 does not indicate that the value is @emph{not} a constant,
5046 but merely that GCC cannot prove it is a constant with the specified
5047 value of the @option{-O} option.
5049 You would typically use this function in an embedded application where
5050 memory was a critical resource. If you have some complex calculation,
5051 you may want it to be folded if it involves constants, but need to call
5052 a function if it does not. For example:
5055 #define Scale_Value(X) \
5056 (__builtin_constant_p (X) \
5057 ? ((X) * SCALE + OFFSET) : Scale (X))
5060 You may use this built-in function in either a macro or an inline
5061 function. However, if you use it in an inlined function and pass an
5062 argument of the function as the argument to the built-in, GCC will
5063 never return 1 when you call the inline function with a string constant
5064 or compound literal (@pxref{Compound Literals}) and will not return 1
5065 when you pass a constant numeric value to the inline function unless you
5066 specify the @option{-O} option.
5068 You may also use @code{__builtin_constant_p} in initializers for static
5069 data. For instance, you can write
5072 static const int table[] = @{
5073 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5079 This is an acceptable initializer even if @var{EXPRESSION} is not a
5080 constant expression. GCC must be more conservative about evaluating the
5081 built-in in this case, because it has no opportunity to perform
5084 Previous versions of GCC did not accept this built-in in data
5085 initializers. The earliest version where it is completely safe is
5089 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5090 @opindex fprofile-arcs
5091 You may use @code{__builtin_expect} to provide the compiler with
5092 branch prediction information. In general, you should prefer to
5093 use actual profile feedback for this (@option{-fprofile-arcs}), as
5094 programmers are notoriously bad at predicting how their programs
5095 actually perform. However, there are applications in which this
5096 data is hard to collect.
5098 The return value is the value of @var{exp}, which should be an
5099 integral expression. The value of @var{c} must be a compile-time
5100 constant. The semantics of the built-in are that it is expected
5101 that @var{exp} == @var{c}. For example:
5104 if (__builtin_expect (x, 0))
5109 would indicate that we do not expect to call @code{foo}, since
5110 we expect @code{x} to be zero. Since you are limited to integral
5111 expressions for @var{exp}, you should use constructions such as
5114 if (__builtin_expect (ptr != NULL, 1))
5119 when testing pointer or floating-point values.
5122 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5123 This function is used to minimize cache-miss latency by moving data into
5124 a cache before it is accessed.
5125 You can insert calls to @code{__builtin_prefetch} into code for which
5126 you know addresses of data in memory that is likely to be accessed soon.
5127 If the target supports them, data prefetch instructions will be generated.
5128 If the prefetch is done early enough before the access then the data will
5129 be in the cache by the time it is accessed.
5131 The value of @var{addr} is the address of the memory to prefetch.
5132 There are two optional arguments, @var{rw} and @var{locality}.
5133 The value of @var{rw} is a compile-time constant one or zero; one
5134 means that the prefetch is preparing for a write to the memory address
5135 and zero, the default, means that the prefetch is preparing for a read.
5136 The value @var{locality} must be a compile-time constant integer between
5137 zero and three. A value of zero means that the data has no temporal
5138 locality, so it need not be left in the cache after the access. A value
5139 of three means that the data has a high degree of temporal locality and
5140 should be left in all levels of cache possible. Values of one and two
5141 mean, respectively, a low or moderate degree of temporal locality. The
5145 for (i = 0; i < n; i++)
5148 __builtin_prefetch (&a[i+j], 1, 1);
5149 __builtin_prefetch (&b[i+j], 0, 1);
5154 Data prefetch does not generate faults if @var{addr} is invalid, but
5155 the address expression itself must be valid. For example, a prefetch
5156 of @code{p->next} will not fault if @code{p->next} is not a valid
5157 address, but evaluation will fault if @code{p} is not a valid address.
5159 If the target does not support data prefetch, the address expression
5160 is evaluated if it includes side effects but no other code is generated
5161 and GCC does not issue a warning.
5164 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5165 Returns a positive infinity, if supported by the floating-point format,
5166 else @code{DBL_MAX}. This function is suitable for implementing the
5167 ISO C macro @code{HUGE_VAL}.
5170 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5171 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5174 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5175 Similar to @code{__builtin_huge_val}, except the return
5176 type is @code{long double}.
5179 @deftypefn {Built-in Function} double __builtin_inf (void)
5180 Similar to @code{__builtin_huge_val}, except a warning is generated
5181 if the target floating-point format does not support infinities.
5182 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5185 @deftypefn {Built-in Function} float __builtin_inff (void)
5186 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5189 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5190 Similar to @code{__builtin_inf}, except the return
5191 type is @code{long double}.
5194 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5195 This is an implementation of the ISO C99 function @code{nan}.
5197 Since ISO C99 defines this function in terms of @code{strtod}, which we
5198 do not implement, a description of the parsing is in order. The string
5199 is parsed as by @code{strtol}; that is, the base is recognized by
5200 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5201 in the significand such that the least significant bit of the number
5202 is at the least significant bit of the significand. The number is
5203 truncated to fit the significand field provided. The significand is
5204 forced to be a quiet NaN.
5206 This function, if given a string literal, is evaluated early enough
5207 that it is considered a compile-time constant.
5210 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
5211 Similar to @code{__builtin_nan}, except the return type is @code{float}.
5214 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
5215 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
5218 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
5219 Similar to @code{__builtin_nan}, except the significand is forced
5220 to be a signaling NaN. The @code{nans} function is proposed by
5221 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
5224 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
5225 Similar to @code{__builtin_nans}, except the return type is @code{float}.
5228 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
5229 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
5232 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
5233 Returns one plus the index of the least significant 1-bit of @var{x}, or
5234 if @var{x} is zero, returns zero.
5237 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
5238 Returns the number of leading 0-bits in @var{x}, starting at the most
5239 significant bit position. If @var{x} is 0, the result is undefined.
5242 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
5243 Returns the number of trailing 0-bits in @var{x}, starting at the least
5244 significant bit position. If @var{x} is 0, the result is undefined.
5247 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
5248 Returns the number of 1-bits in @var{x}.
5251 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
5252 Returns the parity of @var{x}, i.@:e. the number of 1-bits in @var{x}
5256 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
5257 Similar to @code{__builtin_ffs}, except the argument type is
5258 @code{unsigned long}.
5261 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
5262 Similar to @code{__builtin_clz}, except the argument type is
5263 @code{unsigned long}.
5266 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
5267 Similar to @code{__builtin_ctz}, except the argument type is
5268 @code{unsigned long}.
5271 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
5272 Similar to @code{__builtin_popcount}, except the argument type is
5273 @code{unsigned long}.
5276 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
5277 Similar to @code{__builtin_parity}, except the argument type is
5278 @code{unsigned long}.
5281 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
5282 Similar to @code{__builtin_ffs}, except the argument type is
5283 @code{unsigned long long}.
5286 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
5287 Similar to @code{__builtin_clz}, except the argument type is
5288 @code{unsigned long long}.
5291 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
5292 Similar to @code{__builtin_ctz}, except the argument type is
5293 @code{unsigned long long}.
5296 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
5297 Similar to @code{__builtin_popcount}, except the argument type is
5298 @code{unsigned long long}.
5301 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
5302 Similar to @code{__builtin_parity}, except the argument type is
5303 @code{unsigned long long}.
5307 @node Target Builtins
5308 @section Built-in Functions Specific to Particular Target Machines
5310 On some target machines, GCC supports many built-in functions specific
5311 to those machines. Generally these generate calls to specific machine
5312 instructions, but allow the compiler to schedule those calls.
5315 * Alpha Built-in Functions::
5316 * ARM Built-in Functions::
5317 * FR-V Built-in Functions::
5318 * X86 Built-in Functions::
5319 * MIPS Paired-Single Support::
5320 * PowerPC AltiVec Built-in Functions::
5323 @node Alpha Built-in Functions
5324 @subsection Alpha Built-in Functions
5326 These built-in functions are available for the Alpha family of
5327 processors, depending on the command-line switches used.
5329 The following built-in functions are always available. They
5330 all generate the machine instruction that is part of the name.
5333 long __builtin_alpha_implver (void)
5334 long __builtin_alpha_rpcc (void)
5335 long __builtin_alpha_amask (long)
5336 long __builtin_alpha_cmpbge (long, long)
5337 long __builtin_alpha_extbl (long, long)
5338 long __builtin_alpha_extwl (long, long)
5339 long __builtin_alpha_extll (long, long)
5340 long __builtin_alpha_extql (long, long)
5341 long __builtin_alpha_extwh (long, long)
5342 long __builtin_alpha_extlh (long, long)
5343 long __builtin_alpha_extqh (long, long)
5344 long __builtin_alpha_insbl (long, long)
5345 long __builtin_alpha_inswl (long, long)
5346 long __builtin_alpha_insll (long, long)
5347 long __builtin_alpha_insql (long, long)
5348 long __builtin_alpha_inswh (long, long)
5349 long __builtin_alpha_inslh (long, long)
5350 long __builtin_alpha_insqh (long, long)
5351 long __builtin_alpha_mskbl (long, long)
5352 long __builtin_alpha_mskwl (long, long)
5353 long __builtin_alpha_mskll (long, long)
5354 long __builtin_alpha_mskql (long, long)
5355 long __builtin_alpha_mskwh (long, long)
5356 long __builtin_alpha_msklh (long, long)
5357 long __builtin_alpha_mskqh (long, long)
5358 long __builtin_alpha_umulh (long, long)
5359 long __builtin_alpha_zap (long, long)
5360 long __builtin_alpha_zapnot (long, long)
5363 The following built-in functions are always with @option{-mmax}
5364 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
5365 later. They all generate the machine instruction that is part
5369 long __builtin_alpha_pklb (long)
5370 long __builtin_alpha_pkwb (long)
5371 long __builtin_alpha_unpkbl (long)
5372 long __builtin_alpha_unpkbw (long)
5373 long __builtin_alpha_minub8 (long, long)
5374 long __builtin_alpha_minsb8 (long, long)
5375 long __builtin_alpha_minuw4 (long, long)
5376 long __builtin_alpha_minsw4 (long, long)
5377 long __builtin_alpha_maxub8 (long, long)
5378 long __builtin_alpha_maxsb8 (long, long)
5379 long __builtin_alpha_maxuw4 (long, long)
5380 long __builtin_alpha_maxsw4 (long, long)
5381 long __builtin_alpha_perr (long, long)
5384 The following built-in functions are always with @option{-mcix}
5385 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
5386 later. They all generate the machine instruction that is part
5390 long __builtin_alpha_cttz (long)
5391 long __builtin_alpha_ctlz (long)
5392 long __builtin_alpha_ctpop (long)
5395 The following builtins are available on systems that use the OSF/1
5396 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
5397 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
5398 @code{rdval} and @code{wrval}.
5401 void *__builtin_thread_pointer (void)
5402 void __builtin_set_thread_pointer (void *)
5405 @node ARM Built-in Functions
5406 @subsection ARM Built-in Functions
5408 These built-in functions are available for the ARM family of
5409 processors, when the @option{-mcpu=iwmmxt} switch is used:
5412 typedef int v2si __attribute__ ((vector_size (8)));
5413 typedef short v4hi __attribute__ ((vector_size (8)));
5414 typedef char v8qi __attribute__ ((vector_size (8)));
5416 int __builtin_arm_getwcx (int)
5417 void __builtin_arm_setwcx (int, int)
5418 int __builtin_arm_textrmsb (v8qi, int)
5419 int __builtin_arm_textrmsh (v4hi, int)
5420 int __builtin_arm_textrmsw (v2si, int)
5421 int __builtin_arm_textrmub (v8qi, int)
5422 int __builtin_arm_textrmuh (v4hi, int)
5423 int __builtin_arm_textrmuw (v2si, int)
5424 v8qi __builtin_arm_tinsrb (v8qi, int)
5425 v4hi __builtin_arm_tinsrh (v4hi, int)
5426 v2si __builtin_arm_tinsrw (v2si, int)
5427 long long __builtin_arm_tmia (long long, int, int)
5428 long long __builtin_arm_tmiabb (long long, int, int)
5429 long long __builtin_arm_tmiabt (long long, int, int)
5430 long long __builtin_arm_tmiaph (long long, int, int)
5431 long long __builtin_arm_tmiatb (long long, int, int)
5432 long long __builtin_arm_tmiatt (long long, int, int)
5433 int __builtin_arm_tmovmskb (v8qi)
5434 int __builtin_arm_tmovmskh (v4hi)
5435 int __builtin_arm_tmovmskw (v2si)
5436 long long __builtin_arm_waccb (v8qi)
5437 long long __builtin_arm_wacch (v4hi)
5438 long long __builtin_arm_waccw (v2si)
5439 v8qi __builtin_arm_waddb (v8qi, v8qi)
5440 v8qi __builtin_arm_waddbss (v8qi, v8qi)
5441 v8qi __builtin_arm_waddbus (v8qi, v8qi)
5442 v4hi __builtin_arm_waddh (v4hi, v4hi)
5443 v4hi __builtin_arm_waddhss (v4hi, v4hi)
5444 v4hi __builtin_arm_waddhus (v4hi, v4hi)
5445 v2si __builtin_arm_waddw (v2si, v2si)
5446 v2si __builtin_arm_waddwss (v2si, v2si)
5447 v2si __builtin_arm_waddwus (v2si, v2si)
5448 v8qi __builtin_arm_walign (v8qi, v8qi, int)
5449 long long __builtin_arm_wand(long long, long long)
5450 long long __builtin_arm_wandn (long long, long long)
5451 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
5452 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
5453 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
5454 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
5455 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
5456 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
5457 v2si __builtin_arm_wcmpeqw (v2si, v2si)
5458 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
5459 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
5460 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
5461 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
5462 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
5463 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
5464 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
5465 long long __builtin_arm_wmacsz (v4hi, v4hi)
5466 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
5467 long long __builtin_arm_wmacuz (v4hi, v4hi)
5468 v4hi __builtin_arm_wmadds (v4hi, v4hi)
5469 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
5470 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
5471 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
5472 v2si __builtin_arm_wmaxsw (v2si, v2si)
5473 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
5474 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
5475 v2si __builtin_arm_wmaxuw (v2si, v2si)
5476 v8qi __builtin_arm_wminsb (v8qi, v8qi)
5477 v4hi __builtin_arm_wminsh (v4hi, v4hi)
5478 v2si __builtin_arm_wminsw (v2si, v2si)
5479 v8qi __builtin_arm_wminub (v8qi, v8qi)
5480 v4hi __builtin_arm_wminuh (v4hi, v4hi)
5481 v2si __builtin_arm_wminuw (v2si, v2si)
5482 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
5483 v4hi __builtin_arm_wmulul (v4hi, v4hi)
5484 v4hi __builtin_arm_wmulum (v4hi, v4hi)
5485 long long __builtin_arm_wor (long long, long long)
5486 v2si __builtin_arm_wpackdss (long long, long long)
5487 v2si __builtin_arm_wpackdus (long long, long long)
5488 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
5489 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
5490 v4hi __builtin_arm_wpackwss (v2si, v2si)
5491 v4hi __builtin_arm_wpackwus (v2si, v2si)
5492 long long __builtin_arm_wrord (long long, long long)
5493 long long __builtin_arm_wrordi (long long, int)
5494 v4hi __builtin_arm_wrorh (v4hi, long long)
5495 v4hi __builtin_arm_wrorhi (v4hi, int)
5496 v2si __builtin_arm_wrorw (v2si, long long)
5497 v2si __builtin_arm_wrorwi (v2si, int)
5498 v2si __builtin_arm_wsadb (v8qi, v8qi)
5499 v2si __builtin_arm_wsadbz (v8qi, v8qi)
5500 v2si __builtin_arm_wsadh (v4hi, v4hi)
5501 v2si __builtin_arm_wsadhz (v4hi, v4hi)
5502 v4hi __builtin_arm_wshufh (v4hi, int)
5503 long long __builtin_arm_wslld (long long, long long)
5504 long long __builtin_arm_wslldi (long long, int)
5505 v4hi __builtin_arm_wsllh (v4hi, long long)
5506 v4hi __builtin_arm_wsllhi (v4hi, int)
5507 v2si __builtin_arm_wsllw (v2si, long long)
5508 v2si __builtin_arm_wsllwi (v2si, int)
5509 long long __builtin_arm_wsrad (long long, long long)
5510 long long __builtin_arm_wsradi (long long, int)
5511 v4hi __builtin_arm_wsrah (v4hi, long long)
5512 v4hi __builtin_arm_wsrahi (v4hi, int)
5513 v2si __builtin_arm_wsraw (v2si, long long)
5514 v2si __builtin_arm_wsrawi (v2si, int)
5515 long long __builtin_arm_wsrld (long long, long long)
5516 long long __builtin_arm_wsrldi (long long, int)
5517 v4hi __builtin_arm_wsrlh (v4hi, long long)
5518 v4hi __builtin_arm_wsrlhi (v4hi, int)
5519 v2si __builtin_arm_wsrlw (v2si, long long)
5520 v2si __builtin_arm_wsrlwi (v2si, int)
5521 v8qi __builtin_arm_wsubb (v8qi, v8qi)
5522 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
5523 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
5524 v4hi __builtin_arm_wsubh (v4hi, v4hi)
5525 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
5526 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
5527 v2si __builtin_arm_wsubw (v2si, v2si)
5528 v2si __builtin_arm_wsubwss (v2si, v2si)
5529 v2si __builtin_arm_wsubwus (v2si, v2si)
5530 v4hi __builtin_arm_wunpckehsb (v8qi)
5531 v2si __builtin_arm_wunpckehsh (v4hi)
5532 long long __builtin_arm_wunpckehsw (v2si)
5533 v4hi __builtin_arm_wunpckehub (v8qi)
5534 v2si __builtin_arm_wunpckehuh (v4hi)
5535 long long __builtin_arm_wunpckehuw (v2si)
5536 v4hi __builtin_arm_wunpckelsb (v8qi)
5537 v2si __builtin_arm_wunpckelsh (v4hi)
5538 long long __builtin_arm_wunpckelsw (v2si)
5539 v4hi __builtin_arm_wunpckelub (v8qi)
5540 v2si __builtin_arm_wunpckeluh (v4hi)
5541 long long __builtin_arm_wunpckeluw (v2si)
5542 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
5543 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
5544 v2si __builtin_arm_wunpckihw (v2si, v2si)
5545 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
5546 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
5547 v2si __builtin_arm_wunpckilw (v2si, v2si)
5548 long long __builtin_arm_wxor (long long, long long)
5549 long long __builtin_arm_wzero ()
5552 @node FR-V Built-in Functions
5553 @subsection FR-V Built-in Functions
5555 GCC provides many FR-V-specific built-in functions. In general,
5556 these functions are intended to be compatible with those described
5557 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
5558 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
5559 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
5560 pointer rather than by value.
5562 Most of the functions are named after specific FR-V instructions.
5563 Such functions are said to be ``directly mapped'' and are summarized
5564 here in tabular form.
5568 * Directly-mapped Integer Functions::
5569 * Directly-mapped Media Functions::
5570 * Other Built-in Functions::
5573 @node Argument Types
5574 @subsubsection Argument Types
5576 The arguments to the built-in functions can be divided into three groups:
5577 register numbers, compile-time constants and run-time values. In order
5578 to make this classification clear at a glance, the arguments and return
5579 values are given the following pseudo types:
5581 @multitable @columnfractions .20 .30 .15 .35
5582 @item Pseudo type @tab Real C type @tab Constant? @tab Description
5583 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
5584 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
5585 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
5586 @item @code{uw2} @tab @code{unsigned long long} @tab No
5587 @tab an unsigned doubleword
5588 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
5589 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
5590 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
5591 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
5594 These pseudo types are not defined by GCC, they are simply a notational
5595 convenience used in this manual.
5597 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
5598 and @code{sw2} are evaluated at run time. They correspond to
5599 register operands in the underlying FR-V instructions.
5601 @code{const} arguments represent immediate operands in the underlying
5602 FR-V instructions. They must be compile-time constants.
5604 @code{acc} arguments are evaluated at compile time and specify the number
5605 of an accumulator register. For example, an @code{acc} argument of 2
5606 will select the ACC2 register.
5608 @code{iacc} arguments are similar to @code{acc} arguments but specify the
5609 number of an IACC register. See @pxref{Other Built-in Functions}
5612 @node Directly-mapped Integer Functions
5613 @subsubsection Directly-mapped Integer Functions
5615 The functions listed below map directly to FR-V I-type instructions.
5617 @multitable @columnfractions .45 .32 .23
5618 @item Function prototype @tab Example usage @tab Assembly output
5619 @item @code{sw1 __ADDSS (sw1, sw1)}
5620 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
5621 @tab @code{ADDSS @var{a},@var{b},@var{c}}
5622 @item @code{sw1 __SCAN (sw1, sw1)}
5623 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
5624 @tab @code{SCAN @var{a},@var{b},@var{c}}
5625 @item @code{sw1 __SCUTSS (sw1)}
5626 @tab @code{@var{b} = __SCUTSS (@var{a})}
5627 @tab @code{SCUTSS @var{a},@var{b}}
5628 @item @code{sw1 __SLASS (sw1, sw1)}
5629 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
5630 @tab @code{SLASS @var{a},@var{b},@var{c}}
5631 @item @code{void __SMASS (sw1, sw1)}
5632 @tab @code{__SMASS (@var{a}, @var{b})}
5633 @tab @code{SMASS @var{a},@var{b}}
5634 @item @code{void __SMSSS (sw1, sw1)}
5635 @tab @code{__SMSSS (@var{a}, @var{b})}
5636 @tab @code{SMSSS @var{a},@var{b}}
5637 @item @code{void __SMU (sw1, sw1)}
5638 @tab @code{__SMU (@var{a}, @var{b})}
5639 @tab @code{SMU @var{a},@var{b}}
5640 @item @code{sw2 __SMUL (sw1, sw1)}
5641 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
5642 @tab @code{SMUL @var{a},@var{b},@var{c}}
5643 @item @code{sw1 __SUBSS (sw1, sw1)}
5644 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
5645 @tab @code{SUBSS @var{a},@var{b},@var{c}}
5646 @item @code{uw2 __UMUL (uw1, uw1)}
5647 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
5648 @tab @code{UMUL @var{a},@var{b},@var{c}}
5651 @node Directly-mapped Media Functions
5652 @subsubsection Directly-mapped Media Functions
5654 The functions listed below map directly to FR-V M-type instructions.
5656 @multitable @columnfractions .45 .32 .23
5657 @item Function prototype @tab Example usage @tab Assembly output
5658 @item @code{uw1 __MABSHS (sw1)}
5659 @tab @code{@var{b} = __MABSHS (@var{a})}
5660 @tab @code{MABSHS @var{a},@var{b}}
5661 @item @code{void __MADDACCS (acc, acc)}
5662 @tab @code{__MADDACCS (@var{b}, @var{a})}
5663 @tab @code{MADDACCS @var{a},@var{b}}
5664 @item @code{sw1 __MADDHSS (sw1, sw1)}
5665 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
5666 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
5667 @item @code{uw1 __MADDHUS (uw1, uw1)}
5668 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
5669 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
5670 @item @code{uw1 __MAND (uw1, uw1)}
5671 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
5672 @tab @code{MAND @var{a},@var{b},@var{c}}
5673 @item @code{void __MASACCS (acc, acc)}
5674 @tab @code{__MASACCS (@var{b}, @var{a})}
5675 @tab @code{MASACCS @var{a},@var{b}}
5676 @item @code{uw1 __MAVEH (uw1, uw1)}
5677 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
5678 @tab @code{MAVEH @var{a},@var{b},@var{c}}
5679 @item @code{uw2 __MBTOH (uw1)}
5680 @tab @code{@var{b} = __MBTOH (@var{a})}
5681 @tab @code{MBTOH @var{a},@var{b}}
5682 @item @code{void __MBTOHE (uw1 *, uw1)}
5683 @tab @code{__MBTOHE (&@var{b}, @var{a})}
5684 @tab @code{MBTOHE @var{a},@var{b}}
5685 @item @code{void __MCLRACC (acc)}
5686 @tab @code{__MCLRACC (@var{a})}
5687 @tab @code{MCLRACC @var{a}}
5688 @item @code{void __MCLRACCA (void)}
5689 @tab @code{__MCLRACCA ()}
5690 @tab @code{MCLRACCA}
5691 @item @code{uw1 __Mcop1 (uw1, uw1)}
5692 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
5693 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
5694 @item @code{uw1 __Mcop2 (uw1, uw1)}
5695 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
5696 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
5697 @item @code{uw1 __MCPLHI (uw2, const)}
5698 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
5699 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
5700 @item @code{uw1 __MCPLI (uw2, const)}
5701 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
5702 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
5703 @item @code{void __MCPXIS (acc, sw1, sw1)}
5704 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
5705 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
5706 @item @code{void __MCPXIU (acc, uw1, uw1)}
5707 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
5708 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
5709 @item @code{void __MCPXRS (acc, sw1, sw1)}
5710 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
5711 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
5712 @item @code{void __MCPXRU (acc, uw1, uw1)}
5713 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
5714 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
5715 @item @code{uw1 __MCUT (acc, uw1)}
5716 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
5717 @tab @code{MCUT @var{a},@var{b},@var{c}}
5718 @item @code{uw1 __MCUTSS (acc, sw1)}
5719 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
5720 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
5721 @item @code{void __MDADDACCS (acc, acc)}
5722 @tab @code{__MDADDACCS (@var{b}, @var{a})}
5723 @tab @code{MDADDACCS @var{a},@var{b}}
5724 @item @code{void __MDASACCS (acc, acc)}
5725 @tab @code{__MDASACCS (@var{b}, @var{a})}
5726 @tab @code{MDASACCS @var{a},@var{b}}
5727 @item @code{uw2 __MDCUTSSI (acc, const)}
5728 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
5729 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
5730 @item @code{uw2 __MDPACKH (uw2, uw2)}
5731 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
5732 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
5733 @item @code{uw2 __MDROTLI (uw2, const)}
5734 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
5735 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
5736 @item @code{void __MDSUBACCS (acc, acc)}
5737 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
5738 @tab @code{MDSUBACCS @var{a},@var{b}}
5739 @item @code{void __MDUNPACKH (uw1 *, uw2)}
5740 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
5741 @tab @code{MDUNPACKH @var{a},@var{b}}
5742 @item @code{uw2 __MEXPDHD (uw1, const)}
5743 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
5744 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
5745 @item @code{uw1 __MEXPDHW (uw1, const)}
5746 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
5747 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
5748 @item @code{uw1 __MHDSETH (uw1, const)}
5749 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
5750 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
5751 @item @code{sw1 __MHDSETS (const)}
5752 @tab @code{@var{b} = __MHDSETS (@var{a})}
5753 @tab @code{MHDSETS #@var{a},@var{b}}
5754 @item @code{uw1 __MHSETHIH (uw1, const)}
5755 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
5756 @tab @code{MHSETHIH #@var{a},@var{b}}
5757 @item @code{sw1 __MHSETHIS (sw1, const)}
5758 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
5759 @tab @code{MHSETHIS #@var{a},@var{b}}
5760 @item @code{uw1 __MHSETLOH (uw1, const)}
5761 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
5762 @tab @code{MHSETLOH #@var{a},@var{b}}
5763 @item @code{sw1 __MHSETLOS (sw1, const)}
5764 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
5765 @tab @code{MHSETLOS #@var{a},@var{b}}
5766 @item @code{uw1 __MHTOB (uw2)}
5767 @tab @code{@var{b} = __MHTOB (@var{a})}
5768 @tab @code{MHTOB @var{a},@var{b}}
5769 @item @code{void __MMACHS (acc, sw1, sw1)}
5770 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
5771 @tab @code{MMACHS @var{a},@var{b},@var{c}}
5772 @item @code{void __MMACHU (acc, uw1, uw1)}
5773 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
5774 @tab @code{MMACHU @var{a},@var{b},@var{c}}
5775 @item @code{void __MMRDHS (acc, sw1, sw1)}
5776 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
5777 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
5778 @item @code{void __MMRDHU (acc, uw1, uw1)}
5779 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
5780 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
5781 @item @code{void __MMULHS (acc, sw1, sw1)}
5782 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
5783 @tab @code{MMULHS @var{a},@var{b},@var{c}}
5784 @item @code{void __MMULHU (acc, uw1, uw1)}
5785 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
5786 @tab @code{MMULHU @var{a},@var{b},@var{c}}
5787 @item @code{void __MMULXHS (acc, sw1, sw1)}
5788 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
5789 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
5790 @item @code{void __MMULXHU (acc, uw1, uw1)}
5791 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
5792 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
5793 @item @code{uw1 __MNOT (uw1)}
5794 @tab @code{@var{b} = __MNOT (@var{a})}
5795 @tab @code{MNOT @var{a},@var{b}}
5796 @item @code{uw1 __MOR (uw1, uw1)}
5797 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
5798 @tab @code{MOR @var{a},@var{b},@var{c}}
5799 @item @code{uw1 __MPACKH (uh, uh)}
5800 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
5801 @tab @code{MPACKH @var{a},@var{b},@var{c}}
5802 @item @code{sw2 __MQADDHSS (sw2, sw2)}
5803 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
5804 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
5805 @item @code{uw2 __MQADDHUS (uw2, uw2)}
5806 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
5807 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
5808 @item @code{void __MQCPXIS (acc, sw2, sw2)}
5809 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
5810 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
5811 @item @code{void __MQCPXIU (acc, uw2, uw2)}
5812 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
5813 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
5814 @item @code{void __MQCPXRS (acc, sw2, sw2)}
5815 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
5816 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
5817 @item @code{void __MQCPXRU (acc, uw2, uw2)}
5818 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
5819 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
5820 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
5821 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
5822 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
5823 @item @code{sw2 __MQLMTHS (sw2, sw2)}
5824 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
5825 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
5826 @item @code{void __MQMACHS (acc, sw2, sw2)}
5827 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
5828 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
5829 @item @code{void __MQMACHU (acc, uw2, uw2)}
5830 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
5831 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
5832 @item @code{void __MQMACXHS (acc, sw2, sw2)}
5833 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
5834 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
5835 @item @code{void __MQMULHS (acc, sw2, sw2)}
5836 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
5837 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
5838 @item @code{void __MQMULHU (acc, uw2, uw2)}
5839 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
5840 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
5841 @item @code{void __MQMULXHS (acc, sw2, sw2)}
5842 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
5843 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
5844 @item @code{void __MQMULXHU (acc, uw2, uw2)}
5845 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
5846 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
5847 @item @code{sw2 __MQSATHS (sw2, sw2)}
5848 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
5849 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
5850 @item @code{uw2 __MQSLLHI (uw2, int)}
5851 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
5852 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
5853 @item @code{sw2 __MQSRAHI (sw2, int)}
5854 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
5855 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
5856 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
5857 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
5858 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
5859 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
5860 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
5861 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
5862 @item @code{void __MQXMACHS (acc, sw2, sw2)}
5863 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
5864 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
5865 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
5866 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
5867 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
5868 @item @code{uw1 __MRDACC (acc)}
5869 @tab @code{@var{b} = __MRDACC (@var{a})}
5870 @tab @code{MRDACC @var{a},@var{b}}
5871 @item @code{uw1 __MRDACCG (acc)}
5872 @tab @code{@var{b} = __MRDACCG (@var{a})}
5873 @tab @code{MRDACCG @var{a},@var{b}}
5874 @item @code{uw1 __MROTLI (uw1, const)}
5875 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
5876 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
5877 @item @code{uw1 __MROTRI (uw1, const)}
5878 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
5879 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
5880 @item @code{sw1 __MSATHS (sw1, sw1)}
5881 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
5882 @tab @code{MSATHS @var{a},@var{b},@var{c}}
5883 @item @code{uw1 __MSATHU (uw1, uw1)}
5884 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
5885 @tab @code{MSATHU @var{a},@var{b},@var{c}}
5886 @item @code{uw1 __MSLLHI (uw1, const)}
5887 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
5888 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
5889 @item @code{sw1 __MSRAHI (sw1, const)}
5890 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
5891 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
5892 @item @code{uw1 __MSRLHI (uw1, const)}
5893 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
5894 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
5895 @item @code{void __MSUBACCS (acc, acc)}
5896 @tab @code{__MSUBACCS (@var{b}, @var{a})}
5897 @tab @code{MSUBACCS @var{a},@var{b}}
5898 @item @code{sw1 __MSUBHSS (sw1, sw1)}
5899 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
5900 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
5901 @item @code{uw1 __MSUBHUS (uw1, uw1)}
5902 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
5903 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
5904 @item @code{void __MTRAP (void)}
5905 @tab @code{__MTRAP ()}
5907 @item @code{uw2 __MUNPACKH (uw1)}
5908 @tab @code{@var{b} = __MUNPACKH (@var{a})}
5909 @tab @code{MUNPACKH @var{a},@var{b}}
5910 @item @code{uw1 __MWCUT (uw2, uw1)}
5911 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
5912 @tab @code{MWCUT @var{a},@var{b},@var{c}}
5913 @item @code{void __MWTACC (acc, uw1)}
5914 @tab @code{__MWTACC (@var{b}, @var{a})}
5915 @tab @code{MWTACC @var{a},@var{b}}
5916 @item @code{void __MWTACCG (acc, uw1)}
5917 @tab @code{__MWTACCG (@var{b}, @var{a})}
5918 @tab @code{MWTACCG @var{a},@var{b}}
5919 @item @code{uw1 __MXOR (uw1, uw1)}
5920 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
5921 @tab @code{MXOR @var{a},@var{b},@var{c}}
5924 @node Other Built-in Functions
5925 @subsubsection Other Built-in Functions
5927 This section describes built-in functions that are not named after
5928 a specific FR-V instruction.
5931 @item sw2 __IACCreadll (iacc @var{reg})
5932 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
5933 for future expansion and must be 0.
5935 @item sw1 __IACCreadl (iacc @var{reg})
5936 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
5937 Other values of @var{reg} are rejected as invalid.
5939 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
5940 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
5941 is reserved for future expansion and must be 0.
5943 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
5944 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
5945 is 1. Other values of @var{reg} are rejected as invalid.
5947 @item void __data_prefetch0 (const void *@var{x})
5948 Use the @code{dcpl} instruction to load the contents of address @var{x}
5949 into the data cache.
5951 @item void __data_prefetch (const void *@var{x})
5952 Use the @code{nldub} instruction to load the contents of address @var{x}
5953 into the data cache. The instruction will be issued in slot I1@.
5956 @node X86 Built-in Functions
5957 @subsection X86 Built-in Functions
5959 These built-in functions are available for the i386 and x86-64 family
5960 of computers, depending on the command-line switches used.
5962 The following machine modes are available for use with MMX built-in functions
5963 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
5964 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
5965 vector of eight 8-bit integers. Some of the built-in functions operate on
5966 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
5968 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
5969 of two 32-bit floating point values.
5971 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
5972 floating point values. Some instructions use a vector of four 32-bit
5973 integers, these use @code{V4SI}. Finally, some instructions operate on an
5974 entire vector register, interpreting it as a 128-bit integer, these use mode
5977 The following built-in functions are made available by @option{-mmmx}.
5978 All of them generate the machine instruction that is part of the name.
5981 v8qi __builtin_ia32_paddb (v8qi, v8qi)
5982 v4hi __builtin_ia32_paddw (v4hi, v4hi)
5983 v2si __builtin_ia32_paddd (v2si, v2si)
5984 v8qi __builtin_ia32_psubb (v8qi, v8qi)
5985 v4hi __builtin_ia32_psubw (v4hi, v4hi)
5986 v2si __builtin_ia32_psubd (v2si, v2si)
5987 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
5988 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
5989 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
5990 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
5991 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
5992 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
5993 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
5994 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
5995 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
5996 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
5997 di __builtin_ia32_pand (di, di)
5998 di __builtin_ia32_pandn (di,di)
5999 di __builtin_ia32_por (di, di)
6000 di __builtin_ia32_pxor (di, di)
6001 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
6002 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
6003 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
6004 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
6005 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
6006 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
6007 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
6008 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
6009 v2si __builtin_ia32_punpckhdq (v2si, v2si)
6010 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
6011 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
6012 v2si __builtin_ia32_punpckldq (v2si, v2si)
6013 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
6014 v4hi __builtin_ia32_packssdw (v2si, v2si)
6015 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
6018 The following built-in functions are made available either with
6019 @option{-msse}, or with a combination of @option{-m3dnow} and
6020 @option{-march=athlon}. All of them generate the machine
6021 instruction that is part of the name.
6024 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
6025 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
6026 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
6027 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
6028 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
6029 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
6030 v8qi __builtin_ia32_pminub (v8qi, v8qi)
6031 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
6032 int __builtin_ia32_pextrw (v4hi, int)
6033 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
6034 int __builtin_ia32_pmovmskb (v8qi)
6035 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
6036 void __builtin_ia32_movntq (di *, di)
6037 void __builtin_ia32_sfence (void)
6040 The following built-in functions are available when @option{-msse} is used.
6041 All of them generate the machine instruction that is part of the name.
6044 int __builtin_ia32_comieq (v4sf, v4sf)
6045 int __builtin_ia32_comineq (v4sf, v4sf)
6046 int __builtin_ia32_comilt (v4sf, v4sf)
6047 int __builtin_ia32_comile (v4sf, v4sf)
6048 int __builtin_ia32_comigt (v4sf, v4sf)
6049 int __builtin_ia32_comige (v4sf, v4sf)
6050 int __builtin_ia32_ucomieq (v4sf, v4sf)
6051 int __builtin_ia32_ucomineq (v4sf, v4sf)
6052 int __builtin_ia32_ucomilt (v4sf, v4sf)
6053 int __builtin_ia32_ucomile (v4sf, v4sf)
6054 int __builtin_ia32_ucomigt (v4sf, v4sf)
6055 int __builtin_ia32_ucomige (v4sf, v4sf)
6056 v4sf __builtin_ia32_addps (v4sf, v4sf)
6057 v4sf __builtin_ia32_subps (v4sf, v4sf)
6058 v4sf __builtin_ia32_mulps (v4sf, v4sf)
6059 v4sf __builtin_ia32_divps (v4sf, v4sf)
6060 v4sf __builtin_ia32_addss (v4sf, v4sf)
6061 v4sf __builtin_ia32_subss (v4sf, v4sf)
6062 v4sf __builtin_ia32_mulss (v4sf, v4sf)
6063 v4sf __builtin_ia32_divss (v4sf, v4sf)
6064 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
6065 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
6066 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
6067 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
6068 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
6069 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
6070 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
6071 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
6072 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
6073 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
6074 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
6075 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
6076 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
6077 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
6078 v4si __builtin_ia32_cmpless (v4sf, v4sf)
6079 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
6080 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
6081 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
6082 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
6083 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
6084 v4sf __builtin_ia32_maxps (v4sf, v4sf)
6085 v4sf __builtin_ia32_maxss (v4sf, v4sf)
6086 v4sf __builtin_ia32_minps (v4sf, v4sf)
6087 v4sf __builtin_ia32_minss (v4sf, v4sf)
6088 v4sf __builtin_ia32_andps (v4sf, v4sf)
6089 v4sf __builtin_ia32_andnps (v4sf, v4sf)
6090 v4sf __builtin_ia32_orps (v4sf, v4sf)
6091 v4sf __builtin_ia32_xorps (v4sf, v4sf)
6092 v4sf __builtin_ia32_movss (v4sf, v4sf)
6093 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
6094 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
6095 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
6096 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
6097 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
6098 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
6099 v2si __builtin_ia32_cvtps2pi (v4sf)
6100 int __builtin_ia32_cvtss2si (v4sf)
6101 v2si __builtin_ia32_cvttps2pi (v4sf)
6102 int __builtin_ia32_cvttss2si (v4sf)
6103 v4sf __builtin_ia32_rcpps (v4sf)
6104 v4sf __builtin_ia32_rsqrtps (v4sf)
6105 v4sf __builtin_ia32_sqrtps (v4sf)
6106 v4sf __builtin_ia32_rcpss (v4sf)
6107 v4sf __builtin_ia32_rsqrtss (v4sf)
6108 v4sf __builtin_ia32_sqrtss (v4sf)
6109 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
6110 void __builtin_ia32_movntps (float *, v4sf)
6111 int __builtin_ia32_movmskps (v4sf)
6114 The following built-in functions are available when @option{-msse} is used.
6117 @item v4sf __builtin_ia32_loadaps (float *)
6118 Generates the @code{movaps} machine instruction as a load from memory.
6119 @item void __builtin_ia32_storeaps (float *, v4sf)
6120 Generates the @code{movaps} machine instruction as a store to memory.
6121 @item v4sf __builtin_ia32_loadups (float *)
6122 Generates the @code{movups} machine instruction as a load from memory.
6123 @item void __builtin_ia32_storeups (float *, v4sf)
6124 Generates the @code{movups} machine instruction as a store to memory.
6125 @item v4sf __builtin_ia32_loadsss (float *)
6126 Generates the @code{movss} machine instruction as a load from memory.
6127 @item void __builtin_ia32_storess (float *, v4sf)
6128 Generates the @code{movss} machine instruction as a store to memory.
6129 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
6130 Generates the @code{movhps} machine instruction as a load from memory.
6131 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
6132 Generates the @code{movlps} machine instruction as a load from memory
6133 @item void __builtin_ia32_storehps (v4sf, v2si *)
6134 Generates the @code{movhps} machine instruction as a store to memory.
6135 @item void __builtin_ia32_storelps (v4sf, v2si *)
6136 Generates the @code{movlps} machine instruction as a store to memory.
6139 The following built-in functions are available when @option{-msse3} is used.
6140 All of them generate the machine instruction that is part of the name.
6143 v2df __builtin_ia32_addsubpd (v2df, v2df)
6144 v2df __builtin_ia32_addsubps (v2df, v2df)
6145 v2df __builtin_ia32_haddpd (v2df, v2df)
6146 v2df __builtin_ia32_haddps (v2df, v2df)
6147 v2df __builtin_ia32_hsubpd (v2df, v2df)
6148 v2df __builtin_ia32_hsubps (v2df, v2df)
6149 v16qi __builtin_ia32_lddqu (char const *)
6150 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
6151 v2df __builtin_ia32_movddup (v2df)
6152 v4sf __builtin_ia32_movshdup (v4sf)
6153 v4sf __builtin_ia32_movsldup (v4sf)
6154 void __builtin_ia32_mwait (unsigned int, unsigned int)
6157 The following built-in functions are available when @option{-msse3} is used.
6160 @item v2df __builtin_ia32_loadddup (double const *)
6161 Generates the @code{movddup} machine instruction as a load from memory.
6164 The following built-in functions are available when @option{-m3dnow} is used.
6165 All of them generate the machine instruction that is part of the name.
6168 void __builtin_ia32_femms (void)
6169 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
6170 v2si __builtin_ia32_pf2id (v2sf)
6171 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
6172 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
6173 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
6174 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
6175 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
6176 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
6177 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
6178 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
6179 v2sf __builtin_ia32_pfrcp (v2sf)
6180 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
6181 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
6182 v2sf __builtin_ia32_pfrsqrt (v2sf)
6183 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
6184 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
6185 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
6186 v2sf __builtin_ia32_pi2fd (v2si)
6187 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
6190 The following built-in functions are available when both @option{-m3dnow}
6191 and @option{-march=athlon} are used. All of them generate the machine
6192 instruction that is part of the name.
6195 v2si __builtin_ia32_pf2iw (v2sf)
6196 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
6197 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
6198 v2sf __builtin_ia32_pi2fw (v2si)
6199 v2sf __builtin_ia32_pswapdsf (v2sf)
6200 v2si __builtin_ia32_pswapdsi (v2si)
6203 @node MIPS Paired-Single Support
6204 @subsection MIPS Paired-Single Support
6206 The MIPS64 architecture includes a number of instructions that
6207 operate on pairs of single-precision floating-point values.
6208 Each pair is packed into a 64-bit floating-point register,
6209 with one element being designated the ``upper half'' and
6210 the other being designated the ``lower half''.
6212 GCC supports paired-single operations using both the generic
6213 vector extensions (@pxref{Vector Extensions}) and a collection of
6214 MIPS-specific built-in functions. Both kinds of support are
6215 enabled by the @option{-mpaired-single} command-line option.
6217 The vector type associated with paired-single values is usually
6218 called @code{v2sf}. It can be defined in C as follows:
6221 typedef float v2sf __attribute__ ((vector_size (8)));
6224 @code{v2sf} values are initialized in the same way as aggregates.
6228 v2sf a = @{1.5, 9.1@};
6231 b = (v2sf) @{e, f@};
6234 @emph{Note:} The CPU's endianness determines which value is stored in
6235 the upper half of a register and which value is stored in the lower half.
6236 On little-endian targets, the first value is the lower one and the second
6237 value is the upper one. The opposite order applies to big-endian targets.
6238 For example, the code above will set the lower half of @code{a} to
6239 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
6242 * Paired-Single Arithmetic::
6243 * Paired-Single Built-in Functions::
6244 * MIPS-3D Built-in Functions::
6247 @node Paired-Single Arithmetic
6248 @subsubsection Paired-Single Arithmetic
6250 The table below lists the @code{v2sf} operations for which hardware
6251 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
6252 values and @code{x} is an integral value.
6254 @multitable @columnfractions .50 .50
6255 @item C code @tab MIPS instruction
6256 @item @code{a + b} @tab @code{add.ps}
6257 @item @code{a - b} @tab @code{sub.ps}
6258 @item @code{-a} @tab @code{neg.ps}
6259 @item @code{a * b} @tab @code{mul.ps}
6260 @item @code{a * b + c} @tab @code{madd.ps}
6261 @item @code{a * b - c} @tab @code{msub.ps}
6262 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
6263 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
6264 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
6267 Note that the multiply-accumulate instructions can be disabled
6268 using the command-line option @code{-mno-fused-madd}.
6270 @node Paired-Single Built-in Functions
6271 @subsubsection Paired-Single Built-in Functions
6273 The following paired-single functions map directly to a particular
6274 MIPS instruction. Please refer to the architecture specification
6275 for details on what each instruction does.
6278 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
6279 Pair lower lower (@code{pll.ps}).
6281 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
6282 Pair upper lower (@code{pul.ps}).
6284 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
6285 Pair lower upper (@code{plu.ps}).
6287 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
6288 Pair upper upper (@code{puu.ps}).
6290 @item v2sf __builtin_mips_cvt_ps_s (float, float)
6291 Convert pair to paired single (@code{cvt.ps.s}).
6293 @item float __builtin_mips_cvt_s_pl (v2sf)
6294 Convert pair lower to single (@code{cvt.s.pl}).
6296 @item float __builtin_mips_cvt_s_pu (v2sf)
6297 Convert pair upper to single (@code{cvt.s.pu}).
6299 @item v2sf __builtin_mips_abs_ps (v2sf)
6300 Absolute value (@code{abs.ps}).
6302 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
6303 Align variable (@code{alnv.ps}).
6305 @emph{Note:} The value of the third parameter must be 0 or 4
6306 modulo 8, otherwise the result will be unpredictable. Please read the
6307 instruction description for details.
6310 The following multi-instruction functions are also available.
6311 In each case, @var{cond} can be any of the 16 floating-point conditions:
6312 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
6313 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
6314 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
6317 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6318 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6319 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
6320 @code{movt.ps}/@code{movf.ps}).
6322 The @code{movt} functions return the value @var{x} computed by:
6325 c.@var{cond}.ps @var{cc},@var{a},@var{b}
6326 mov.ps @var{x},@var{c}
6327 movt.ps @var{x},@var{d},@var{cc}
6330 The @code{movf} functions are similar but use @code{movf.ps} instead
6333 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6334 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6335 Comparison of two paired-single values (@code{c.@var{cond}.ps},
6336 @code{bc1t}/@code{bc1f}).
6338 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
6339 and return either the upper or lower half of the result. For example:
6343 if (__builtin_mips_upper_c_eq_ps (a, b))
6344 upper_halves_are_equal ();
6346 upper_halves_are_unequal ();
6348 if (__builtin_mips_lower_c_eq_ps (a, b))
6349 lower_halves_are_equal ();
6351 lower_halves_are_unequal ();
6355 @node MIPS-3D Built-in Functions
6356 @subsubsection MIPS-3D Built-in Functions
6358 The MIPS-3D Application-Specific Extension (ASE) includes additional
6359 paired-single instructions that are designed to improve the performance
6360 of 3D graphics operations. Support for these instructions is controlled
6361 by the @option{-mips3d} command-line option.
6363 The functions listed below map directly to a particular MIPS-3D
6364 instruction. Please refer to the architecture specification for
6365 more details on what each instruction does.
6368 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
6369 Reduction add (@code{addr.ps}).
6371 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
6372 Reduction multiply (@code{mulr.ps}).
6374 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
6375 Convert paired single to paired word (@code{cvt.pw.ps}).
6377 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
6378 Convert paired word to paired single (@code{cvt.ps.pw}).
6380 @item float __builtin_mips_recip1_s (float)
6381 @itemx double __builtin_mips_recip1_d (double)
6382 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
6383 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
6385 @item float __builtin_mips_recip2_s (float, float)
6386 @itemx double __builtin_mips_recip2_d (double, double)
6387 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
6388 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
6390 @item float __builtin_mips_rsqrt1_s (float)
6391 @itemx double __builtin_mips_rsqrt1_d (double)
6392 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
6393 Reduced precision reciprocal square root (sequence step 1)
6394 (@code{rsqrt1.@var{fmt}}).
6396 @item float __builtin_mips_rsqrt2_s (float, float)
6397 @itemx double __builtin_mips_rsqrt2_d (double, double)
6398 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
6399 Reduced precision reciprocal square root (sequence step 2)
6400 (@code{rsqrt2.@var{fmt}}).
6403 The following multi-instruction functions are also available.
6404 In each case, @var{cond} can be any of the 16 floating-point conditions:
6405 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
6406 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
6407 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
6410 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
6411 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
6412 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
6413 @code{bc1t}/@code{bc1f}).
6415 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
6416 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
6421 if (__builtin_mips_cabs_eq_s (a, b))
6427 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6428 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6429 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
6430 @code{bc1t}/@code{bc1f}).
6432 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
6433 and return either the upper or lower half of the result. For example:
6437 if (__builtin_mips_upper_cabs_eq_ps (a, b))
6438 upper_halves_are_equal ();
6440 upper_halves_are_unequal ();
6442 if (__builtin_mips_lower_cabs_eq_ps (a, b))
6443 lower_halves_are_equal ();
6445 lower_halves_are_unequal ();
6448 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6449 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6450 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
6451 @code{movt.ps}/@code{movf.ps}).
6453 The @code{movt} functions return the value @var{x} computed by:
6456 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
6457 mov.ps @var{x},@var{c}
6458 movt.ps @var{x},@var{d},@var{cc}
6461 The @code{movf} functions are similar but use @code{movf.ps} instead
6464 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6465 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6466 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6467 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
6468 Comparison of two paired-single values
6469 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
6470 @code{bc1any2t}/@code{bc1any2f}).
6472 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
6473 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
6474 result is true and the @code{all} forms return true if both results are true.
6479 if (__builtin_mips_any_c_eq_ps (a, b))
6484 if (__builtin_mips_all_c_eq_ps (a, b))
6490 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6491 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6492 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6493 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
6494 Comparison of four paired-single values
6495 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
6496 @code{bc1any4t}/@code{bc1any4f}).
6498 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
6499 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
6500 The @code{any} forms return true if any of the four results are true
6501 and the @code{all} forms return true if all four results are true.
6506 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
6511 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
6518 @node PowerPC AltiVec Built-in Functions
6519 @subsection PowerPC AltiVec Built-in Functions
6521 GCC provides an interface for the PowerPC family of processors to access
6522 the AltiVec operations described in Motorola's AltiVec Programming
6523 Interface Manual. The interface is made available by including
6524 @code{<altivec.h>} and using @option{-maltivec} and
6525 @option{-mabi=altivec}. The interface supports the following vector
6529 vector unsigned char
6533 vector unsigned short
6544 GCC's implementation of the high-level language interface available from
6545 C and C++ code differs from Motorola's documentation in several ways.
6550 A vector constant is a list of constant expressions within curly braces.
6553 A vector initializer requires no cast if the vector constant is of the
6554 same type as the variable it is initializing.
6557 If @code{signed} or @code{unsigned} is omitted, the vector type defaults
6558 to @code{signed} for @code{vector int} or @code{vector short} and to
6559 @code{unsigned} for @code{vector char}.
6562 Compiling with @option{-maltivec} adds keywords @code{__vector},
6563 @code{__pixel}, and @code{__bool}. Macros @option{vector},
6564 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
6568 GCC allows using a @code{typedef} name as the type specifier for a
6572 For C, overloaded functions are implemented with macros so the following
6576 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
6579 Since @code{vec_add} is a macro, the vector constant in the example
6580 is treated as four separate arguments. Wrap the entire argument in
6581 parentheses for this to work.
6584 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
6585 Internally, GCC uses built-in functions to achieve the functionality in
6586 the aforementioned header file, but they are not supported and are
6587 subject to change without notice.
6589 The following interfaces are supported for the generic and specific
6590 AltiVec operations and the AltiVec predicates. In cases where there
6591 is a direct mapping between generic and specific operations, only the
6592 generic names are shown here, although the specific operations can also
6595 Arguments that are documented as @code{const int} require literal
6596 integral values within the range required for that operation.
6599 vector signed char vec_abs (vector signed char);
6600 vector signed short vec_abs (vector signed short);
6601 vector signed int vec_abs (vector signed int);
6602 vector float vec_abs (vector float);
6604 vector signed char vec_abss (vector signed char);
6605 vector signed short vec_abss (vector signed short);
6606 vector signed int vec_abss (vector signed int);
6608 vector signed char vec_add (vector bool char, vector signed char);
6609 vector signed char vec_add (vector signed char, vector bool char);
6610 vector signed char vec_add (vector signed char, vector signed char);
6611 vector unsigned char vec_add (vector bool char, vector unsigned char);
6612 vector unsigned char vec_add (vector unsigned char, vector bool char);
6613 vector unsigned char vec_add (vector unsigned char,
6614 vector unsigned char);
6615 vector signed short vec_add (vector bool short, vector signed short);
6616 vector signed short vec_add (vector signed short, vector bool short);
6617 vector signed short vec_add (vector signed short, vector signed short);
6618 vector unsigned short vec_add (vector bool short,
6619 vector unsigned short);
6620 vector unsigned short vec_add (vector unsigned short,
6622 vector unsigned short vec_add (vector unsigned short,
6623 vector unsigned short);
6624 vector signed int vec_add (vector bool int, vector signed int);
6625 vector signed int vec_add (vector signed int, vector bool int);
6626 vector signed int vec_add (vector signed int, vector signed int);
6627 vector unsigned int vec_add (vector bool int, vector unsigned int);
6628 vector unsigned int vec_add (vector unsigned int, vector bool int);
6629 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
6630 vector float vec_add (vector float, vector float);
6632 vector float vec_vaddfp (vector float, vector float);
6634 vector signed int vec_vadduwm (vector bool int, vector signed int);
6635 vector signed int vec_vadduwm (vector signed int, vector bool int);
6636 vector signed int vec_vadduwm (vector signed int, vector signed int);
6637 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
6638 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
6639 vector unsigned int vec_vadduwm (vector unsigned int,
6640 vector unsigned int);
6642 vector signed short vec_vadduhm (vector bool short,
6643 vector signed short);
6644 vector signed short vec_vadduhm (vector signed short,
6646 vector signed short vec_vadduhm (vector signed short,
6647 vector signed short);
6648 vector unsigned short vec_vadduhm (vector bool short,
6649 vector unsigned short);
6650 vector unsigned short vec_vadduhm (vector unsigned short,
6652 vector unsigned short vec_vadduhm (vector unsigned short,
6653 vector unsigned short);
6655 vector signed char vec_vaddubm (vector bool char, vector signed char);
6656 vector signed char vec_vaddubm (vector signed char, vector bool char);
6657 vector signed char vec_vaddubm (vector signed char, vector signed char);
6658 vector unsigned char vec_vaddubm (vector bool char,
6659 vector unsigned char);
6660 vector unsigned char vec_vaddubm (vector unsigned char,
6662 vector unsigned char vec_vaddubm (vector unsigned char,
6663 vector unsigned char);
6665 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
6667 vector unsigned char vec_adds (vector bool char, vector unsigned char);
6668 vector unsigned char vec_adds (vector unsigned char, vector bool char);
6669 vector unsigned char vec_adds (vector unsigned char,
6670 vector unsigned char);
6671 vector signed char vec_adds (vector bool char, vector signed char);
6672 vector signed char vec_adds (vector signed char, vector bool char);
6673 vector signed char vec_adds (vector signed char, vector signed char);
6674 vector unsigned short vec_adds (vector bool short,
6675 vector unsigned short);
6676 vector unsigned short vec_adds (vector unsigned short,
6678 vector unsigned short vec_adds (vector unsigned short,
6679 vector unsigned short);
6680 vector signed short vec_adds (vector bool short, vector signed short);
6681 vector signed short vec_adds (vector signed short, vector bool short);
6682 vector signed short vec_adds (vector signed short, vector signed short);
6683 vector unsigned int vec_adds (vector bool int, vector unsigned int);
6684 vector unsigned int vec_adds (vector unsigned int, vector bool int);
6685 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
6686 vector signed int vec_adds (vector bool int, vector signed int);
6687 vector signed int vec_adds (vector signed int, vector bool int);
6688 vector signed int vec_adds (vector signed int, vector signed int);
6690 vector signed int vec_vaddsws (vector bool int, vector signed int);
6691 vector signed int vec_vaddsws (vector signed int, vector bool int);
6692 vector signed int vec_vaddsws (vector signed int, vector signed int);
6694 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
6695 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
6696 vector unsigned int vec_vadduws (vector unsigned int,
6697 vector unsigned int);
6699 vector signed short vec_vaddshs (vector bool short,
6700 vector signed short);
6701 vector signed short vec_vaddshs (vector signed short,
6703 vector signed short vec_vaddshs (vector signed short,
6704 vector signed short);
6706 vector unsigned short vec_vadduhs (vector bool short,
6707 vector unsigned short);
6708 vector unsigned short vec_vadduhs (vector unsigned short,
6710 vector unsigned short vec_vadduhs (vector unsigned short,
6711 vector unsigned short);
6713 vector signed char vec_vaddsbs (vector bool char, vector signed char);
6714 vector signed char vec_vaddsbs (vector signed char, vector bool char);
6715 vector signed char vec_vaddsbs (vector signed char, vector signed char);
6717 vector unsigned char vec_vaddubs (vector bool char,
6718 vector unsigned char);
6719 vector unsigned char vec_vaddubs (vector unsigned char,
6721 vector unsigned char vec_vaddubs (vector unsigned char,
6722 vector unsigned char);
6724 vector float vec_and (vector float, vector float);
6725 vector float vec_and (vector float, vector bool int);
6726 vector float vec_and (vector bool int, vector float);
6727 vector bool int vec_and (vector bool int, vector bool int);
6728 vector signed int vec_and (vector bool int, vector signed int);
6729 vector signed int vec_and (vector signed int, vector bool int);
6730 vector signed int vec_and (vector signed int, vector signed int);
6731 vector unsigned int vec_and (vector bool int, vector unsigned int);
6732 vector unsigned int vec_and (vector unsigned int, vector bool int);
6733 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
6734 vector bool short vec_and (vector bool short, vector bool short);
6735 vector signed short vec_and (vector bool short, vector signed short);
6736 vector signed short vec_and (vector signed short, vector bool short);
6737 vector signed short vec_and (vector signed short, vector signed short);
6738 vector unsigned short vec_and (vector bool short,
6739 vector unsigned short);
6740 vector unsigned short vec_and (vector unsigned short,
6742 vector unsigned short vec_and (vector unsigned short,
6743 vector unsigned short);
6744 vector signed char vec_and (vector bool char, vector signed char);
6745 vector bool char vec_and (vector bool char, vector bool char);
6746 vector signed char vec_and (vector signed char, vector bool char);
6747 vector signed char vec_and (vector signed char, vector signed char);
6748 vector unsigned char vec_and (vector bool char, vector unsigned char);
6749 vector unsigned char vec_and (vector unsigned char, vector bool char);
6750 vector unsigned char vec_and (vector unsigned char,
6751 vector unsigned char);
6753 vector float vec_andc (vector float, vector float);
6754 vector float vec_andc (vector float, vector bool int);
6755 vector float vec_andc (vector bool int, vector float);
6756 vector bool int vec_andc (vector bool int, vector bool int);
6757 vector signed int vec_andc (vector bool int, vector signed int);
6758 vector signed int vec_andc (vector signed int, vector bool int);
6759 vector signed int vec_andc (vector signed int, vector signed int);
6760 vector unsigned int vec_andc (vector bool int, vector unsigned int);
6761 vector unsigned int vec_andc (vector unsigned int, vector bool int);
6762 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
6763 vector bool short vec_andc (vector bool short, vector bool short);
6764 vector signed short vec_andc (vector bool short, vector signed short);
6765 vector signed short vec_andc (vector signed short, vector bool short);
6766 vector signed short vec_andc (vector signed short, vector signed short);
6767 vector unsigned short vec_andc (vector bool short,
6768 vector unsigned short);
6769 vector unsigned short vec_andc (vector unsigned short,
6771 vector unsigned short vec_andc (vector unsigned short,
6772 vector unsigned short);
6773 vector signed char vec_andc (vector bool char, vector signed char);
6774 vector bool char vec_andc (vector bool char, vector bool char);
6775 vector signed char vec_andc (vector signed char, vector bool char);
6776 vector signed char vec_andc (vector signed char, vector signed char);
6777 vector unsigned char vec_andc (vector bool char, vector unsigned char);
6778 vector unsigned char vec_andc (vector unsigned char, vector bool char);
6779 vector unsigned char vec_andc (vector unsigned char,
6780 vector unsigned char);
6782 vector unsigned char vec_avg (vector unsigned char,
6783 vector unsigned char);
6784 vector signed char vec_avg (vector signed char, vector signed char);
6785 vector unsigned short vec_avg (vector unsigned short,
6786 vector unsigned short);
6787 vector signed short vec_avg (vector signed short, vector signed short);
6788 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
6789 vector signed int vec_avg (vector signed int, vector signed int);
6791 vector signed int vec_vavgsw (vector signed int, vector signed int);
6793 vector unsigned int vec_vavguw (vector unsigned int,
6794 vector unsigned int);
6796 vector signed short vec_vavgsh (vector signed short,
6797 vector signed short);
6799 vector unsigned short vec_vavguh (vector unsigned short,
6800 vector unsigned short);
6802 vector signed char vec_vavgsb (vector signed char, vector signed char);
6804 vector unsigned char vec_vavgub (vector unsigned char,
6805 vector unsigned char);
6807 vector float vec_ceil (vector float);
6809 vector signed int vec_cmpb (vector float, vector float);
6811 vector bool char vec_cmpeq (vector signed char, vector signed char);
6812 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
6813 vector bool short vec_cmpeq (vector signed short, vector signed short);
6814 vector bool short vec_cmpeq (vector unsigned short,
6815 vector unsigned short);
6816 vector bool int vec_cmpeq (vector signed int, vector signed int);
6817 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
6818 vector bool int vec_cmpeq (vector float, vector float);
6820 vector bool int vec_vcmpeqfp (vector float, vector float);
6822 vector bool int vec_vcmpequw (vector signed int, vector signed int);
6823 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
6825 vector bool short vec_vcmpequh (vector signed short,
6826 vector signed short);
6827 vector bool short vec_vcmpequh (vector unsigned short,
6828 vector unsigned short);
6830 vector bool char vec_vcmpequb (vector signed char, vector signed char);
6831 vector bool char vec_vcmpequb (vector unsigned char,
6832 vector unsigned char);
6834 vector bool int vec_cmpge (vector float, vector float);
6836 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
6837 vector bool char vec_cmpgt (vector signed char, vector signed char);
6838 vector bool short vec_cmpgt (vector unsigned short,
6839 vector unsigned short);
6840 vector bool short vec_cmpgt (vector signed short, vector signed short);
6841 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
6842 vector bool int vec_cmpgt (vector signed int, vector signed int);
6843 vector bool int vec_cmpgt (vector float, vector float);
6845 vector bool int vec_vcmpgtfp (vector float, vector float);
6847 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
6849 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
6851 vector bool short vec_vcmpgtsh (vector signed short,
6852 vector signed short);
6854 vector bool short vec_vcmpgtuh (vector unsigned short,
6855 vector unsigned short);
6857 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
6859 vector bool char vec_vcmpgtub (vector unsigned char,
6860 vector unsigned char);
6862 vector bool int vec_cmple (vector float, vector float);
6864 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
6865 vector bool char vec_cmplt (vector signed char, vector signed char);
6866 vector bool short vec_cmplt (vector unsigned short,
6867 vector unsigned short);
6868 vector bool short vec_cmplt (vector signed short, vector signed short);
6869 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
6870 vector bool int vec_cmplt (vector signed int, vector signed int);
6871 vector bool int vec_cmplt (vector float, vector float);
6873 vector float vec_ctf (vector unsigned int, const int);
6874 vector float vec_ctf (vector signed int, const int);
6876 vector float vec_vcfsx (vector signed int, const int);
6878 vector float vec_vcfux (vector unsigned int, const int);
6880 vector signed int vec_cts (vector float, const int);
6882 vector unsigned int vec_ctu (vector float, const int);
6884 void vec_dss (const int);
6886 void vec_dssall (void);
6888 void vec_dst (const vector unsigned char *, int, const int);
6889 void vec_dst (const vector signed char *, int, const int);
6890 void vec_dst (const vector bool char *, int, const int);
6891 void vec_dst (const vector unsigned short *, int, const int);
6892 void vec_dst (const vector signed short *, int, const int);
6893 void vec_dst (const vector bool short *, int, const int);
6894 void vec_dst (const vector pixel *, int, const int);
6895 void vec_dst (const vector unsigned int *, int, const int);
6896 void vec_dst (const vector signed int *, int, const int);
6897 void vec_dst (const vector bool int *, int, const int);
6898 void vec_dst (const vector float *, int, const int);
6899 void vec_dst (const unsigned char *, int, const int);
6900 void vec_dst (const signed char *, int, const int);
6901 void vec_dst (const unsigned short *, int, const int);
6902 void vec_dst (const short *, int, const int);
6903 void vec_dst (const unsigned int *, int, const int);
6904 void vec_dst (const int *, int, const int);
6905 void vec_dst (const unsigned long *, int, const int);
6906 void vec_dst (const long *, int, const int);
6907 void vec_dst (const float *, int, const int);
6909 void vec_dstst (const vector unsigned char *, int, const int);
6910 void vec_dstst (const vector signed char *, int, const int);
6911 void vec_dstst (const vector bool char *, int, const int);
6912 void vec_dstst (const vector unsigned short *, int, const int);
6913 void vec_dstst (const vector signed short *, int, const int);
6914 void vec_dstst (const vector bool short *, int, const int);
6915 void vec_dstst (const vector pixel *, int, const int);
6916 void vec_dstst (const vector unsigned int *, int, const int);
6917 void vec_dstst (const vector signed int *, int, const int);
6918 void vec_dstst (const vector bool int *, int, const int);
6919 void vec_dstst (const vector float *, int, const int);
6920 void vec_dstst (const unsigned char *, int, const int);
6921 void vec_dstst (const signed char *, int, const int);
6922 void vec_dstst (const unsigned short *, int, const int);
6923 void vec_dstst (const short *, int, const int);
6924 void vec_dstst (const unsigned int *, int, const int);
6925 void vec_dstst (const int *, int, const int);
6926 void vec_dstst (const unsigned long *, int, const int);
6927 void vec_dstst (const long *, int, const int);
6928 void vec_dstst (const float *, int, const int);
6930 void vec_dststt (const vector unsigned char *, int, const int);
6931 void vec_dststt (const vector signed char *, int, const int);
6932 void vec_dststt (const vector bool char *, int, const int);
6933 void vec_dststt (const vector unsigned short *, int, const int);
6934 void vec_dststt (const vector signed short *, int, const int);
6935 void vec_dststt (const vector bool short *, int, const int);
6936 void vec_dststt (const vector pixel *, int, const int);
6937 void vec_dststt (const vector unsigned int *, int, const int);
6938 void vec_dststt (const vector signed int *, int, const int);
6939 void vec_dststt (const vector bool int *, int, const int);
6940 void vec_dststt (const vector float *, int, const int);
6941 void vec_dststt (const unsigned char *, int, const int);
6942 void vec_dststt (const signed char *, int, const int);
6943 void vec_dststt (const unsigned short *, int, const int);
6944 void vec_dststt (const short *, int, const int);
6945 void vec_dststt (const unsigned int *, int, const int);
6946 void vec_dststt (const int *, int, const int);
6947 void vec_dststt (const unsigned long *, int, const int);
6948 void vec_dststt (const long *, int, const int);
6949 void vec_dststt (const float *, int, const int);
6951 void vec_dstt (const vector unsigned char *, int, const int);
6952 void vec_dstt (const vector signed char *, int, const int);
6953 void vec_dstt (const vector bool char *, int, const int);
6954 void vec_dstt (const vector unsigned short *, int, const int);
6955 void vec_dstt (const vector signed short *, int, const int);
6956 void vec_dstt (const vector bool short *, int, const int);
6957 void vec_dstt (const vector pixel *, int, const int);
6958 void vec_dstt (const vector unsigned int *, int, const int);
6959 void vec_dstt (const vector signed int *, int, const int);
6960 void vec_dstt (const vector bool int *, int, const int);
6961 void vec_dstt (const vector float *, int, const int);
6962 void vec_dstt (const unsigned char *, int, const int);
6963 void vec_dstt (const signed char *, int, const int);
6964 void vec_dstt (const unsigned short *, int, const int);
6965 void vec_dstt (const short *, int, const int);
6966 void vec_dstt (const unsigned int *, int, const int);
6967 void vec_dstt (const int *, int, const int);
6968 void vec_dstt (const unsigned long *, int, const int);
6969 void vec_dstt (const long *, int, const int);
6970 void vec_dstt (const float *, int, const int);
6972 vector float vec_expte (vector float);
6974 vector float vec_floor (vector float);
6976 vector float vec_ld (int, const vector float *);
6977 vector float vec_ld (int, const float *);
6978 vector bool int vec_ld (int, const vector bool int *);
6979 vector signed int vec_ld (int, const vector signed int *);
6980 vector signed int vec_ld (int, const int *);
6981 vector signed int vec_ld (int, const long *);
6982 vector unsigned int vec_ld (int, const vector unsigned int *);
6983 vector unsigned int vec_ld (int, const unsigned int *);
6984 vector unsigned int vec_ld (int, const unsigned long *);
6985 vector bool short vec_ld (int, const vector bool short *);
6986 vector pixel vec_ld (int, const vector pixel *);
6987 vector signed short vec_ld (int, const vector signed short *);
6988 vector signed short vec_ld (int, const short *);
6989 vector unsigned short vec_ld (int, const vector unsigned short *);
6990 vector unsigned short vec_ld (int, const unsigned short *);
6991 vector bool char vec_ld (int, const vector bool char *);
6992 vector signed char vec_ld (int, const vector signed char *);
6993 vector signed char vec_ld (int, const signed char *);
6994 vector unsigned char vec_ld (int, const vector unsigned char *);
6995 vector unsigned char vec_ld (int, const unsigned char *);
6997 vector signed char vec_lde (int, const signed char *);
6998 vector unsigned char vec_lde (int, const unsigned char *);
6999 vector signed short vec_lde (int, const short *);
7000 vector unsigned short vec_lde (int, const unsigned short *);
7001 vector float vec_lde (int, const float *);
7002 vector signed int vec_lde (int, const int *);
7003 vector unsigned int vec_lde (int, const unsigned int *);
7004 vector signed int vec_lde (int, const long *);
7005 vector unsigned int vec_lde (int, const unsigned long *);
7007 vector float vec_lvewx (int, float *);
7008 vector signed int vec_lvewx (int, int *);
7009 vector unsigned int vec_lvewx (int, unsigned int *);
7010 vector signed int vec_lvewx (int, long *);
7011 vector unsigned int vec_lvewx (int, unsigned long *);
7013 vector signed short vec_lvehx (int, short *);
7014 vector unsigned short vec_lvehx (int, unsigned short *);
7016 vector signed char vec_lvebx (int, char *);
7017 vector unsigned char vec_lvebx (int, unsigned char *);
7019 vector float vec_ldl (int, const vector float *);
7020 vector float vec_ldl (int, const float *);
7021 vector bool int vec_ldl (int, const vector bool int *);
7022 vector signed int vec_ldl (int, const vector signed int *);
7023 vector signed int vec_ldl (int, const int *);
7024 vector signed int vec_ldl (int, const long *);
7025 vector unsigned int vec_ldl (int, const vector unsigned int *);
7026 vector unsigned int vec_ldl (int, const unsigned int *);
7027 vector unsigned int vec_ldl (int, const unsigned long *);
7028 vector bool short vec_ldl (int, const vector bool short *);
7029 vector pixel vec_ldl (int, const vector pixel *);
7030 vector signed short vec_ldl (int, const vector signed short *);
7031 vector signed short vec_ldl (int, const short *);
7032 vector unsigned short vec_ldl (int, const vector unsigned short *);
7033 vector unsigned short vec_ldl (int, const unsigned short *);
7034 vector bool char vec_ldl (int, const vector bool char *);
7035 vector signed char vec_ldl (int, const vector signed char *);
7036 vector signed char vec_ldl (int, const signed char *);
7037 vector unsigned char vec_ldl (int, const vector unsigned char *);
7038 vector unsigned char vec_ldl (int, const unsigned char *);
7040 vector float vec_loge (vector float);
7042 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
7043 vector unsigned char vec_lvsl (int, const volatile signed char *);
7044 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
7045 vector unsigned char vec_lvsl (int, const volatile short *);
7046 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
7047 vector unsigned char vec_lvsl (int, const volatile int *);
7048 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
7049 vector unsigned char vec_lvsl (int, const volatile long *);
7050 vector unsigned char vec_lvsl (int, const volatile float *);
7052 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
7053 vector unsigned char vec_lvsr (int, const volatile signed char *);
7054 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
7055 vector unsigned char vec_lvsr (int, const volatile short *);
7056 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
7057 vector unsigned char vec_lvsr (int, const volatile int *);
7058 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
7059 vector unsigned char vec_lvsr (int, const volatile long *);
7060 vector unsigned char vec_lvsr (int, const volatile float *);
7062 vector float vec_madd (vector float, vector float, vector float);
7064 vector signed short vec_madds (vector signed short,
7065 vector signed short,
7066 vector signed short);
7068 vector unsigned char vec_max (vector bool char, vector unsigned char);
7069 vector unsigned char vec_max (vector unsigned char, vector bool char);
7070 vector unsigned char vec_max (vector unsigned char,
7071 vector unsigned char);
7072 vector signed char vec_max (vector bool char, vector signed char);
7073 vector signed char vec_max (vector signed char, vector bool char);
7074 vector signed char vec_max (vector signed char, vector signed char);
7075 vector unsigned short vec_max (vector bool short,
7076 vector unsigned short);
7077 vector unsigned short vec_max (vector unsigned short,
7079 vector unsigned short vec_max (vector unsigned short,
7080 vector unsigned short);
7081 vector signed short vec_max (vector bool short, vector signed short);
7082 vector signed short vec_max (vector signed short, vector bool short);
7083 vector signed short vec_max (vector signed short, vector signed short);
7084 vector unsigned int vec_max (vector bool int, vector unsigned int);
7085 vector unsigned int vec_max (vector unsigned int, vector bool int);
7086 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
7087 vector signed int vec_max (vector bool int, vector signed int);
7088 vector signed int vec_max (vector signed int, vector bool int);
7089 vector signed int vec_max (vector signed int, vector signed int);
7090 vector float vec_max (vector float, vector float);
7092 vector float vec_vmaxfp (vector float, vector float);
7094 vector signed int vec_vmaxsw (vector bool int, vector signed int);
7095 vector signed int vec_vmaxsw (vector signed int, vector bool int);
7096 vector signed int vec_vmaxsw (vector signed int, vector signed int);
7098 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
7099 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
7100 vector unsigned int vec_vmaxuw (vector unsigned int,
7101 vector unsigned int);
7103 vector signed short vec_vmaxsh (vector bool short, vector signed short);
7104 vector signed short vec_vmaxsh (vector signed short, vector bool short);
7105 vector signed short vec_vmaxsh (vector signed short,
7106 vector signed short);
7108 vector unsigned short vec_vmaxuh (vector bool short,
7109 vector unsigned short);
7110 vector unsigned short vec_vmaxuh (vector unsigned short,
7112 vector unsigned short vec_vmaxuh (vector unsigned short,
7113 vector unsigned short);
7115 vector signed char vec_vmaxsb (vector bool char, vector signed char);
7116 vector signed char vec_vmaxsb (vector signed char, vector bool char);
7117 vector signed char vec_vmaxsb (vector signed char, vector signed char);
7119 vector unsigned char vec_vmaxub (vector bool char,
7120 vector unsigned char);
7121 vector unsigned char vec_vmaxub (vector unsigned char,
7123 vector unsigned char vec_vmaxub (vector unsigned char,
7124 vector unsigned char);
7126 vector bool char vec_mergeh (vector bool char, vector bool char);
7127 vector signed char vec_mergeh (vector signed char, vector signed char);
7128 vector unsigned char vec_mergeh (vector unsigned char,
7129 vector unsigned char);
7130 vector bool short vec_mergeh (vector bool short, vector bool short);
7131 vector pixel vec_mergeh (vector pixel, vector pixel);
7132 vector signed short vec_mergeh (vector signed short,
7133 vector signed short);
7134 vector unsigned short vec_mergeh (vector unsigned short,
7135 vector unsigned short);
7136 vector float vec_mergeh (vector float, vector float);
7137 vector bool int vec_mergeh (vector bool int, vector bool int);
7138 vector signed int vec_mergeh (vector signed int, vector signed int);
7139 vector unsigned int vec_mergeh (vector unsigned int,
7140 vector unsigned int);
7142 vector float vec_vmrghw (vector float, vector float);
7143 vector bool int vec_vmrghw (vector bool int, vector bool int);
7144 vector signed int vec_vmrghw (vector signed int, vector signed int);
7145 vector unsigned int vec_vmrghw (vector unsigned int,
7146 vector unsigned int);
7148 vector bool short vec_vmrghh (vector bool short, vector bool short);
7149 vector signed short vec_vmrghh (vector signed short,
7150 vector signed short);
7151 vector unsigned short vec_vmrghh (vector unsigned short,
7152 vector unsigned short);
7153 vector pixel vec_vmrghh (vector pixel, vector pixel);
7155 vector bool char vec_vmrghb (vector bool char, vector bool char);
7156 vector signed char vec_vmrghb (vector signed char, vector signed char);
7157 vector unsigned char vec_vmrghb (vector unsigned char,
7158 vector unsigned char);
7160 vector bool char vec_mergel (vector bool char, vector bool char);
7161 vector signed char vec_mergel (vector signed char, vector signed char);
7162 vector unsigned char vec_mergel (vector unsigned char,
7163 vector unsigned char);
7164 vector bool short vec_mergel (vector bool short, vector bool short);
7165 vector pixel vec_mergel (vector pixel, vector pixel);
7166 vector signed short vec_mergel (vector signed short,
7167 vector signed short);
7168 vector unsigned short vec_mergel (vector unsigned short,
7169 vector unsigned short);
7170 vector float vec_mergel (vector float, vector float);
7171 vector bool int vec_mergel (vector bool int, vector bool int);
7172 vector signed int vec_mergel (vector signed int, vector signed int);
7173 vector unsigned int vec_mergel (vector unsigned int,
7174 vector unsigned int);
7176 vector float vec_vmrglw (vector float, vector float);
7177 vector signed int vec_vmrglw (vector signed int, vector signed int);
7178 vector unsigned int vec_vmrglw (vector unsigned int,
7179 vector unsigned int);
7180 vector bool int vec_vmrglw (vector bool int, vector bool int);
7182 vector bool short vec_vmrglh (vector bool short, vector bool short);
7183 vector signed short vec_vmrglh (vector signed short,
7184 vector signed short);
7185 vector unsigned short vec_vmrglh (vector unsigned short,
7186 vector unsigned short);
7187 vector pixel vec_vmrglh (vector pixel, vector pixel);
7189 vector bool char vec_vmrglb (vector bool char, vector bool char);
7190 vector signed char vec_vmrglb (vector signed char, vector signed char);
7191 vector unsigned char vec_vmrglb (vector unsigned char,
7192 vector unsigned char);
7194 vector unsigned short vec_mfvscr (void);
7196 vector unsigned char vec_min (vector bool char, vector unsigned char);
7197 vector unsigned char vec_min (vector unsigned char, vector bool char);
7198 vector unsigned char vec_min (vector unsigned char,
7199 vector unsigned char);
7200 vector signed char vec_min (vector bool char, vector signed char);
7201 vector signed char vec_min (vector signed char, vector bool char);
7202 vector signed char vec_min (vector signed char, vector signed char);
7203 vector unsigned short vec_min (vector bool short,
7204 vector unsigned short);
7205 vector unsigned short vec_min (vector unsigned short,
7207 vector unsigned short vec_min (vector unsigned short,
7208 vector unsigned short);
7209 vector signed short vec_min (vector bool short, vector signed short);
7210 vector signed short vec_min (vector signed short, vector bool short);
7211 vector signed short vec_min (vector signed short, vector signed short);
7212 vector unsigned int vec_min (vector bool int, vector unsigned int);
7213 vector unsigned int vec_min (vector unsigned int, vector bool int);
7214 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
7215 vector signed int vec_min (vector bool int, vector signed int);
7216 vector signed int vec_min (vector signed int, vector bool int);
7217 vector signed int vec_min (vector signed int, vector signed int);
7218 vector float vec_min (vector float, vector float);
7220 vector float vec_vminfp (vector float, vector float);
7222 vector signed int vec_vminsw (vector bool int, vector signed int);
7223 vector signed int vec_vminsw (vector signed int, vector bool int);
7224 vector signed int vec_vminsw (vector signed int, vector signed int);
7226 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
7227 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
7228 vector unsigned int vec_vminuw (vector unsigned int,
7229 vector unsigned int);
7231 vector signed short vec_vminsh (vector bool short, vector signed short);
7232 vector signed short vec_vminsh (vector signed short, vector bool short);
7233 vector signed short vec_vminsh (vector signed short,
7234 vector signed short);
7236 vector unsigned short vec_vminuh (vector bool short,
7237 vector unsigned short);
7238 vector unsigned short vec_vminuh (vector unsigned short,
7240 vector unsigned short vec_vminuh (vector unsigned short,
7241 vector unsigned short);
7243 vector signed char vec_vminsb (vector bool char, vector signed char);
7244 vector signed char vec_vminsb (vector signed char, vector bool char);
7245 vector signed char vec_vminsb (vector signed char, vector signed char);
7247 vector unsigned char vec_vminub (vector bool char,
7248 vector unsigned char);
7249 vector unsigned char vec_vminub (vector unsigned char,
7251 vector unsigned char vec_vminub (vector unsigned char,
7252 vector unsigned char);
7254 vector signed short vec_mladd (vector signed short,
7255 vector signed short,
7256 vector signed short);
7257 vector signed short vec_mladd (vector signed short,
7258 vector unsigned short,
7259 vector unsigned short);
7260 vector signed short vec_mladd (vector unsigned short,
7261 vector signed short,
7262 vector signed short);
7263 vector unsigned short vec_mladd (vector unsigned short,
7264 vector unsigned short,
7265 vector unsigned short);
7267 vector signed short vec_mradds (vector signed short,
7268 vector signed short,
7269 vector signed short);
7271 vector unsigned int vec_msum (vector unsigned char,
7272 vector unsigned char,
7273 vector unsigned int);
7274 vector signed int vec_msum (vector signed char,
7275 vector unsigned char,
7277 vector unsigned int vec_msum (vector unsigned short,
7278 vector unsigned short,
7279 vector unsigned int);
7280 vector signed int vec_msum (vector signed short,
7281 vector signed short,
7284 vector signed int vec_vmsumshm (vector signed short,
7285 vector signed short,
7288 vector unsigned int vec_vmsumuhm (vector unsigned short,
7289 vector unsigned short,
7290 vector unsigned int);
7292 vector signed int vec_vmsummbm (vector signed char,
7293 vector unsigned char,
7296 vector unsigned int vec_vmsumubm (vector unsigned char,
7297 vector unsigned char,
7298 vector unsigned int);
7300 vector unsigned int vec_msums (vector unsigned short,
7301 vector unsigned short,
7302 vector unsigned int);
7303 vector signed int vec_msums (vector signed short,
7304 vector signed short,
7307 vector signed int vec_vmsumshs (vector signed short,
7308 vector signed short,
7311 vector unsigned int vec_vmsumuhs (vector unsigned short,
7312 vector unsigned short,
7313 vector unsigned int);
7315 void vec_mtvscr (vector signed int);
7316 void vec_mtvscr (vector unsigned int);
7317 void vec_mtvscr (vector bool int);
7318 void vec_mtvscr (vector signed short);
7319 void vec_mtvscr (vector unsigned short);
7320 void vec_mtvscr (vector bool short);
7321 void vec_mtvscr (vector pixel);
7322 void vec_mtvscr (vector signed char);
7323 void vec_mtvscr (vector unsigned char);
7324 void vec_mtvscr (vector bool char);
7326 vector unsigned short vec_mule (vector unsigned char,
7327 vector unsigned char);
7328 vector signed short vec_mule (vector signed char,
7329 vector signed char);
7330 vector unsigned int vec_mule (vector unsigned short,
7331 vector unsigned short);
7332 vector signed int vec_mule (vector signed short, vector signed short);
7334 vector signed int vec_vmulesh (vector signed short,
7335 vector signed short);
7337 vector unsigned int vec_vmuleuh (vector unsigned short,
7338 vector unsigned short);
7340 vector signed short vec_vmulesb (vector signed char,
7341 vector signed char);
7343 vector unsigned short vec_vmuleub (vector unsigned char,
7344 vector unsigned char);
7346 vector unsigned short vec_mulo (vector unsigned char,
7347 vector unsigned char);
7348 vector signed short vec_mulo (vector signed char, vector signed char);
7349 vector unsigned int vec_mulo (vector unsigned short,
7350 vector unsigned short);
7351 vector signed int vec_mulo (vector signed short, vector signed short);
7353 vector signed int vec_vmulosh (vector signed short,
7354 vector signed short);
7356 vector unsigned int vec_vmulouh (vector unsigned short,
7357 vector unsigned short);
7359 vector signed short vec_vmulosb (vector signed char,
7360 vector signed char);
7362 vector unsigned short vec_vmuloub (vector unsigned char,
7363 vector unsigned char);
7365 vector float vec_nmsub (vector float, vector float, vector float);
7367 vector float vec_nor (vector float, vector float);
7368 vector signed int vec_nor (vector signed int, vector signed int);
7369 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
7370 vector bool int vec_nor (vector bool int, vector bool int);
7371 vector signed short vec_nor (vector signed short, vector signed short);
7372 vector unsigned short vec_nor (vector unsigned short,
7373 vector unsigned short);
7374 vector bool short vec_nor (vector bool short, vector bool short);
7375 vector signed char vec_nor (vector signed char, vector signed char);
7376 vector unsigned char vec_nor (vector unsigned char,
7377 vector unsigned char);
7378 vector bool char vec_nor (vector bool char, vector bool char);
7380 vector float vec_or (vector float, vector float);
7381 vector float vec_or (vector float, vector bool int);
7382 vector float vec_or (vector bool int, vector float);
7383 vector bool int vec_or (vector bool int, vector bool int);
7384 vector signed int vec_or (vector bool int, vector signed int);
7385 vector signed int vec_or (vector signed int, vector bool int);
7386 vector signed int vec_or (vector signed int, vector signed int);
7387 vector unsigned int vec_or (vector bool int, vector unsigned int);
7388 vector unsigned int vec_or (vector unsigned int, vector bool int);
7389 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
7390 vector bool short vec_or (vector bool short, vector bool short);
7391 vector signed short vec_or (vector bool short, vector signed short);
7392 vector signed short vec_or (vector signed short, vector bool short);
7393 vector signed short vec_or (vector signed short, vector signed short);
7394 vector unsigned short vec_or (vector bool short, vector unsigned short);
7395 vector unsigned short vec_or (vector unsigned short, vector bool short);
7396 vector unsigned short vec_or (vector unsigned short,
7397 vector unsigned short);
7398 vector signed char vec_or (vector bool char, vector signed char);
7399 vector bool char vec_or (vector bool char, vector bool char);
7400 vector signed char vec_or (vector signed char, vector bool char);
7401 vector signed char vec_or (vector signed char, vector signed char);
7402 vector unsigned char vec_or (vector bool char, vector unsigned char);
7403 vector unsigned char vec_or (vector unsigned char, vector bool char);
7404 vector unsigned char vec_or (vector unsigned char,
7405 vector unsigned char);
7407 vector signed char vec_pack (vector signed short, vector signed short);
7408 vector unsigned char vec_pack (vector unsigned short,
7409 vector unsigned short);
7410 vector bool char vec_pack (vector bool short, vector bool short);
7411 vector signed short vec_pack (vector signed int, vector signed int);
7412 vector unsigned short vec_pack (vector unsigned int,
7413 vector unsigned int);
7414 vector bool short vec_pack (vector bool int, vector bool int);
7416 vector bool short vec_vpkuwum (vector bool int, vector bool int);
7417 vector signed short vec_vpkuwum (vector signed int, vector signed int);
7418 vector unsigned short vec_vpkuwum (vector unsigned int,
7419 vector unsigned int);
7421 vector bool char vec_vpkuhum (vector bool short, vector bool short);
7422 vector signed char vec_vpkuhum (vector signed short,
7423 vector signed short);
7424 vector unsigned char vec_vpkuhum (vector unsigned short,
7425 vector unsigned short);
7427 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
7429 vector unsigned char vec_packs (vector unsigned short,
7430 vector unsigned short);
7431 vector signed char vec_packs (vector signed short, vector signed short);
7432 vector unsigned short vec_packs (vector unsigned int,
7433 vector unsigned int);
7434 vector signed short vec_packs (vector signed int, vector signed int);
7436 vector signed short vec_vpkswss (vector signed int, vector signed int);
7438 vector unsigned short vec_vpkuwus (vector unsigned int,
7439 vector unsigned int);
7441 vector signed char vec_vpkshss (vector signed short,
7442 vector signed short);
7444 vector unsigned char vec_vpkuhus (vector unsigned short,
7445 vector unsigned short);
7447 vector unsigned char vec_packsu (vector unsigned short,
7448 vector unsigned short);
7449 vector unsigned char vec_packsu (vector signed short,
7450 vector signed short);
7451 vector unsigned short vec_packsu (vector unsigned int,
7452 vector unsigned int);
7453 vector unsigned short vec_packsu (vector signed int, vector signed int);
7455 vector unsigned short vec_vpkswus (vector signed int,
7458 vector unsigned char vec_vpkshus (vector signed short,
7459 vector signed short);
7461 vector float vec_perm (vector float,
7463 vector unsigned char);
7464 vector signed int vec_perm (vector signed int,
7466 vector unsigned char);
7467 vector unsigned int vec_perm (vector unsigned int,
7468 vector unsigned int,
7469 vector unsigned char);
7470 vector bool int vec_perm (vector bool int,
7472 vector unsigned char);
7473 vector signed short vec_perm (vector signed short,
7474 vector signed short,
7475 vector unsigned char);
7476 vector unsigned short vec_perm (vector unsigned short,
7477 vector unsigned short,
7478 vector unsigned char);
7479 vector bool short vec_perm (vector bool short,
7481 vector unsigned char);
7482 vector pixel vec_perm (vector pixel,
7484 vector unsigned char);
7485 vector signed char vec_perm (vector signed char,
7487 vector unsigned char);
7488 vector unsigned char vec_perm (vector unsigned char,
7489 vector unsigned char,
7490 vector unsigned char);
7491 vector bool char vec_perm (vector bool char,
7493 vector unsigned char);
7495 vector float vec_re (vector float);
7497 vector signed char vec_rl (vector signed char,
7498 vector unsigned char);
7499 vector unsigned char vec_rl (vector unsigned char,
7500 vector unsigned char);
7501 vector signed short vec_rl (vector signed short, vector unsigned short);
7502 vector unsigned short vec_rl (vector unsigned short,
7503 vector unsigned short);
7504 vector signed int vec_rl (vector signed int, vector unsigned int);
7505 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
7507 vector signed int vec_vrlw (vector signed int, vector unsigned int);
7508 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
7510 vector signed short vec_vrlh (vector signed short,
7511 vector unsigned short);
7512 vector unsigned short vec_vrlh (vector unsigned short,
7513 vector unsigned short);
7515 vector signed char vec_vrlb (vector signed char, vector unsigned char);
7516 vector unsigned char vec_vrlb (vector unsigned char,
7517 vector unsigned char);
7519 vector float vec_round (vector float);
7521 vector float vec_rsqrte (vector float);
7523 vector float vec_sel (vector float, vector float, vector bool int);
7524 vector float vec_sel (vector float, vector float, vector unsigned int);
7525 vector signed int vec_sel (vector signed int,
7528 vector signed int vec_sel (vector signed int,
7530 vector unsigned int);
7531 vector unsigned int vec_sel (vector unsigned int,
7532 vector unsigned int,
7534 vector unsigned int vec_sel (vector unsigned int,
7535 vector unsigned int,
7536 vector unsigned int);
7537 vector bool int vec_sel (vector bool int,
7540 vector bool int vec_sel (vector bool int,
7542 vector unsigned int);
7543 vector signed short vec_sel (vector signed short,
7544 vector signed short,
7546 vector signed short vec_sel (vector signed short,
7547 vector signed short,
7548 vector unsigned short);
7549 vector unsigned short vec_sel (vector unsigned short,
7550 vector unsigned short,
7552 vector unsigned short vec_sel (vector unsigned short,
7553 vector unsigned short,
7554 vector unsigned short);
7555 vector bool short vec_sel (vector bool short,
7558 vector bool short vec_sel (vector bool short,
7560 vector unsigned short);
7561 vector signed char vec_sel (vector signed char,
7564 vector signed char vec_sel (vector signed char,
7566 vector unsigned char);
7567 vector unsigned char vec_sel (vector unsigned char,
7568 vector unsigned char,
7570 vector unsigned char vec_sel (vector unsigned char,
7571 vector unsigned char,
7572 vector unsigned char);
7573 vector bool char vec_sel (vector bool char,
7576 vector bool char vec_sel (vector bool char,
7578 vector unsigned char);
7580 vector signed char vec_sl (vector signed char,
7581 vector unsigned char);
7582 vector unsigned char vec_sl (vector unsigned char,
7583 vector unsigned char);
7584 vector signed short vec_sl (vector signed short, vector unsigned short);
7585 vector unsigned short vec_sl (vector unsigned short,
7586 vector unsigned short);
7587 vector signed int vec_sl (vector signed int, vector unsigned int);
7588 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
7590 vector signed int vec_vslw (vector signed int, vector unsigned int);
7591 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
7593 vector signed short vec_vslh (vector signed short,
7594 vector unsigned short);
7595 vector unsigned short vec_vslh (vector unsigned short,
7596 vector unsigned short);
7598 vector signed char vec_vslb (vector signed char, vector unsigned char);
7599 vector unsigned char vec_vslb (vector unsigned char,
7600 vector unsigned char);
7602 vector float vec_sld (vector float, vector float, const int);
7603 vector signed int vec_sld (vector signed int,
7606 vector unsigned int vec_sld (vector unsigned int,
7607 vector unsigned int,
7609 vector bool int vec_sld (vector bool int,
7612 vector signed short vec_sld (vector signed short,
7613 vector signed short,
7615 vector unsigned short vec_sld (vector unsigned short,
7616 vector unsigned short,
7618 vector bool short vec_sld (vector bool short,
7621 vector pixel vec_sld (vector pixel,
7624 vector signed char vec_sld (vector signed char,
7627 vector unsigned char vec_sld (vector unsigned char,
7628 vector unsigned char,
7630 vector bool char vec_sld (vector bool char,
7634 vector signed int vec_sll (vector signed int,
7635 vector unsigned int);
7636 vector signed int vec_sll (vector signed int,
7637 vector unsigned short);
7638 vector signed int vec_sll (vector signed int,
7639 vector unsigned char);
7640 vector unsigned int vec_sll (vector unsigned int,
7641 vector unsigned int);
7642 vector unsigned int vec_sll (vector unsigned int,
7643 vector unsigned short);
7644 vector unsigned int vec_sll (vector unsigned int,
7645 vector unsigned char);
7646 vector bool int vec_sll (vector bool int,
7647 vector unsigned int);
7648 vector bool int vec_sll (vector bool int,
7649 vector unsigned short);
7650 vector bool int vec_sll (vector bool int,
7651 vector unsigned char);
7652 vector signed short vec_sll (vector signed short,
7653 vector unsigned int);
7654 vector signed short vec_sll (vector signed short,
7655 vector unsigned short);
7656 vector signed short vec_sll (vector signed short,
7657 vector unsigned char);
7658 vector unsigned short vec_sll (vector unsigned short,
7659 vector unsigned int);
7660 vector unsigned short vec_sll (vector unsigned short,
7661 vector unsigned short);
7662 vector unsigned short vec_sll (vector unsigned short,
7663 vector unsigned char);
7664 vector bool short vec_sll (vector bool short, vector unsigned int);
7665 vector bool short vec_sll (vector bool short, vector unsigned short);
7666 vector bool short vec_sll (vector bool short, vector unsigned char);
7667 vector pixel vec_sll (vector pixel, vector unsigned int);
7668 vector pixel vec_sll (vector pixel, vector unsigned short);
7669 vector pixel vec_sll (vector pixel, vector unsigned char);
7670 vector signed char vec_sll (vector signed char, vector unsigned int);
7671 vector signed char vec_sll (vector signed char, vector unsigned short);
7672 vector signed char vec_sll (vector signed char, vector unsigned char);
7673 vector unsigned char vec_sll (vector unsigned char,
7674 vector unsigned int);
7675 vector unsigned char vec_sll (vector unsigned char,
7676 vector unsigned short);
7677 vector unsigned char vec_sll (vector unsigned char,
7678 vector unsigned char);
7679 vector bool char vec_sll (vector bool char, vector unsigned int);
7680 vector bool char vec_sll (vector bool char, vector unsigned short);
7681 vector bool char vec_sll (vector bool char, vector unsigned char);
7683 vector float vec_slo (vector float, vector signed char);
7684 vector float vec_slo (vector float, vector unsigned char);
7685 vector signed int vec_slo (vector signed int, vector signed char);
7686 vector signed int vec_slo (vector signed int, vector unsigned char);
7687 vector unsigned int vec_slo (vector unsigned int, vector signed char);
7688 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
7689 vector signed short vec_slo (vector signed short, vector signed char);
7690 vector signed short vec_slo (vector signed short, vector unsigned char);
7691 vector unsigned short vec_slo (vector unsigned short,
7692 vector signed char);
7693 vector unsigned short vec_slo (vector unsigned short,
7694 vector unsigned char);
7695 vector pixel vec_slo (vector pixel, vector signed char);
7696 vector pixel vec_slo (vector pixel, vector unsigned char);
7697 vector signed char vec_slo (vector signed char, vector signed char);
7698 vector signed char vec_slo (vector signed char, vector unsigned char);
7699 vector unsigned char vec_slo (vector unsigned char, vector signed char);
7700 vector unsigned char vec_slo (vector unsigned char,
7701 vector unsigned char);
7703 vector signed char vec_splat (vector signed char, const int);
7704 vector unsigned char vec_splat (vector unsigned char, const int);
7705 vector bool char vec_splat (vector bool char, const int);
7706 vector signed short vec_splat (vector signed short, const int);
7707 vector unsigned short vec_splat (vector unsigned short, const int);
7708 vector bool short vec_splat (vector bool short, const int);
7709 vector pixel vec_splat (vector pixel, const int);
7710 vector float vec_splat (vector float, const int);
7711 vector signed int vec_splat (vector signed int, const int);
7712 vector unsigned int vec_splat (vector unsigned int, const int);
7713 vector bool int vec_splat (vector bool int, const int);
7715 vector float vec_vspltw (vector float, const int);
7716 vector signed int vec_vspltw (vector signed int, const int);
7717 vector unsigned int vec_vspltw (vector unsigned int, const int);
7718 vector bool int vec_vspltw (vector bool int, const int);
7720 vector bool short vec_vsplth (vector bool short, const int);
7721 vector signed short vec_vsplth (vector signed short, const int);
7722 vector unsigned short vec_vsplth (vector unsigned short, const int);
7723 vector pixel vec_vsplth (vector pixel, const int);
7725 vector signed char vec_vspltb (vector signed char, const int);
7726 vector unsigned char vec_vspltb (vector unsigned char, const int);
7727 vector bool char vec_vspltb (vector bool char, const int);
7729 vector signed char vec_splat_s8 (const int);
7731 vector signed short vec_splat_s16 (const int);
7733 vector signed int vec_splat_s32 (const int);
7735 vector unsigned char vec_splat_u8 (const int);
7737 vector unsigned short vec_splat_u16 (const int);
7739 vector unsigned int vec_splat_u32 (const int);
7741 vector signed char vec_sr (vector signed char, vector unsigned char);
7742 vector unsigned char vec_sr (vector unsigned char,
7743 vector unsigned char);
7744 vector signed short vec_sr (vector signed short,
7745 vector unsigned short);
7746 vector unsigned short vec_sr (vector unsigned short,
7747 vector unsigned short);
7748 vector signed int vec_sr (vector signed int, vector unsigned int);
7749 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
7751 vector signed int vec_vsrw (vector signed int, vector unsigned int);
7752 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
7754 vector signed short vec_vsrh (vector signed short,
7755 vector unsigned short);
7756 vector unsigned short vec_vsrh (vector unsigned short,
7757 vector unsigned short);
7759 vector signed char vec_vsrb (vector signed char, vector unsigned char);
7760 vector unsigned char vec_vsrb (vector unsigned char,
7761 vector unsigned char);
7763 vector signed char vec_sra (vector signed char, vector unsigned char);
7764 vector unsigned char vec_sra (vector unsigned char,
7765 vector unsigned char);
7766 vector signed short vec_sra (vector signed short,
7767 vector unsigned short);
7768 vector unsigned short vec_sra (vector unsigned short,
7769 vector unsigned short);
7770 vector signed int vec_sra (vector signed int, vector unsigned int);
7771 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
7773 vector signed int vec_vsraw (vector signed int, vector unsigned int);
7774 vector unsigned int vec_vsraw (vector unsigned int,
7775 vector unsigned int);
7777 vector signed short vec_vsrah (vector signed short,
7778 vector unsigned short);
7779 vector unsigned short vec_vsrah (vector unsigned short,
7780 vector unsigned short);
7782 vector signed char vec_vsrab (vector signed char, vector unsigned char);
7783 vector unsigned char vec_vsrab (vector unsigned char,
7784 vector unsigned char);
7786 vector signed int vec_srl (vector signed int, vector unsigned int);
7787 vector signed int vec_srl (vector signed int, vector unsigned short);
7788 vector signed int vec_srl (vector signed int, vector unsigned char);
7789 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
7790 vector unsigned int vec_srl (vector unsigned int,
7791 vector unsigned short);
7792 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
7793 vector bool int vec_srl (vector bool int, vector unsigned int);
7794 vector bool int vec_srl (vector bool int, vector unsigned short);
7795 vector bool int vec_srl (vector bool int, vector unsigned char);
7796 vector signed short vec_srl (vector signed short, vector unsigned int);
7797 vector signed short vec_srl (vector signed short,
7798 vector unsigned short);
7799 vector signed short vec_srl (vector signed short, vector unsigned char);
7800 vector unsigned short vec_srl (vector unsigned short,
7801 vector unsigned int);
7802 vector unsigned short vec_srl (vector unsigned short,
7803 vector unsigned short);
7804 vector unsigned short vec_srl (vector unsigned short,
7805 vector unsigned char);
7806 vector bool short vec_srl (vector bool short, vector unsigned int);
7807 vector bool short vec_srl (vector bool short, vector unsigned short);
7808 vector bool short vec_srl (vector bool short, vector unsigned char);
7809 vector pixel vec_srl (vector pixel, vector unsigned int);
7810 vector pixel vec_srl (vector pixel, vector unsigned short);
7811 vector pixel vec_srl (vector pixel, vector unsigned char);
7812 vector signed char vec_srl (vector signed char, vector unsigned int);
7813 vector signed char vec_srl (vector signed char, vector unsigned short);
7814 vector signed char vec_srl (vector signed char, vector unsigned char);
7815 vector unsigned char vec_srl (vector unsigned char,
7816 vector unsigned int);
7817 vector unsigned char vec_srl (vector unsigned char,
7818 vector unsigned short);
7819 vector unsigned char vec_srl (vector unsigned char,
7820 vector unsigned char);
7821 vector bool char vec_srl (vector bool char, vector unsigned int);
7822 vector bool char vec_srl (vector bool char, vector unsigned short);
7823 vector bool char vec_srl (vector bool char, vector unsigned char);
7825 vector float vec_sro (vector float, vector signed char);
7826 vector float vec_sro (vector float, vector unsigned char);
7827 vector signed int vec_sro (vector signed int, vector signed char);
7828 vector signed int vec_sro (vector signed int, vector unsigned char);
7829 vector unsigned int vec_sro (vector unsigned int, vector signed char);
7830 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
7831 vector signed short vec_sro (vector signed short, vector signed char);
7832 vector signed short vec_sro (vector signed short, vector unsigned char);
7833 vector unsigned short vec_sro (vector unsigned short,
7834 vector signed char);
7835 vector unsigned short vec_sro (vector unsigned short,
7836 vector unsigned char);
7837 vector pixel vec_sro (vector pixel, vector signed char);
7838 vector pixel vec_sro (vector pixel, vector unsigned char);
7839 vector signed char vec_sro (vector signed char, vector signed char);
7840 vector signed char vec_sro (vector signed char, vector unsigned char);
7841 vector unsigned char vec_sro (vector unsigned char, vector signed char);
7842 vector unsigned char vec_sro (vector unsigned char,
7843 vector unsigned char);
7845 void vec_st (vector float, int, vector float *);
7846 void vec_st (vector float, int, float *);
7847 void vec_st (vector signed int, int, vector signed int *);
7848 void vec_st (vector signed int, int, int *);
7849 void vec_st (vector unsigned int, int, vector unsigned int *);
7850 void vec_st (vector unsigned int, int, unsigned int *);
7851 void vec_st (vector bool int, int, vector bool int *);
7852 void vec_st (vector bool int, int, unsigned int *);
7853 void vec_st (vector bool int, int, int *);
7854 void vec_st (vector signed short, int, vector signed short *);
7855 void vec_st (vector signed short, int, short *);
7856 void vec_st (vector unsigned short, int, vector unsigned short *);
7857 void vec_st (vector unsigned short, int, unsigned short *);
7858 void vec_st (vector bool short, int, vector bool short *);
7859 void vec_st (vector bool short, int, unsigned short *);
7860 void vec_st (vector pixel, int, vector pixel *);
7861 void vec_st (vector pixel, int, unsigned short *);
7862 void vec_st (vector pixel, int, short *);
7863 void vec_st (vector bool short, int, short *);
7864 void vec_st (vector signed char, int, vector signed char *);
7865 void vec_st (vector signed char, int, signed char *);
7866 void vec_st (vector unsigned char, int, vector unsigned char *);
7867 void vec_st (vector unsigned char, int, unsigned char *);
7868 void vec_st (vector bool char, int, vector bool char *);
7869 void vec_st (vector bool char, int, unsigned char *);
7870 void vec_st (vector bool char, int, signed char *);
7872 void vec_ste (vector signed char, int, signed char *);
7873 void vec_ste (vector unsigned char, int, unsigned char *);
7874 void vec_ste (vector bool char, int, signed char *);
7875 void vec_ste (vector bool char, int, unsigned char *);
7876 void vec_ste (vector signed short, int, short *);
7877 void vec_ste (vector unsigned short, int, unsigned short *);
7878 void vec_ste (vector bool short, int, short *);
7879 void vec_ste (vector bool short, int, unsigned short *);
7880 void vec_ste (vector pixel, int, short *);
7881 void vec_ste (vector pixel, int, unsigned short *);
7882 void vec_ste (vector float, int, float *);
7883 void vec_ste (vector signed int, int, int *);
7884 void vec_ste (vector unsigned int, int, unsigned int *);
7885 void vec_ste (vector bool int, int, int *);
7886 void vec_ste (vector bool int, int, unsigned int *);
7888 void vec_stvewx (vector float, int, float *);
7889 void vec_stvewx (vector signed int, int, int *);
7890 void vec_stvewx (vector unsigned int, int, unsigned int *);
7891 void vec_stvewx (vector bool int, int, int *);
7892 void vec_stvewx (vector bool int, int, unsigned int *);
7894 void vec_stvehx (vector signed short, int, short *);
7895 void vec_stvehx (vector unsigned short, int, unsigned short *);
7896 void vec_stvehx (vector bool short, int, short *);
7897 void vec_stvehx (vector bool short, int, unsigned short *);
7898 void vec_stvehx (vector pixel, int, short *);
7899 void vec_stvehx (vector pixel, int, unsigned short *);
7901 void vec_stvebx (vector signed char, int, signed char *);
7902 void vec_stvebx (vector unsigned char, int, unsigned char *);
7903 void vec_stvebx (vector bool char, int, signed char *);
7904 void vec_stvebx (vector bool char, int, unsigned char *);
7906 void vec_stl (vector float, int, vector float *);
7907 void vec_stl (vector float, int, float *);
7908 void vec_stl (vector signed int, int, vector signed int *);
7909 void vec_stl (vector signed int, int, int *);
7910 void vec_stl (vector unsigned int, int, vector unsigned int *);
7911 void vec_stl (vector unsigned int, int, unsigned int *);
7912 void vec_stl (vector bool int, int, vector bool int *);
7913 void vec_stl (vector bool int, int, unsigned int *);
7914 void vec_stl (vector bool int, int, int *);
7915 void vec_stl (vector signed short, int, vector signed short *);
7916 void vec_stl (vector signed short, int, short *);
7917 void vec_stl (vector unsigned short, int, vector unsigned short *);
7918 void vec_stl (vector unsigned short, int, unsigned short *);
7919 void vec_stl (vector bool short, int, vector bool short *);
7920 void vec_stl (vector bool short, int, unsigned short *);
7921 void vec_stl (vector bool short, int, short *);
7922 void vec_stl (vector pixel, int, vector pixel *);
7923 void vec_stl (vector pixel, int, unsigned short *);
7924 void vec_stl (vector pixel, int, short *);
7925 void vec_stl (vector signed char, int, vector signed char *);
7926 void vec_stl (vector signed char, int, signed char *);
7927 void vec_stl (vector unsigned char, int, vector unsigned char *);
7928 void vec_stl (vector unsigned char, int, unsigned char *);
7929 void vec_stl (vector bool char, int, vector bool char *);
7930 void vec_stl (vector bool char, int, unsigned char *);
7931 void vec_stl (vector bool char, int, signed char *);
7933 vector signed char vec_sub (vector bool char, vector signed char);
7934 vector signed char vec_sub (vector signed char, vector bool char);
7935 vector signed char vec_sub (vector signed char, vector signed char);
7936 vector unsigned char vec_sub (vector bool char, vector unsigned char);
7937 vector unsigned char vec_sub (vector unsigned char, vector bool char);
7938 vector unsigned char vec_sub (vector unsigned char,
7939 vector unsigned char);
7940 vector signed short vec_sub (vector bool short, vector signed short);
7941 vector signed short vec_sub (vector signed short, vector bool short);
7942 vector signed short vec_sub (vector signed short, vector signed short);
7943 vector unsigned short vec_sub (vector bool short,
7944 vector unsigned short);
7945 vector unsigned short vec_sub (vector unsigned short,
7947 vector unsigned short vec_sub (vector unsigned short,
7948 vector unsigned short);
7949 vector signed int vec_sub (vector bool int, vector signed int);
7950 vector signed int vec_sub (vector signed int, vector bool int);
7951 vector signed int vec_sub (vector signed int, vector signed int);
7952 vector unsigned int vec_sub (vector bool int, vector unsigned int);
7953 vector unsigned int vec_sub (vector unsigned int, vector bool int);
7954 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
7955 vector float vec_sub (vector float, vector float);
7957 vector float vec_vsubfp (vector float, vector float);
7959 vector signed int vec_vsubuwm (vector bool int, vector signed int);
7960 vector signed int vec_vsubuwm (vector signed int, vector bool int);
7961 vector signed int vec_vsubuwm (vector signed int, vector signed int);
7962 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
7963 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
7964 vector unsigned int vec_vsubuwm (vector unsigned int,
7965 vector unsigned int);
7967 vector signed short vec_vsubuhm (vector bool short,
7968 vector signed short);
7969 vector signed short vec_vsubuhm (vector signed short,
7971 vector signed short vec_vsubuhm (vector signed short,
7972 vector signed short);
7973 vector unsigned short vec_vsubuhm (vector bool short,
7974 vector unsigned short);
7975 vector unsigned short vec_vsubuhm (vector unsigned short,
7977 vector unsigned short vec_vsubuhm (vector unsigned short,
7978 vector unsigned short);
7980 vector signed char vec_vsububm (vector bool char, vector signed char);
7981 vector signed char vec_vsububm (vector signed char, vector bool char);
7982 vector signed char vec_vsububm (vector signed char, vector signed char);
7983 vector unsigned char vec_vsububm (vector bool char,
7984 vector unsigned char);
7985 vector unsigned char vec_vsububm (vector unsigned char,
7987 vector unsigned char vec_vsububm (vector unsigned char,
7988 vector unsigned char);
7990 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
7992 vector unsigned char vec_subs (vector bool char, vector unsigned char);
7993 vector unsigned char vec_subs (vector unsigned char, vector bool char);
7994 vector unsigned char vec_subs (vector unsigned char,
7995 vector unsigned char);
7996 vector signed char vec_subs (vector bool char, vector signed char);
7997 vector signed char vec_subs (vector signed char, vector bool char);
7998 vector signed char vec_subs (vector signed char, vector signed char);
7999 vector unsigned short vec_subs (vector bool short,
8000 vector unsigned short);
8001 vector unsigned short vec_subs (vector unsigned short,
8003 vector unsigned short vec_subs (vector unsigned short,
8004 vector unsigned short);
8005 vector signed short vec_subs (vector bool short, vector signed short);
8006 vector signed short vec_subs (vector signed short, vector bool short);
8007 vector signed short vec_subs (vector signed short, vector signed short);
8008 vector unsigned int vec_subs (vector bool int, vector unsigned int);
8009 vector unsigned int vec_subs (vector unsigned int, vector bool int);
8010 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
8011 vector signed int vec_subs (vector bool int, vector signed int);
8012 vector signed int vec_subs (vector signed int, vector bool int);
8013 vector signed int vec_subs (vector signed int, vector signed int);
8015 vector signed int vec_vsubsws (vector bool int, vector signed int);
8016 vector signed int vec_vsubsws (vector signed int, vector bool int);
8017 vector signed int vec_vsubsws (vector signed int, vector signed int);
8019 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
8020 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
8021 vector unsigned int vec_vsubuws (vector unsigned int,
8022 vector unsigned int);
8024 vector signed short vec_vsubshs (vector bool short,
8025 vector signed short);
8026 vector signed short vec_vsubshs (vector signed short,
8028 vector signed short vec_vsubshs (vector signed short,
8029 vector signed short);
8031 vector unsigned short vec_vsubuhs (vector bool short,
8032 vector unsigned short);
8033 vector unsigned short vec_vsubuhs (vector unsigned short,
8035 vector unsigned short vec_vsubuhs (vector unsigned short,
8036 vector unsigned short);
8038 vector signed char vec_vsubsbs (vector bool char, vector signed char);
8039 vector signed char vec_vsubsbs (vector signed char, vector bool char);
8040 vector signed char vec_vsubsbs (vector signed char, vector signed char);
8042 vector unsigned char vec_vsububs (vector bool char,
8043 vector unsigned char);
8044 vector unsigned char vec_vsububs (vector unsigned char,
8046 vector unsigned char vec_vsububs (vector unsigned char,
8047 vector unsigned char);
8049 vector unsigned int vec_sum4s (vector unsigned char,
8050 vector unsigned int);
8051 vector signed int vec_sum4s (vector signed char, vector signed int);
8052 vector signed int vec_sum4s (vector signed short, vector signed int);
8054 vector signed int vec_vsum4shs (vector signed short, vector signed int);
8056 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
8058 vector unsigned int vec_vsum4ubs (vector unsigned char,
8059 vector unsigned int);
8061 vector signed int vec_sum2s (vector signed int, vector signed int);
8063 vector signed int vec_sums (vector signed int, vector signed int);
8065 vector float vec_trunc (vector float);
8067 vector signed short vec_unpackh (vector signed char);
8068 vector bool short vec_unpackh (vector bool char);
8069 vector signed int vec_unpackh (vector signed short);
8070 vector bool int vec_unpackh (vector bool short);
8071 vector unsigned int vec_unpackh (vector pixel);
8073 vector bool int vec_vupkhsh (vector bool short);
8074 vector signed int vec_vupkhsh (vector signed short);
8076 vector unsigned int vec_vupkhpx (vector pixel);
8078 vector bool short vec_vupkhsb (vector bool char);
8079 vector signed short vec_vupkhsb (vector signed char);
8081 vector signed short vec_unpackl (vector signed char);
8082 vector bool short vec_unpackl (vector bool char);
8083 vector unsigned int vec_unpackl (vector pixel);
8084 vector signed int vec_unpackl (vector signed short);
8085 vector bool int vec_unpackl (vector bool short);
8087 vector unsigned int vec_vupklpx (vector pixel);
8089 vector bool int vec_vupklsh (vector bool short);
8090 vector signed int vec_vupklsh (vector signed short);
8092 vector bool short vec_vupklsb (vector bool char);
8093 vector signed short vec_vupklsb (vector signed char);
8095 vector float vec_xor (vector float, vector float);
8096 vector float vec_xor (vector float, vector bool int);
8097 vector float vec_xor (vector bool int, vector float);
8098 vector bool int vec_xor (vector bool int, vector bool int);
8099 vector signed int vec_xor (vector bool int, vector signed int);
8100 vector signed int vec_xor (vector signed int, vector bool int);
8101 vector signed int vec_xor (vector signed int, vector signed int);
8102 vector unsigned int vec_xor (vector bool int, vector unsigned int);
8103 vector unsigned int vec_xor (vector unsigned int, vector bool int);
8104 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
8105 vector bool short vec_xor (vector bool short, vector bool short);
8106 vector signed short vec_xor (vector bool short, vector signed short);
8107 vector signed short vec_xor (vector signed short, vector bool short);
8108 vector signed short vec_xor (vector signed short, vector signed short);
8109 vector unsigned short vec_xor (vector bool short,
8110 vector unsigned short);
8111 vector unsigned short vec_xor (vector unsigned short,
8113 vector unsigned short vec_xor (vector unsigned short,
8114 vector unsigned short);
8115 vector signed char vec_xor (vector bool char, vector signed char);
8116 vector bool char vec_xor (vector bool char, vector bool char);
8117 vector signed char vec_xor (vector signed char, vector bool char);
8118 vector signed char vec_xor (vector signed char, vector signed char);
8119 vector unsigned char vec_xor (vector bool char, vector unsigned char);
8120 vector unsigned char vec_xor (vector unsigned char, vector bool char);
8121 vector unsigned char vec_xor (vector unsigned char,
8122 vector unsigned char);
8124 int vec_all_eq (vector signed char, vector bool char);
8125 int vec_all_eq (vector signed char, vector signed char);
8126 int vec_all_eq (vector unsigned char, vector bool char);
8127 int vec_all_eq (vector unsigned char, vector unsigned char);
8128 int vec_all_eq (vector bool char, vector bool char);
8129 int vec_all_eq (vector bool char, vector unsigned char);
8130 int vec_all_eq (vector bool char, vector signed char);
8131 int vec_all_eq (vector signed short, vector bool short);
8132 int vec_all_eq (vector signed short, vector signed short);
8133 int vec_all_eq (vector unsigned short, vector bool short);
8134 int vec_all_eq (vector unsigned short, vector unsigned short);
8135 int vec_all_eq (vector bool short, vector bool short);
8136 int vec_all_eq (vector bool short, vector unsigned short);
8137 int vec_all_eq (vector bool short, vector signed short);
8138 int vec_all_eq (vector pixel, vector pixel);
8139 int vec_all_eq (vector signed int, vector bool int);
8140 int vec_all_eq (vector signed int, vector signed int);
8141 int vec_all_eq (vector unsigned int, vector bool int);
8142 int vec_all_eq (vector unsigned int, vector unsigned int);
8143 int vec_all_eq (vector bool int, vector bool int);
8144 int vec_all_eq (vector bool int, vector unsigned int);
8145 int vec_all_eq (vector bool int, vector signed int);
8146 int vec_all_eq (vector float, vector float);
8148 int vec_all_ge (vector bool char, vector unsigned char);
8149 int vec_all_ge (vector unsigned char, vector bool char);
8150 int vec_all_ge (vector unsigned char, vector unsigned char);
8151 int vec_all_ge (vector bool char, vector signed char);
8152 int vec_all_ge (vector signed char, vector bool char);
8153 int vec_all_ge (vector signed char, vector signed char);
8154 int vec_all_ge (vector bool short, vector unsigned short);
8155 int vec_all_ge (vector unsigned short, vector bool short);
8156 int vec_all_ge (vector unsigned short, vector unsigned short);
8157 int vec_all_ge (vector signed short, vector signed short);
8158 int vec_all_ge (vector bool short, vector signed short);
8159 int vec_all_ge (vector signed short, vector bool short);
8160 int vec_all_ge (vector bool int, vector unsigned int);
8161 int vec_all_ge (vector unsigned int, vector bool int);
8162 int vec_all_ge (vector unsigned int, vector unsigned int);
8163 int vec_all_ge (vector bool int, vector signed int);
8164 int vec_all_ge (vector signed int, vector bool int);
8165 int vec_all_ge (vector signed int, vector signed int);
8166 int vec_all_ge (vector float, vector float);
8168 int vec_all_gt (vector bool char, vector unsigned char);
8169 int vec_all_gt (vector unsigned char, vector bool char);
8170 int vec_all_gt (vector unsigned char, vector unsigned char);
8171 int vec_all_gt (vector bool char, vector signed char);
8172 int vec_all_gt (vector signed char, vector bool char);
8173 int vec_all_gt (vector signed char, vector signed char);
8174 int vec_all_gt (vector bool short, vector unsigned short);
8175 int vec_all_gt (vector unsigned short, vector bool short);
8176 int vec_all_gt (vector unsigned short, vector unsigned short);
8177 int vec_all_gt (vector bool short, vector signed short);
8178 int vec_all_gt (vector signed short, vector bool short);
8179 int vec_all_gt (vector signed short, vector signed short);
8180 int vec_all_gt (vector bool int, vector unsigned int);
8181 int vec_all_gt (vector unsigned int, vector bool int);
8182 int vec_all_gt (vector unsigned int, vector unsigned int);
8183 int vec_all_gt (vector bool int, vector signed int);
8184 int vec_all_gt (vector signed int, vector bool int);
8185 int vec_all_gt (vector signed int, vector signed int);
8186 int vec_all_gt (vector float, vector float);
8188 int vec_all_in (vector float, vector float);
8190 int vec_all_le (vector bool char, vector unsigned char);
8191 int vec_all_le (vector unsigned char, vector bool char);
8192 int vec_all_le (vector unsigned char, vector unsigned char);
8193 int vec_all_le (vector bool char, vector signed char);
8194 int vec_all_le (vector signed char, vector bool char);
8195 int vec_all_le (vector signed char, vector signed char);
8196 int vec_all_le (vector bool short, vector unsigned short);
8197 int vec_all_le (vector unsigned short, vector bool short);
8198 int vec_all_le (vector unsigned short, vector unsigned short);
8199 int vec_all_le (vector bool short, vector signed short);
8200 int vec_all_le (vector signed short, vector bool short);
8201 int vec_all_le (vector signed short, vector signed short);
8202 int vec_all_le (vector bool int, vector unsigned int);
8203 int vec_all_le (vector unsigned int, vector bool int);
8204 int vec_all_le (vector unsigned int, vector unsigned int);
8205 int vec_all_le (vector bool int, vector signed int);
8206 int vec_all_le (vector signed int, vector bool int);
8207 int vec_all_le (vector signed int, vector signed int);
8208 int vec_all_le (vector float, vector float);
8210 int vec_all_lt (vector bool char, vector unsigned char);
8211 int vec_all_lt (vector unsigned char, vector bool char);
8212 int vec_all_lt (vector unsigned char, vector unsigned char);
8213 int vec_all_lt (vector bool char, vector signed char);
8214 int vec_all_lt (vector signed char, vector bool char);
8215 int vec_all_lt (vector signed char, vector signed char);
8216 int vec_all_lt (vector bool short, vector unsigned short);
8217 int vec_all_lt (vector unsigned short, vector bool short);
8218 int vec_all_lt (vector unsigned short, vector unsigned short);
8219 int vec_all_lt (vector bool short, vector signed short);
8220 int vec_all_lt (vector signed short, vector bool short);
8221 int vec_all_lt (vector signed short, vector signed short);
8222 int vec_all_lt (vector bool int, vector unsigned int);
8223 int vec_all_lt (vector unsigned int, vector bool int);
8224 int vec_all_lt (vector unsigned int, vector unsigned int);
8225 int vec_all_lt (vector bool int, vector signed int);
8226 int vec_all_lt (vector signed int, vector bool int);
8227 int vec_all_lt (vector signed int, vector signed int);
8228 int vec_all_lt (vector float, vector float);
8230 int vec_all_nan (vector float);
8232 int vec_all_ne (vector signed char, vector bool char);
8233 int vec_all_ne (vector signed char, vector signed char);
8234 int vec_all_ne (vector unsigned char, vector bool char);
8235 int vec_all_ne (vector unsigned char, vector unsigned char);
8236 int vec_all_ne (vector bool char, vector bool char);
8237 int vec_all_ne (vector bool char, vector unsigned char);
8238 int vec_all_ne (vector bool char, vector signed char);
8239 int vec_all_ne (vector signed short, vector bool short);
8240 int vec_all_ne (vector signed short, vector signed short);
8241 int vec_all_ne (vector unsigned short, vector bool short);
8242 int vec_all_ne (vector unsigned short, vector unsigned short);
8243 int vec_all_ne (vector bool short, vector bool short);
8244 int vec_all_ne (vector bool short, vector unsigned short);
8245 int vec_all_ne (vector bool short, vector signed short);
8246 int vec_all_ne (vector pixel, vector pixel);
8247 int vec_all_ne (vector signed int, vector bool int);
8248 int vec_all_ne (vector signed int, vector signed int);
8249 int vec_all_ne (vector unsigned int, vector bool int);
8250 int vec_all_ne (vector unsigned int, vector unsigned int);
8251 int vec_all_ne (vector bool int, vector bool int);
8252 int vec_all_ne (vector bool int, vector unsigned int);
8253 int vec_all_ne (vector bool int, vector signed int);
8254 int vec_all_ne (vector float, vector float);
8256 int vec_all_nge (vector float, vector float);
8258 int vec_all_ngt (vector float, vector float);
8260 int vec_all_nle (vector float, vector float);
8262 int vec_all_nlt (vector float, vector float);
8264 int vec_all_numeric (vector float);
8266 int vec_any_eq (vector signed char, vector bool char);
8267 int vec_any_eq (vector signed char, vector signed char);
8268 int vec_any_eq (vector unsigned char, vector bool char);
8269 int vec_any_eq (vector unsigned char, vector unsigned char);
8270 int vec_any_eq (vector bool char, vector bool char);
8271 int vec_any_eq (vector bool char, vector unsigned char);
8272 int vec_any_eq (vector bool char, vector signed char);
8273 int vec_any_eq (vector signed short, vector bool short);
8274 int vec_any_eq (vector signed short, vector signed short);
8275 int vec_any_eq (vector unsigned short, vector bool short);
8276 int vec_any_eq (vector unsigned short, vector unsigned short);
8277 int vec_any_eq (vector bool short, vector bool short);
8278 int vec_any_eq (vector bool short, vector unsigned short);
8279 int vec_any_eq (vector bool short, vector signed short);
8280 int vec_any_eq (vector pixel, vector pixel);
8281 int vec_any_eq (vector signed int, vector bool int);
8282 int vec_any_eq (vector signed int, vector signed int);
8283 int vec_any_eq (vector unsigned int, vector bool int);
8284 int vec_any_eq (vector unsigned int, vector unsigned int);
8285 int vec_any_eq (vector bool int, vector bool int);
8286 int vec_any_eq (vector bool int, vector unsigned int);
8287 int vec_any_eq (vector bool int, vector signed int);
8288 int vec_any_eq (vector float, vector float);
8290 int vec_any_ge (vector signed char, vector bool char);
8291 int vec_any_ge (vector unsigned char, vector bool char);
8292 int vec_any_ge (vector unsigned char, vector unsigned char);
8293 int vec_any_ge (vector signed char, vector signed char);
8294 int vec_any_ge (vector bool char, vector unsigned char);
8295 int vec_any_ge (vector bool char, vector signed char);
8296 int vec_any_ge (vector unsigned short, vector bool short);
8297 int vec_any_ge (vector unsigned short, vector unsigned short);
8298 int vec_any_ge (vector signed short, vector signed short);
8299 int vec_any_ge (vector signed short, vector bool short);
8300 int vec_any_ge (vector bool short, vector unsigned short);
8301 int vec_any_ge (vector bool short, vector signed short);
8302 int vec_any_ge (vector signed int, vector bool int);
8303 int vec_any_ge (vector unsigned int, vector bool int);
8304 int vec_any_ge (vector unsigned int, vector unsigned int);
8305 int vec_any_ge (vector signed int, vector signed int);
8306 int vec_any_ge (vector bool int, vector unsigned int);
8307 int vec_any_ge (vector bool int, vector signed int);
8308 int vec_any_ge (vector float, vector float);
8310 int vec_any_gt (vector bool char, vector unsigned char);
8311 int vec_any_gt (vector unsigned char, vector bool char);
8312 int vec_any_gt (vector unsigned char, vector unsigned char);
8313 int vec_any_gt (vector bool char, vector signed char);
8314 int vec_any_gt (vector signed char, vector bool char);
8315 int vec_any_gt (vector signed char, vector signed char);
8316 int vec_any_gt (vector bool short, vector unsigned short);
8317 int vec_any_gt (vector unsigned short, vector bool short);
8318 int vec_any_gt (vector unsigned short, vector unsigned short);
8319 int vec_any_gt (vector bool short, vector signed short);
8320 int vec_any_gt (vector signed short, vector bool short);
8321 int vec_any_gt (vector signed short, vector signed short);
8322 int vec_any_gt (vector bool int, vector unsigned int);
8323 int vec_any_gt (vector unsigned int, vector bool int);
8324 int vec_any_gt (vector unsigned int, vector unsigned int);
8325 int vec_any_gt (vector bool int, vector signed int);
8326 int vec_any_gt (vector signed int, vector bool int);
8327 int vec_any_gt (vector signed int, vector signed int);
8328 int vec_any_gt (vector float, vector float);
8330 int vec_any_le (vector bool char, vector unsigned char);
8331 int vec_any_le (vector unsigned char, vector bool char);
8332 int vec_any_le (vector unsigned char, vector unsigned char);
8333 int vec_any_le (vector bool char, vector signed char);
8334 int vec_any_le (vector signed char, vector bool char);
8335 int vec_any_le (vector signed char, vector signed char);
8336 int vec_any_le (vector bool short, vector unsigned short);
8337 int vec_any_le (vector unsigned short, vector bool short);
8338 int vec_any_le (vector unsigned short, vector unsigned short);
8339 int vec_any_le (vector bool short, vector signed short);
8340 int vec_any_le (vector signed short, vector bool short);
8341 int vec_any_le (vector signed short, vector signed short);
8342 int vec_any_le (vector bool int, vector unsigned int);
8343 int vec_any_le (vector unsigned int, vector bool int);
8344 int vec_any_le (vector unsigned int, vector unsigned int);
8345 int vec_any_le (vector bool int, vector signed int);
8346 int vec_any_le (vector signed int, vector bool int);
8347 int vec_any_le (vector signed int, vector signed int);
8348 int vec_any_le (vector float, vector float);
8350 int vec_any_lt (vector bool char, vector unsigned char);
8351 int vec_any_lt (vector unsigned char, vector bool char);
8352 int vec_any_lt (vector unsigned char, vector unsigned char);
8353 int vec_any_lt (vector bool char, vector signed char);
8354 int vec_any_lt (vector signed char, vector bool char);
8355 int vec_any_lt (vector signed char, vector signed char);
8356 int vec_any_lt (vector bool short, vector unsigned short);
8357 int vec_any_lt (vector unsigned short, vector bool short);
8358 int vec_any_lt (vector unsigned short, vector unsigned short);
8359 int vec_any_lt (vector bool short, vector signed short);
8360 int vec_any_lt (vector signed short, vector bool short);
8361 int vec_any_lt (vector signed short, vector signed short);
8362 int vec_any_lt (vector bool int, vector unsigned int);
8363 int vec_any_lt (vector unsigned int, vector bool int);
8364 int vec_any_lt (vector unsigned int, vector unsigned int);
8365 int vec_any_lt (vector bool int, vector signed int);
8366 int vec_any_lt (vector signed int, vector bool int);
8367 int vec_any_lt (vector signed int, vector signed int);
8368 int vec_any_lt (vector float, vector float);
8370 int vec_any_nan (vector float);
8372 int vec_any_ne (vector signed char, vector bool char);
8373 int vec_any_ne (vector signed char, vector signed char);
8374 int vec_any_ne (vector unsigned char, vector bool char);
8375 int vec_any_ne (vector unsigned char, vector unsigned char);
8376 int vec_any_ne (vector bool char, vector bool char);
8377 int vec_any_ne (vector bool char, vector unsigned char);
8378 int vec_any_ne (vector bool char, vector signed char);
8379 int vec_any_ne (vector signed short, vector bool short);
8380 int vec_any_ne (vector signed short, vector signed short);
8381 int vec_any_ne (vector unsigned short, vector bool short);
8382 int vec_any_ne (vector unsigned short, vector unsigned short);
8383 int vec_any_ne (vector bool short, vector bool short);
8384 int vec_any_ne (vector bool short, vector unsigned short);
8385 int vec_any_ne (vector bool short, vector signed short);
8386 int vec_any_ne (vector pixel, vector pixel);
8387 int vec_any_ne (vector signed int, vector bool int);
8388 int vec_any_ne (vector signed int, vector signed int);
8389 int vec_any_ne (vector unsigned int, vector bool int);
8390 int vec_any_ne (vector unsigned int, vector unsigned int);
8391 int vec_any_ne (vector bool int, vector bool int);
8392 int vec_any_ne (vector bool int, vector unsigned int);
8393 int vec_any_ne (vector bool int, vector signed int);
8394 int vec_any_ne (vector float, vector float);
8396 int vec_any_nge (vector float, vector float);
8398 int vec_any_ngt (vector float, vector float);
8400 int vec_any_nle (vector float, vector float);
8402 int vec_any_nlt (vector float, vector float);
8404 int vec_any_numeric (vector float);
8406 int vec_any_out (vector float, vector float);
8409 @node Target Format Checks
8410 @section Format Checks Specific to Particular Target Machines
8412 For some target machines, GCC supports additional options to the
8414 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
8417 * Solaris Format Checks::
8420 @node Solaris Format Checks
8421 @subsection Solaris Format Checks
8423 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
8424 check. @code{cmn_err} accepts a subset of the standard @code{printf}
8425 conversions, and the two-argument @code{%b} conversion for displaying
8426 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
8429 @section Pragmas Accepted by GCC
8433 GCC supports several types of pragmas, primarily in order to compile
8434 code originally written for other compilers. Note that in general
8435 we do not recommend the use of pragmas; @xref{Function Attributes},
8436 for further explanation.
8440 * RS/6000 and PowerPC Pragmas::
8443 * Symbol-Renaming Pragmas::
8444 * Structure-Packing Pragmas::
8448 @subsection ARM Pragmas
8450 The ARM target defines pragmas for controlling the default addition of
8451 @code{long_call} and @code{short_call} attributes to functions.
8452 @xref{Function Attributes}, for information about the effects of these
8457 @cindex pragma, long_calls
8458 Set all subsequent functions to have the @code{long_call} attribute.
8461 @cindex pragma, no_long_calls
8462 Set all subsequent functions to have the @code{short_call} attribute.
8464 @item long_calls_off
8465 @cindex pragma, long_calls_off
8466 Do not affect the @code{long_call} or @code{short_call} attributes of
8467 subsequent functions.
8470 @node RS/6000 and PowerPC Pragmas
8471 @subsection RS/6000 and PowerPC Pragmas
8473 The RS/6000 and PowerPC targets define one pragma for controlling
8474 whether or not the @code{longcall} attribute is added to function
8475 declarations by default. This pragma overrides the @option{-mlongcall}
8476 option, but not the @code{longcall} and @code{shortcall} attributes.
8477 @xref{RS/6000 and PowerPC Options}, for more information about when long
8478 calls are and are not necessary.
8482 @cindex pragma, longcall
8483 Apply the @code{longcall} attribute to all subsequent function
8487 Do not apply the @code{longcall} attribute to subsequent function
8491 @c Describe c4x pragmas here.
8492 @c Describe h8300 pragmas here.
8493 @c Describe sh pragmas here.
8494 @c Describe v850 pragmas here.
8496 @node Darwin Pragmas
8497 @subsection Darwin Pragmas
8499 The following pragmas are available for all architectures running the
8500 Darwin operating system. These are useful for compatibility with other
8504 @item mark @var{tokens}@dots{}
8505 @cindex pragma, mark
8506 This pragma is accepted, but has no effect.
8508 @item options align=@var{alignment}
8509 @cindex pragma, options align
8510 This pragma sets the alignment of fields in structures. The values of
8511 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
8512 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
8513 properly; to restore the previous setting, use @code{reset} for the
8516 @item segment @var{tokens}@dots{}
8517 @cindex pragma, segment
8518 This pragma is accepted, but has no effect.
8520 @item unused (@var{var} [, @var{var}]@dots{})
8521 @cindex pragma, unused
8522 This pragma declares variables to be possibly unused. GCC will not
8523 produce warnings for the listed variables. The effect is similar to
8524 that of the @code{unused} attribute, except that this pragma may appear
8525 anywhere within the variables' scopes.
8528 @node Solaris Pragmas
8529 @subsection Solaris Pragmas
8531 The Solaris target supports @code{#pragma redefine_extname}
8532 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
8533 @code{#pragma} directives for compatibility with the system compiler.
8536 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
8537 @cindex pragma, align
8539 Increase the minimum alignment of each @var{variable} to @var{alignment}.
8540 This is the same as GCC's @code{aligned} attribute @pxref{Variable
8543 @item fini (@var{function} [, @var{function}]...)
8544 @cindex pragma, fini
8546 This pragma causes each listed @var{function} to be called after
8547 main, or during shared module unloading, by adding a call to the
8548 @code{.fini} section.
8550 @item init (@var{function} [, @var{function}]...)
8551 @cindex pragma, init
8553 This pragma causes each listed @var{function} to be called during
8554 initialization (before @code{main}) or during shared module loading, by
8555 adding a call to the @code{.init} section.
8559 @node Symbol-Renaming Pragmas
8560 @subsection Symbol-Renaming Pragmas
8562 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
8563 supports two @code{#pragma} directives which change the name used in
8564 assembly for a given declaration. These pragmas are only available on
8565 platforms whose system headers need them. To get this effect on all
8566 platforms supported by GCC, use the asm labels extension (@pxref{Asm
8570 @item redefine_extname @var{oldname} @var{newname}
8571 @cindex pragma, redefine_extname
8573 This pragma gives the C function @var{oldname} the assembly symbol
8574 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
8575 will be defined if this pragma is available (currently only on
8578 @item extern_prefix @var{string}
8579 @cindex pragma, extern_prefix
8581 This pragma causes all subsequent external function and variable
8582 declarations to have @var{string} prepended to their assembly symbols.
8583 This effect may be terminated with another @code{extern_prefix} pragma
8584 whose argument is an empty string. The preprocessor macro
8585 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
8586 available (currently only on Tru64 UNIX).
8589 These pragmas and the asm labels extension interact in a complicated
8590 manner. Here are some corner cases you may want to be aware of.
8593 @item Both pragmas silently apply only to declarations with external
8594 linkage. Asm labels do not have this restriction.
8596 @item In C++, both pragmas silently apply only to declarations with
8597 ``C'' linkage. Again, asm labels do not have this restriction.
8599 @item If any of the three ways of changing the assembly name of a
8600 declaration is applied to a declaration whose assembly name has
8601 already been determined (either by a previous use of one of these
8602 features, or because the compiler needed the assembly name in order to
8603 generate code), and the new name is different, a warning issues and
8604 the name does not change.
8606 @item The @var{oldname} used by @code{#pragma redefine_extname} is
8607 always the C-language name.
8609 @item If @code{#pragma extern_prefix} is in effect, and a declaration
8610 occurs with an asm label attached, the prefix is silently ignored for
8613 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
8614 apply to the same declaration, whichever triggered first wins, and a
8615 warning issues if they contradict each other. (We would like to have
8616 @code{#pragma redefine_extname} always win, for consistency with asm
8617 labels, but if @code{#pragma extern_prefix} triggers first we have no
8618 way of knowing that that happened.)
8621 @node Structure-Packing Pragmas
8622 @subsection Structure-Packing Pragmas
8624 For compatibility with Win32, GCC supports as set of @code{#pragma}
8625 directives which change the maximum alignment of members of structures,
8626 unions, and classes subsequently defined. The @var{n} value below always
8627 is required to be a small power of two and specifies the new alignment
8631 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
8632 @item @code{#pragma pack()} sets the alignment to the one that was in
8633 effect when compilation started (see also command line option
8634 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
8635 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
8636 setting on an internal stack and then optionally sets the new alignment.
8637 @item @code{#pragma pack(pop)} restores the alignment setting to the one
8638 saved at the top of the internal stack (and removes that stack entry).
8639 Note that @code{#pragma pack([@var{n}])} does not influence this internal
8640 stack; thus it is possible to have @code{#pragma pack(push)} followed by
8641 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
8642 @code{#pragma pack(pop)}.
8645 @node Unnamed Fields
8646 @section Unnamed struct/union fields within structs/unions
8650 For compatibility with other compilers, GCC allows you to define
8651 a structure or union that contains, as fields, structures and unions
8652 without names. For example:
8665 In this example, the user would be able to access members of the unnamed
8666 union with code like @samp{foo.b}. Note that only unnamed structs and
8667 unions are allowed, you may not have, for example, an unnamed
8670 You must never create such structures that cause ambiguous field definitions.
8671 For example, this structure:
8682 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
8683 Such constructs are not supported and must be avoided. In the future,
8684 such constructs may be detected and treated as compilation errors.
8686 @opindex fms-extensions
8687 Unless @option{-fms-extensions} is used, the unnamed field must be a
8688 structure or union definition without a tag (for example, @samp{struct
8689 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
8690 also be a definition with a tag such as @samp{struct foo @{ int a;
8691 @};}, a reference to a previously defined structure or union such as
8692 @samp{struct foo;}, or a reference to a @code{typedef} name for a
8693 previously defined structure or union type.
8696 @section Thread-Local Storage
8697 @cindex Thread-Local Storage
8698 @cindex @acronym{TLS}
8701 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
8702 are allocated such that there is one instance of the variable per extant
8703 thread. The run-time model GCC uses to implement this originates
8704 in the IA-64 processor-specific ABI, but has since been migrated
8705 to other processors as well. It requires significant support from
8706 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
8707 system libraries (@file{libc.so} and @file{libpthread.so}), so it
8708 is not available everywhere.
8710 At the user level, the extension is visible with a new storage
8711 class keyword: @code{__thread}. For example:
8715 extern __thread struct state s;
8716 static __thread char *p;
8719 The @code{__thread} specifier may be used alone, with the @code{extern}
8720 or @code{static} specifiers, but with no other storage class specifier.
8721 When used with @code{extern} or @code{static}, @code{__thread} must appear
8722 immediately after the other storage class specifier.
8724 The @code{__thread} specifier may be applied to any global, file-scoped
8725 static, function-scoped static, or static data member of a class. It may
8726 not be applied to block-scoped automatic or non-static data member.
8728 When the address-of operator is applied to a thread-local variable, it is
8729 evaluated at run-time and returns the address of the current thread's
8730 instance of that variable. An address so obtained may be used by any
8731 thread. When a thread terminates, any pointers to thread-local variables
8732 in that thread become invalid.
8734 No static initialization may refer to the address of a thread-local variable.
8736 In C++, if an initializer is present for a thread-local variable, it must
8737 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
8740 See @uref{http://people.redhat.com/drepper/tls.pdf,
8741 ELF Handling For Thread-Local Storage} for a detailed explanation of
8742 the four thread-local storage addressing models, and how the run-time
8743 is expected to function.
8746 * C99 Thread-Local Edits::
8747 * C++98 Thread-Local Edits::
8750 @node C99 Thread-Local Edits
8751 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
8753 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
8754 that document the exact semantics of the language extension.
8758 @cite{5.1.2 Execution environments}
8760 Add new text after paragraph 1
8763 Within either execution environment, a @dfn{thread} is a flow of
8764 control within a program. It is implementation defined whether
8765 or not there may be more than one thread associated with a program.
8766 It is implementation defined how threads beyond the first are
8767 created, the name and type of the function called at thread
8768 startup, and how threads may be terminated. However, objects
8769 with thread storage duration shall be initialized before thread
8774 @cite{6.2.4 Storage durations of objects}
8776 Add new text before paragraph 3
8779 An object whose identifier is declared with the storage-class
8780 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
8781 Its lifetime is the entire execution of the thread, and its
8782 stored value is initialized only once, prior to thread startup.
8786 @cite{6.4.1 Keywords}
8788 Add @code{__thread}.
8791 @cite{6.7.1 Storage-class specifiers}
8793 Add @code{__thread} to the list of storage class specifiers in
8796 Change paragraph 2 to
8799 With the exception of @code{__thread}, at most one storage-class
8800 specifier may be given [@dots{}]. The @code{__thread} specifier may
8801 be used alone, or immediately following @code{extern} or
8805 Add new text after paragraph 6
8808 The declaration of an identifier for a variable that has
8809 block scope that specifies @code{__thread} shall also
8810 specify either @code{extern} or @code{static}.
8812 The @code{__thread} specifier shall be used only with
8817 @node C++98 Thread-Local Edits
8818 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
8820 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
8821 that document the exact semantics of the language extension.
8825 @b{[intro.execution]}
8827 New text after paragraph 4
8830 A @dfn{thread} is a flow of control within the abstract machine.
8831 It is implementation defined whether or not there may be more than
8835 New text after paragraph 7
8838 It is unspecified whether additional action must be taken to
8839 ensure when and whether side effects are visible to other threads.
8845 Add @code{__thread}.
8848 @b{[basic.start.main]}
8850 Add after paragraph 5
8853 The thread that begins execution at the @code{main} function is called
8854 the @dfn{main thread}. It is implementation defined how functions
8855 beginning threads other than the main thread are designated or typed.
8856 A function so designated, as well as the @code{main} function, is called
8857 a @dfn{thread startup function}. It is implementation defined what
8858 happens if a thread startup function returns. It is implementation
8859 defined what happens to other threads when any thread calls @code{exit}.
8863 @b{[basic.start.init]}
8865 Add after paragraph 4
8868 The storage for an object of thread storage duration shall be
8869 statically initialized before the first statement of the thread startup
8870 function. An object of thread storage duration shall not require
8871 dynamic initialization.
8875 @b{[basic.start.term]}
8877 Add after paragraph 3
8880 The type of an object with thread storage duration shall not have a
8881 non-trivial destructor, nor shall it be an array type whose elements
8882 (directly or indirectly) have non-trivial destructors.
8888 Add ``thread storage duration'' to the list in paragraph 1.
8893 Thread, static, and automatic storage durations are associated with
8894 objects introduced by declarations [@dots{}].
8897 Add @code{__thread} to the list of specifiers in paragraph 3.
8900 @b{[basic.stc.thread]}
8902 New section before @b{[basic.stc.static]}
8905 The keyword @code{__thread} applied to a non-local object gives the
8906 object thread storage duration.
8908 A local variable or class data member declared both @code{static}
8909 and @code{__thread} gives the variable or member thread storage
8914 @b{[basic.stc.static]}
8919 All objects which have neither thread storage duration, dynamic
8920 storage duration nor are local [@dots{}].
8926 Add @code{__thread} to the list in paragraph 1.
8931 With the exception of @code{__thread}, at most one
8932 @var{storage-class-specifier} shall appear in a given
8933 @var{decl-specifier-seq}. The @code{__thread} specifier may
8934 be used alone, or immediately following the @code{extern} or
8935 @code{static} specifiers. [@dots{}]
8938 Add after paragraph 5
8941 The @code{__thread} specifier can be applied only to the names of objects
8942 and to anonymous unions.
8948 Add after paragraph 6
8951 Non-@code{static} members shall not be @code{__thread}.
8955 @node C++ Extensions
8956 @chapter Extensions to the C++ Language
8957 @cindex extensions, C++ language
8958 @cindex C++ language extensions
8960 The GNU compiler provides these extensions to the C++ language (and you
8961 can also use most of the C language extensions in your C++ programs). If you
8962 want to write code that checks whether these features are available, you can
8963 test for the GNU compiler the same way as for C programs: check for a
8964 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
8965 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
8966 Predefined Macros,cpp,The GNU C Preprocessor}).
8969 * Min and Max:: C++ Minimum and maximum operators.
8970 * Volatiles:: What constitutes an access to a volatile object.
8971 * Restricted Pointers:: C99 restricted pointers and references.
8972 * Vague Linkage:: Where G++ puts inlines, vtables and such.
8973 * C++ Interface:: You can use a single C++ header file for both
8974 declarations and definitions.
8975 * Template Instantiation:: Methods for ensuring that exactly one copy of
8976 each needed template instantiation is emitted.
8977 * Bound member functions:: You can extract a function pointer to the
8978 method denoted by a @samp{->*} or @samp{.*} expression.
8979 * C++ Attributes:: Variable, function, and type attributes for C++ only.
8980 * Strong Using:: Strong using-directives for namespace composition.
8981 * Java Exceptions:: Tweaking exception handling to work with Java.
8982 * Deprecated Features:: Things will disappear from g++.
8983 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
8987 @section Minimum and Maximum Operators in C++
8989 It is very convenient to have operators which return the ``minimum'' or the
8990 ``maximum'' of two arguments. In GNU C++ (but not in GNU C),
8993 @item @var{a} <? @var{b}
8995 @cindex minimum operator
8996 is the @dfn{minimum}, returning the smaller of the numeric values
8997 @var{a} and @var{b};
8999 @item @var{a} >? @var{b}
9001 @cindex maximum operator
9002 is the @dfn{maximum}, returning the larger of the numeric values @var{a}
9006 These operations are not primitive in ordinary C++, since you can
9007 use a macro to return the minimum of two things in C++, as in the
9011 #define MIN(X,Y) ((X) < (Y) ? : (X) : (Y))
9015 You might then use @w{@samp{int min = MIN (i, j);}} to set @var{min} to
9016 the minimum value of variables @var{i} and @var{j}.
9018 However, side effects in @code{X} or @code{Y} may cause unintended
9019 behavior. For example, @code{MIN (i++, j++)} will fail, incrementing
9020 the smaller counter twice. The GNU C @code{typeof} extension allows you
9021 to write safe macros that avoid this kind of problem (@pxref{Typeof}).
9022 However, writing @code{MIN} and @code{MAX} as macros also forces you to
9023 use function-call notation for a fundamental arithmetic operation.
9024 Using GNU C++ extensions, you can write @w{@samp{int min = i <? j;}}
9027 Since @code{<?} and @code{>?} are built into the compiler, they properly
9028 handle expressions with side-effects; @w{@samp{int min = i++ <? j++;}}
9032 @section When is a Volatile Object Accessed?
9033 @cindex accessing volatiles
9034 @cindex volatile read
9035 @cindex volatile write
9036 @cindex volatile access
9038 Both the C and C++ standard have the concept of volatile objects. These
9039 are normally accessed by pointers and used for accessing hardware. The
9040 standards encourage compilers to refrain from optimizations
9041 concerning accesses to volatile objects that it might perform on
9042 non-volatile objects. The C standard leaves it implementation defined
9043 as to what constitutes a volatile access. The C++ standard omits to
9044 specify this, except to say that C++ should behave in a similar manner
9045 to C with respect to volatiles, where possible. The minimum either
9046 standard specifies is that at a sequence point all previous accesses to
9047 volatile objects have stabilized and no subsequent accesses have
9048 occurred. Thus an implementation is free to reorder and combine
9049 volatile accesses which occur between sequence points, but cannot do so
9050 for accesses across a sequence point. The use of volatiles does not
9051 allow you to violate the restriction on updating objects multiple times
9052 within a sequence point.
9054 In most expressions, it is intuitively obvious what is a read and what is
9055 a write. For instance
9058 volatile int *dst = @var{somevalue};
9059 volatile int *src = @var{someothervalue};
9064 will cause a read of the volatile object pointed to by @var{src} and stores the
9065 value into the volatile object pointed to by @var{dst}. There is no
9066 guarantee that these reads and writes are atomic, especially for objects
9067 larger than @code{int}.
9069 Less obvious expressions are where something which looks like an access
9070 is used in a void context. An example would be,
9073 volatile int *src = @var{somevalue};
9077 With C, such expressions are rvalues, and as rvalues cause a read of
9078 the object, GCC interprets this as a read of the volatile being pointed
9079 to. The C++ standard specifies that such expressions do not undergo
9080 lvalue to rvalue conversion, and that the type of the dereferenced
9081 object may be incomplete. The C++ standard does not specify explicitly
9082 that it is this lvalue to rvalue conversion which is responsible for
9083 causing an access. However, there is reason to believe that it is,
9084 because otherwise certain simple expressions become undefined. However,
9085 because it would surprise most programmers, G++ treats dereferencing a
9086 pointer to volatile object of complete type in a void context as a read
9087 of the object. When the object has incomplete type, G++ issues a
9092 struct T @{int m;@};
9093 volatile S *ptr1 = @var{somevalue};
9094 volatile T *ptr2 = @var{somevalue};
9099 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
9100 causes a read of the object pointed to. If you wish to force an error on
9101 the first case, you must force a conversion to rvalue with, for instance
9102 a static cast, @code{static_cast<S>(*ptr1)}.
9104 When using a reference to volatile, G++ does not treat equivalent
9105 expressions as accesses to volatiles, but instead issues a warning that
9106 no volatile is accessed. The rationale for this is that otherwise it
9107 becomes difficult to determine where volatile access occur, and not
9108 possible to ignore the return value from functions returning volatile
9109 references. Again, if you wish to force a read, cast the reference to
9112 @node Restricted Pointers
9113 @section Restricting Pointer Aliasing
9114 @cindex restricted pointers
9115 @cindex restricted references
9116 @cindex restricted this pointer
9118 As with the C front end, G++ understands the C99 feature of restricted pointers,
9119 specified with the @code{__restrict__}, or @code{__restrict} type
9120 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
9121 language flag, @code{restrict} is not a keyword in C++.
9123 In addition to allowing restricted pointers, you can specify restricted
9124 references, which indicate that the reference is not aliased in the local
9128 void fn (int *__restrict__ rptr, int &__restrict__ rref)
9135 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
9136 @var{rref} refers to a (different) unaliased integer.
9138 You may also specify whether a member function's @var{this} pointer is
9139 unaliased by using @code{__restrict__} as a member function qualifier.
9142 void T::fn () __restrict__
9149 Within the body of @code{T::fn}, @var{this} will have the effective
9150 definition @code{T *__restrict__ const this}. Notice that the
9151 interpretation of a @code{__restrict__} member function qualifier is
9152 different to that of @code{const} or @code{volatile} qualifier, in that it
9153 is applied to the pointer rather than the object. This is consistent with
9154 other compilers which implement restricted pointers.
9156 As with all outermost parameter qualifiers, @code{__restrict__} is
9157 ignored in function definition matching. This means you only need to
9158 specify @code{__restrict__} in a function definition, rather than
9159 in a function prototype as well.
9162 @section Vague Linkage
9163 @cindex vague linkage
9165 There are several constructs in C++ which require space in the object
9166 file but are not clearly tied to a single translation unit. We say that
9167 these constructs have ``vague linkage''. Typically such constructs are
9168 emitted wherever they are needed, though sometimes we can be more
9172 @item Inline Functions
9173 Inline functions are typically defined in a header file which can be
9174 included in many different compilations. Hopefully they can usually be
9175 inlined, but sometimes an out-of-line copy is necessary, if the address
9176 of the function is taken or if inlining fails. In general, we emit an
9177 out-of-line copy in all translation units where one is needed. As an
9178 exception, we only emit inline virtual functions with the vtable, since
9179 it will always require a copy.
9181 Local static variables and string constants used in an inline function
9182 are also considered to have vague linkage, since they must be shared
9183 between all inlined and out-of-line instances of the function.
9187 C++ virtual functions are implemented in most compilers using a lookup
9188 table, known as a vtable. The vtable contains pointers to the virtual
9189 functions provided by a class, and each object of the class contains a
9190 pointer to its vtable (or vtables, in some multiple-inheritance
9191 situations). If the class declares any non-inline, non-pure virtual
9192 functions, the first one is chosen as the ``key method'' for the class,
9193 and the vtable is only emitted in the translation unit where the key
9196 @emph{Note:} If the chosen key method is later defined as inline, the
9197 vtable will still be emitted in every translation unit which defines it.
9198 Make sure that any inline virtuals are declared inline in the class
9199 body, even if they are not defined there.
9201 @item type_info objects
9204 C++ requires information about types to be written out in order to
9205 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
9206 For polymorphic classes (classes with virtual functions), the type_info
9207 object is written out along with the vtable so that @samp{dynamic_cast}
9208 can determine the dynamic type of a class object at runtime. For all
9209 other types, we write out the type_info object when it is used: when
9210 applying @samp{typeid} to an expression, throwing an object, or
9211 referring to a type in a catch clause or exception specification.
9213 @item Template Instantiations
9214 Most everything in this section also applies to template instantiations,
9215 but there are other options as well.
9216 @xref{Template Instantiation,,Where's the Template?}.
9220 When used with GNU ld version 2.8 or later on an ELF system such as
9221 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
9222 these constructs will be discarded at link time. This is known as
9225 On targets that don't support COMDAT, but do support weak symbols, GCC
9226 will use them. This way one copy will override all the others, but
9227 the unused copies will still take up space in the executable.
9229 For targets which do not support either COMDAT or weak symbols,
9230 most entities with vague linkage will be emitted as local symbols to
9231 avoid duplicate definition errors from the linker. This will not happen
9232 for local statics in inlines, however, as having multiple copies will
9233 almost certainly break things.
9235 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
9236 another way to control placement of these constructs.
9239 @section #pragma interface and implementation
9241 @cindex interface and implementation headers, C++
9242 @cindex C++ interface and implementation headers
9243 @cindex pragmas, interface and implementation
9245 @code{#pragma interface} and @code{#pragma implementation} provide the
9246 user with a way of explicitly directing the compiler to emit entities
9247 with vague linkage (and debugging information) in a particular
9250 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
9251 most cases, because of COMDAT support and the ``key method'' heuristic
9252 mentioned in @ref{Vague Linkage}. Using them can actually cause your
9253 program to grow due to unnecessary out-of-line copies of inline
9254 functions. Currently (3.4) the only benefit of these
9255 @code{#pragma}s is reduced duplication of debugging information, and
9256 that should be addressed soon on DWARF 2 targets with the use of
9260 @item #pragma interface
9261 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
9262 @kindex #pragma interface
9263 Use this directive in @emph{header files} that define object classes, to save
9264 space in most of the object files that use those classes. Normally,
9265 local copies of certain information (backup copies of inline member
9266 functions, debugging information, and the internal tables that implement
9267 virtual functions) must be kept in each object file that includes class
9268 definitions. You can use this pragma to avoid such duplication. When a
9269 header file containing @samp{#pragma interface} is included in a
9270 compilation, this auxiliary information will not be generated (unless
9271 the main input source file itself uses @samp{#pragma implementation}).
9272 Instead, the object files will contain references to be resolved at link
9275 The second form of this directive is useful for the case where you have
9276 multiple headers with the same name in different directories. If you
9277 use this form, you must specify the same string to @samp{#pragma
9280 @item #pragma implementation
9281 @itemx #pragma implementation "@var{objects}.h"
9282 @kindex #pragma implementation
9283 Use this pragma in a @emph{main input file}, when you want full output from
9284 included header files to be generated (and made globally visible). The
9285 included header file, in turn, should use @samp{#pragma interface}.
9286 Backup copies of inline member functions, debugging information, and the
9287 internal tables used to implement virtual functions are all generated in
9288 implementation files.
9290 @cindex implied @code{#pragma implementation}
9291 @cindex @code{#pragma implementation}, implied
9292 @cindex naming convention, implementation headers
9293 If you use @samp{#pragma implementation} with no argument, it applies to
9294 an include file with the same basename@footnote{A file's @dfn{basename}
9295 was the name stripped of all leading path information and of trailing
9296 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
9297 file. For example, in @file{allclass.cc}, giving just
9298 @samp{#pragma implementation}
9299 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
9301 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
9302 an implementation file whenever you would include it from
9303 @file{allclass.cc} even if you never specified @samp{#pragma
9304 implementation}. This was deemed to be more trouble than it was worth,
9305 however, and disabled.
9307 Use the string argument if you want a single implementation file to
9308 include code from multiple header files. (You must also use
9309 @samp{#include} to include the header file; @samp{#pragma
9310 implementation} only specifies how to use the file---it doesn't actually
9313 There is no way to split up the contents of a single header file into
9314 multiple implementation files.
9317 @cindex inlining and C++ pragmas
9318 @cindex C++ pragmas, effect on inlining
9319 @cindex pragmas in C++, effect on inlining
9320 @samp{#pragma implementation} and @samp{#pragma interface} also have an
9321 effect on function inlining.
9323 If you define a class in a header file marked with @samp{#pragma
9324 interface}, the effect on an inline function defined in that class is
9325 similar to an explicit @code{extern} declaration---the compiler emits
9326 no code at all to define an independent version of the function. Its
9327 definition is used only for inlining with its callers.
9329 @opindex fno-implement-inlines
9330 Conversely, when you include the same header file in a main source file
9331 that declares it as @samp{#pragma implementation}, the compiler emits
9332 code for the function itself; this defines a version of the function
9333 that can be found via pointers (or by callers compiled without
9334 inlining). If all calls to the function can be inlined, you can avoid
9335 emitting the function by compiling with @option{-fno-implement-inlines}.
9336 If any calls were not inlined, you will get linker errors.
9338 @node Template Instantiation
9339 @section Where's the Template?
9340 @cindex template instantiation
9342 C++ templates are the first language feature to require more
9343 intelligence from the environment than one usually finds on a UNIX
9344 system. Somehow the compiler and linker have to make sure that each
9345 template instance occurs exactly once in the executable if it is needed,
9346 and not at all otherwise. There are two basic approaches to this
9347 problem, which are referred to as the Borland model and the Cfront model.
9351 Borland C++ solved the template instantiation problem by adding the code
9352 equivalent of common blocks to their linker; the compiler emits template
9353 instances in each translation unit that uses them, and the linker
9354 collapses them together. The advantage of this model is that the linker
9355 only has to consider the object files themselves; there is no external
9356 complexity to worry about. This disadvantage is that compilation time
9357 is increased because the template code is being compiled repeatedly.
9358 Code written for this model tends to include definitions of all
9359 templates in the header file, since they must be seen to be
9363 The AT&T C++ translator, Cfront, solved the template instantiation
9364 problem by creating the notion of a template repository, an
9365 automatically maintained place where template instances are stored. A
9366 more modern version of the repository works as follows: As individual
9367 object files are built, the compiler places any template definitions and
9368 instantiations encountered in the repository. At link time, the link
9369 wrapper adds in the objects in the repository and compiles any needed
9370 instances that were not previously emitted. The advantages of this
9371 model are more optimal compilation speed and the ability to use the
9372 system linker; to implement the Borland model a compiler vendor also
9373 needs to replace the linker. The disadvantages are vastly increased
9374 complexity, and thus potential for error; for some code this can be
9375 just as transparent, but in practice it can been very difficult to build
9376 multiple programs in one directory and one program in multiple
9377 directories. Code written for this model tends to separate definitions
9378 of non-inline member templates into a separate file, which should be
9379 compiled separately.
9382 When used with GNU ld version 2.8 or later on an ELF system such as
9383 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
9384 Borland model. On other systems, G++ implements neither automatic
9387 A future version of G++ will support a hybrid model whereby the compiler
9388 will emit any instantiations for which the template definition is
9389 included in the compile, and store template definitions and
9390 instantiation context information into the object file for the rest.
9391 The link wrapper will extract that information as necessary and invoke
9392 the compiler to produce the remaining instantiations. The linker will
9393 then combine duplicate instantiations.
9395 In the mean time, you have the following options for dealing with
9396 template instantiations:
9401 Compile your template-using code with @option{-frepo}. The compiler will
9402 generate files with the extension @samp{.rpo} listing all of the
9403 template instantiations used in the corresponding object files which
9404 could be instantiated there; the link wrapper, @samp{collect2}, will
9405 then update the @samp{.rpo} files to tell the compiler where to place
9406 those instantiations and rebuild any affected object files. The
9407 link-time overhead is negligible after the first pass, as the compiler
9408 will continue to place the instantiations in the same files.
9410 This is your best option for application code written for the Borland
9411 model, as it will just work. Code written for the Cfront model will
9412 need to be modified so that the template definitions are available at
9413 one or more points of instantiation; usually this is as simple as adding
9414 @code{#include <tmethods.cc>} to the end of each template header.
9416 For library code, if you want the library to provide all of the template
9417 instantiations it needs, just try to link all of its object files
9418 together; the link will fail, but cause the instantiations to be
9419 generated as a side effect. Be warned, however, that this may cause
9420 conflicts if multiple libraries try to provide the same instantiations.
9421 For greater control, use explicit instantiation as described in the next
9425 @opindex fno-implicit-templates
9426 Compile your code with @option{-fno-implicit-templates} to disable the
9427 implicit generation of template instances, and explicitly instantiate
9428 all the ones you use. This approach requires more knowledge of exactly
9429 which instances you need than do the others, but it's less
9430 mysterious and allows greater control. You can scatter the explicit
9431 instantiations throughout your program, perhaps putting them in the
9432 translation units where the instances are used or the translation units
9433 that define the templates themselves; you can put all of the explicit
9434 instantiations you need into one big file; or you can create small files
9441 template class Foo<int>;
9442 template ostream& operator <<
9443 (ostream&, const Foo<int>&);
9446 for each of the instances you need, and create a template instantiation
9449 If you are using Cfront-model code, you can probably get away with not
9450 using @option{-fno-implicit-templates} when compiling files that don't
9451 @samp{#include} the member template definitions.
9453 If you use one big file to do the instantiations, you may want to
9454 compile it without @option{-fno-implicit-templates} so you get all of the
9455 instances required by your explicit instantiations (but not by any
9456 other files) without having to specify them as well.
9458 G++ has extended the template instantiation syntax given in the ISO
9459 standard to allow forward declaration of explicit instantiations
9460 (with @code{extern}), instantiation of the compiler support data for a
9461 template class (i.e.@: the vtable) without instantiating any of its
9462 members (with @code{inline}), and instantiation of only the static data
9463 members of a template class, without the support data or member
9464 functions (with (@code{static}):
9467 extern template int max (int, int);
9468 inline template class Foo<int>;
9469 static template class Foo<int>;
9473 Do nothing. Pretend G++ does implement automatic instantiation
9474 management. Code written for the Borland model will work fine, but
9475 each translation unit will contain instances of each of the templates it
9476 uses. In a large program, this can lead to an unacceptable amount of code
9480 @node Bound member functions
9481 @section Extracting the function pointer from a bound pointer to member function
9483 @cindex pointer to member function
9484 @cindex bound pointer to member function
9486 In C++, pointer to member functions (PMFs) are implemented using a wide
9487 pointer of sorts to handle all the possible call mechanisms; the PMF
9488 needs to store information about how to adjust the @samp{this} pointer,
9489 and if the function pointed to is virtual, where to find the vtable, and
9490 where in the vtable to look for the member function. If you are using
9491 PMFs in an inner loop, you should really reconsider that decision. If
9492 that is not an option, you can extract the pointer to the function that
9493 would be called for a given object/PMF pair and call it directly inside
9494 the inner loop, to save a bit of time.
9496 Note that you will still be paying the penalty for the call through a
9497 function pointer; on most modern architectures, such a call defeats the
9498 branch prediction features of the CPU@. This is also true of normal
9499 virtual function calls.
9501 The syntax for this extension is
9505 extern int (A::*fp)();
9506 typedef int (*fptr)(A *);
9508 fptr p = (fptr)(a.*fp);
9511 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
9512 no object is needed to obtain the address of the function. They can be
9513 converted to function pointers directly:
9516 fptr p1 = (fptr)(&A::foo);
9519 @opindex Wno-pmf-conversions
9520 You must specify @option{-Wno-pmf-conversions} to use this extension.
9522 @node C++ Attributes
9523 @section C++-Specific Variable, Function, and Type Attributes
9525 Some attributes only make sense for C++ programs.
9528 @item init_priority (@var{priority})
9529 @cindex init_priority attribute
9532 In Standard C++, objects defined at namespace scope are guaranteed to be
9533 initialized in an order in strict accordance with that of their definitions
9534 @emph{in a given translation unit}. No guarantee is made for initializations
9535 across translation units. However, GNU C++ allows users to control the
9536 order of initialization of objects defined at namespace scope with the
9537 @code{init_priority} attribute by specifying a relative @var{priority},
9538 a constant integral expression currently bounded between 101 and 65535
9539 inclusive. Lower numbers indicate a higher priority.
9541 In the following example, @code{A} would normally be created before
9542 @code{B}, but the @code{init_priority} attribute has reversed that order:
9545 Some_Class A __attribute__ ((init_priority (2000)));
9546 Some_Class B __attribute__ ((init_priority (543)));
9550 Note that the particular values of @var{priority} do not matter; only their
9553 @item java_interface
9554 @cindex java_interface attribute
9556 This type attribute informs C++ that the class is a Java interface. It may
9557 only be applied to classes declared within an @code{extern "Java"} block.
9558 Calls to methods declared in this interface will be dispatched using GCJ's
9559 interface table mechanism, instead of regular virtual table dispatch.
9563 See also @xref{Strong Using}.
9566 @section Strong Using
9568 @strong{Caution:} The semantics of this extension are not fully
9569 defined. Users should refrain from using this extension as its
9570 semantics may change subtly over time. It is possible that this
9571 extension wil be removed in future versions of G++.
9573 A using-directive with @code{__attribute ((strong))} is stronger
9574 than a normal using-directive in two ways:
9578 Templates from the used namespace can be specialized as though they were members of the using namespace.
9581 The using namespace is considered an associated namespace of all
9582 templates in the used namespace for purposes of argument-dependent
9586 This is useful for composing a namespace transparently from
9587 implementation namespaces. For example:
9592 template <class T> struct A @{ @};
9594 using namespace debug __attribute ((__strong__));
9595 template <> struct A<int> @{ @}; // @r{ok to specialize}
9597 template <class T> void f (A<T>);
9602 f (std::A<float>()); // @r{lookup finds} std::f
9607 @node Java Exceptions
9608 @section Java Exceptions
9610 The Java language uses a slightly different exception handling model
9611 from C++. Normally, GNU C++ will automatically detect when you are
9612 writing C++ code that uses Java exceptions, and handle them
9613 appropriately. However, if C++ code only needs to execute destructors
9614 when Java exceptions are thrown through it, GCC will guess incorrectly.
9615 Sample problematic code is:
9618 struct S @{ ~S(); @};
9619 extern void bar(); // @r{is written in Java, and may throw exceptions}
9628 The usual effect of an incorrect guess is a link failure, complaining of
9629 a missing routine called @samp{__gxx_personality_v0}.
9631 You can inform the compiler that Java exceptions are to be used in a
9632 translation unit, irrespective of what it might think, by writing
9633 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
9634 @samp{#pragma} must appear before any functions that throw or catch
9635 exceptions, or run destructors when exceptions are thrown through them.
9637 You cannot mix Java and C++ exceptions in the same translation unit. It
9638 is believed to be safe to throw a C++ exception from one file through
9639 another file compiled for the Java exception model, or vice versa, but
9640 there may be bugs in this area.
9642 @node Deprecated Features
9643 @section Deprecated Features
9645 In the past, the GNU C++ compiler was extended to experiment with new
9646 features, at a time when the C++ language was still evolving. Now that
9647 the C++ standard is complete, some of those features are superseded by
9648 superior alternatives. Using the old features might cause a warning in
9649 some cases that the feature will be dropped in the future. In other
9650 cases, the feature might be gone already.
9652 While the list below is not exhaustive, it documents some of the options
9653 that are now deprecated:
9656 @item -fexternal-templates
9657 @itemx -falt-external-templates
9658 These are two of the many ways for G++ to implement template
9659 instantiation. @xref{Template Instantiation}. The C++ standard clearly
9660 defines how template definitions have to be organized across
9661 implementation units. G++ has an implicit instantiation mechanism that
9662 should work just fine for standard-conforming code.
9664 @item -fstrict-prototype
9665 @itemx -fno-strict-prototype
9666 Previously it was possible to use an empty prototype parameter list to
9667 indicate an unspecified number of parameters (like C), rather than no
9668 parameters, as C++ demands. This feature has been removed, except where
9669 it is required for backwards compatibility @xref{Backwards Compatibility}.
9672 The named return value extension has been deprecated, and is now
9675 The use of initializer lists with new expressions has been deprecated,
9676 and is now removed from G++.
9678 Floating and complex non-type template parameters have been deprecated,
9679 and are now removed from G++.
9681 The implicit typename extension has been deprecated and is now
9684 The use of default arguments in function pointers, function typedefs and
9685 and other places where they are not permitted by the standard is
9686 deprecated and will be removed from a future version of G++.
9688 @node Backwards Compatibility
9689 @section Backwards Compatibility
9690 @cindex Backwards Compatibility
9691 @cindex ARM [Annotated C++ Reference Manual]
9693 Now that there is a definitive ISO standard C++, G++ has a specification
9694 to adhere to. The C++ language evolved over time, and features that
9695 used to be acceptable in previous drafts of the standard, such as the ARM
9696 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
9697 compilation of C++ written to such drafts, G++ contains some backwards
9698 compatibilities. @emph{All such backwards compatibility features are
9699 liable to disappear in future versions of G++.} They should be considered
9700 deprecated @xref{Deprecated Features}.
9704 If a variable is declared at for scope, it used to remain in scope until
9705 the end of the scope which contained the for statement (rather than just
9706 within the for scope). G++ retains this, but issues a warning, if such a
9707 variable is accessed outside the for scope.
9709 @item Implicit C language
9710 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
9711 scope to set the language. On such systems, all header files are
9712 implicitly scoped inside a C language scope. Also, an empty prototype
9713 @code{()} will be treated as an unspecified number of arguments, rather
9714 than no arguments, as C++ demands.