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 The @code{format} attribute allows you to identify your own functions
1768 which take format strings as arguments, so that GCC can check the
1769 calls to these functions for errors. The compiler always (unless
1770 @option{-ffreestanding} is used) checks formats
1771 for the standard library functions @code{printf}, @code{fprintf},
1772 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
1773 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
1774 warnings are requested (using @option{-Wformat}), so there is no need to
1775 modify the header file @file{stdio.h}. In C99 mode, the functions
1776 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
1777 @code{vsscanf} are also checked. Except in strictly conforming C
1778 standard modes, the X/Open function @code{strfmon} is also checked as
1779 are @code{printf_unlocked} and @code{fprintf_unlocked}.
1780 @xref{C Dialect Options,,Options Controlling C Dialect}.
1782 The target may provide additional types of format checks.
1783 @xref{Target Format Checks,,Format Checks Specific to Particular
1786 @item format_arg (@var{string-index})
1787 @cindex @code{format_arg} function attribute
1788 @opindex Wformat-nonliteral
1789 The @code{format_arg} attribute specifies that a function takes a format
1790 string for a @code{printf}, @code{scanf}, @code{strftime} or
1791 @code{strfmon} style function and modifies it (for example, to translate
1792 it into another language), so the result can be passed to a
1793 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
1794 function (with the remaining arguments to the format function the same
1795 as they would have been for the unmodified string). For example, the
1800 my_dgettext (char *my_domain, const char *my_format)
1801 __attribute__ ((format_arg (2)));
1805 causes the compiler to check the arguments in calls to a @code{printf},
1806 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
1807 format string argument is a call to the @code{my_dgettext} function, for
1808 consistency with the format string argument @code{my_format}. If the
1809 @code{format_arg} attribute had not been specified, all the compiler
1810 could tell in such calls to format functions would be that the format
1811 string argument is not constant; this would generate a warning when
1812 @option{-Wformat-nonliteral} is used, but the calls could not be checked
1813 without the attribute.
1815 The parameter @var{string-index} specifies which argument is the format
1816 string argument (starting from one). Since non-static C++ methods have
1817 an implicit @code{this} argument, the arguments of such methods should
1818 be counted from two.
1820 The @code{format-arg} attribute allows you to identify your own
1821 functions which modify format strings, so that GCC can check the
1822 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
1823 type function whose operands are a call to one of your own function.
1824 The compiler always treats @code{gettext}, @code{dgettext}, and
1825 @code{dcgettext} in this manner except when strict ISO C support is
1826 requested by @option{-ansi} or an appropriate @option{-std} option, or
1827 @option{-ffreestanding} is used. @xref{C Dialect Options,,Options
1828 Controlling C Dialect}.
1830 @item function_vector
1831 @cindex calling functions through the function vector on the H8/300 processors
1832 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1833 function should be called through the function vector. Calling a
1834 function through the function vector will reduce code size, however;
1835 the function vector has a limited size (maximum 128 entries on the H8/300
1836 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
1838 You must use GAS and GLD from GNU binutils version 2.7 or later for
1839 this attribute to work correctly.
1842 @cindex interrupt handler functions
1843 Use this attribute on the ARM, AVR, C4x, M32R/D and Xstormy16 ports to indicate
1844 that the specified function is an interrupt handler. The compiler will
1845 generate function entry and exit sequences suitable for use in an
1846 interrupt handler when this attribute is present.
1848 Note, interrupt handlers for the m68k, H8/300, H8/300H, H8S, and SH processors
1849 can be specified via the @code{interrupt_handler} attribute.
1851 Note, on the AVR, interrupts will be enabled inside the function.
1853 Note, for the ARM, you can specify the kind of interrupt to be handled by
1854 adding an optional parameter to the interrupt attribute like this:
1857 void f () __attribute__ ((interrupt ("IRQ")));
1860 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
1862 @item interrupt_handler
1863 @cindex interrupt handler functions on the m68k, H8/300 and SH processors
1864 Use this attribute on the m68k, H8/300, H8/300H, H8S, and SH to indicate that
1865 the specified function is an interrupt handler. The compiler will generate
1866 function entry and exit sequences suitable for use in an interrupt
1867 handler when this attribute is present.
1869 @item long_call/short_call
1870 @cindex indirect calls on ARM
1871 This attribute specifies how a particular function is called on
1872 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
1873 command line switch and @code{#pragma long_calls} settings. The
1874 @code{long_call} attribute causes the compiler to always call the
1875 function by first loading its address into a register and then using the
1876 contents of that register. The @code{short_call} attribute always places
1877 the offset to the function from the call site into the @samp{BL}
1878 instruction directly.
1880 @item longcall/shortcall
1881 @cindex functions called via pointer on the RS/6000 and PowerPC
1882 On the RS/6000 and PowerPC, the @code{longcall} attribute causes the
1883 compiler to always call this function via a pointer, just as it would if
1884 the @option{-mlongcall} option had been specified. The @code{shortcall}
1885 attribute causes the compiler not to do this. These attributes override
1886 both the @option{-mlongcall} switch and the @code{#pragma longcall}
1889 @xref{RS/6000 and PowerPC Options}, for more information on whether long
1890 calls are necessary.
1893 @cindex @code{malloc} attribute
1894 The @code{malloc} attribute is used to tell the compiler that a function
1895 may be treated as if any non-@code{NULL} pointer it returns cannot
1896 alias any other pointer valid when the function returns.
1897 This will often improve optimization.
1898 Standard functions with this property include @code{malloc} and
1899 @code{calloc}. @code{realloc}-like functions have this property as
1900 long as the old pointer is never referred to (including comparing it
1901 to the new pointer) after the function returns a non-@code{NULL}
1904 @item model (@var{model-name})
1905 @cindex function addressability on the M32R/D
1906 @cindex variable addressability on the IA-64
1908 On the M32R/D, use this attribute to set the addressability of an
1909 object, and of the code generated for a function. The identifier
1910 @var{model-name} is one of @code{small}, @code{medium}, or
1911 @code{large}, representing each of the code models.
1913 Small model objects live in the lower 16MB of memory (so that their
1914 addresses can be loaded with the @code{ld24} instruction), and are
1915 callable with the @code{bl} instruction.
1917 Medium model objects may live anywhere in the 32-bit address space (the
1918 compiler will generate @code{seth/add3} instructions to load their addresses),
1919 and are callable with the @code{bl} instruction.
1921 Large model objects may live anywhere in the 32-bit address space (the
1922 compiler will generate @code{seth/add3} instructions to load their addresses),
1923 and may not be reachable with the @code{bl} instruction (the compiler will
1924 generate the much slower @code{seth/add3/jl} instruction sequence).
1926 On IA-64, use this attribute to set the addressability of an object.
1927 At present, the only supported identifier for @var{model-name} is
1928 @code{small}, indicating addressability via ``small'' (22-bit)
1929 addresses (so that their addresses can be loaded with the @code{addl}
1930 instruction). Caveat: such addressing is by definition not position
1931 independent and hence this attribute must not be used for objects
1932 defined by shared libraries.
1935 @cindex function without a prologue/epilogue code
1936 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
1937 specified function does not need prologue/epilogue sequences generated by
1938 the compiler. It is up to the programmer to provide these sequences.
1941 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
1942 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
1943 use the normal calling convention based on @code{jsr} and @code{rts}.
1944 This attribute can be used to cancel the effect of the @option{-mlong-calls}
1947 @item no_instrument_function
1948 @cindex @code{no_instrument_function} function attribute
1949 @opindex finstrument-functions
1950 If @option{-finstrument-functions} is given, profiling function calls will
1951 be generated at entry and exit of most user-compiled functions.
1952 Functions with this attribute will not be so instrumented.
1955 @cindex @code{noinline} function attribute
1956 This function attribute prevents a function from being considered for
1959 @item nonnull (@var{arg-index}, @dots{})
1960 @cindex @code{nonnull} function attribute
1961 The @code{nonnull} attribute specifies that some function parameters should
1962 be non-null pointers. For instance, the declaration:
1966 my_memcpy (void *dest, const void *src, size_t len)
1967 __attribute__((nonnull (1, 2)));
1971 causes the compiler to check that, in calls to @code{my_memcpy},
1972 arguments @var{dest} and @var{src} are non-null. If the compiler
1973 determines that a null pointer is passed in an argument slot marked
1974 as non-null, and the @option{-Wnonnull} option is enabled, a warning
1975 is issued. The compiler may also choose to make optimizations based
1976 on the knowledge that certain function arguments will not be null.
1978 If no argument index list is given to the @code{nonnull} attribute,
1979 all pointer arguments are marked as non-null. To illustrate, the
1980 following declaration is equivalent to the previous example:
1984 my_memcpy (void *dest, const void *src, size_t len)
1985 __attribute__((nonnull));
1989 @cindex @code{noreturn} function attribute
1990 A few standard library functions, such as @code{abort} and @code{exit},
1991 cannot return. GCC knows this automatically. Some programs define
1992 their own functions that never return. You can declare them
1993 @code{noreturn} to tell the compiler this fact. For example,
1997 void fatal () __attribute__ ((noreturn));
2000 fatal (/* @r{@dots{}} */)
2002 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2008 The @code{noreturn} keyword tells the compiler to assume that
2009 @code{fatal} cannot return. It can then optimize without regard to what
2010 would happen if @code{fatal} ever did return. This makes slightly
2011 better code. More importantly, it helps avoid spurious warnings of
2012 uninitialized variables.
2014 The @code{noreturn} keyword does not affect the exceptional path when that
2015 applies: a @code{noreturn}-marked function may still return to the caller
2016 by throwing an exception.
2018 Do not assume that registers saved by the calling function are
2019 restored before calling the @code{noreturn} function.
2021 It does not make sense for a @code{noreturn} function to have a return
2022 type other than @code{void}.
2024 The attribute @code{noreturn} is not implemented in GCC versions
2025 earlier than 2.5. An alternative way to declare that a function does
2026 not return, which works in the current version and in some older
2027 versions, is as follows:
2030 typedef void voidfn ();
2032 volatile voidfn fatal;
2036 @cindex @code{nothrow} function attribute
2037 The @code{nothrow} attribute is used to inform the compiler that a
2038 function cannot throw an exception. For example, most functions in
2039 the standard C library can be guaranteed not to throw an exception
2040 with the notable exceptions of @code{qsort} and @code{bsearch} that
2041 take function pointer arguments. The @code{nothrow} attribute is not
2042 implemented in GCC versions earlier than 3.2.
2045 @cindex @code{pure} function attribute
2046 Many functions have no effects except the return value and their
2047 return value depends only on the parameters and/or global variables.
2048 Such a function can be subject
2049 to common subexpression elimination and loop optimization just as an
2050 arithmetic operator would be. These functions should be declared
2051 with the attribute @code{pure}. For example,
2054 int square (int) __attribute__ ((pure));
2058 says that the hypothetical function @code{square} is safe to call
2059 fewer times than the program says.
2061 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2062 Interesting non-pure functions are functions with infinite loops or those
2063 depending on volatile memory or other system resource, that may change between
2064 two consecutive calls (such as @code{feof} in a multithreading environment).
2066 The attribute @code{pure} is not implemented in GCC versions earlier
2069 @item regparm (@var{number})
2070 @cindex @code{regparm} attribute
2071 @cindex functions that are passed arguments in registers on the 386
2072 On the Intel 386, the @code{regparm} attribute causes the compiler to
2073 pass up to @var{number} integer arguments in registers EAX,
2074 EDX, and ECX instead of on the stack. Functions that take a
2075 variable number of arguments will continue to be passed all of their
2076 arguments on the stack.
2078 Beware that on some ELF systems this attribute is unsuitable for
2079 global functions in shared libraries with lazy binding (which is the
2080 default). Lazy binding will send the first call via resolving code in
2081 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2082 per the standard calling conventions. Solaris 8 is affected by this.
2083 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2084 safe since the loaders there save all registers. (Lazy binding can be
2085 disabled with the linker or the loader if desired, to avoid the
2089 @cindex save all registers on the H8/300, H8/300H, and H8S
2090 Use this attribute on the H8/300, H8/300H, and H8S to indicate that
2091 all registers except the stack pointer should be saved in the prologue
2092 regardless of whether they are used or not.
2094 @item section ("@var{section-name}")
2095 @cindex @code{section} function attribute
2096 Normally, the compiler places the code it generates in the @code{text} section.
2097 Sometimes, however, you need additional sections, or you need certain
2098 particular functions to appear in special sections. The @code{section}
2099 attribute specifies that a function lives in a particular section.
2100 For example, the declaration:
2103 extern void foobar (void) __attribute__ ((section ("bar")));
2107 puts the function @code{foobar} in the @code{bar} section.
2109 Some file formats do not support arbitrary sections so the @code{section}
2110 attribute is not available on all platforms.
2111 If you need to map the entire contents of a module to a particular
2112 section, consider using the facilities of the linker instead.
2115 @cindex @code{sentinel} function attribute
2116 This function attribute ensures that a parameter in a function call is
2117 an explicit @code{NULL}. The attribute is only valid on variadic
2118 functions. By default, the sentinel is located at position zero, the
2119 last parameter of the function call. If an optional integer position
2120 argument P is supplied to the attribute, the sentinel must be located at
2121 position P counting backwards from the end of the argument list.
2124 __attribute__ ((sentinel))
2126 __attribute__ ((sentinel(0)))
2129 The attribute is automatically set with a position of 0 for the built-in
2130 functions @code{execl} and @code{execlp}. The built-in function
2131 @code{execle} has the attribute set set with a position of 1.
2133 A valid @code{NULL} in this context is defined as zero with any pointer
2134 type. If your system defines the @code{NULL} macro with an integer type
2135 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2136 with a copy that redefines NULL appropriately.
2138 The warnings for missing or incorrect sentinels are enabled with
2142 See long_call/short_call.
2145 See longcall/shortcall.
2148 @cindex signal handler functions on the AVR processors
2149 Use this attribute on the AVR to indicate that the specified
2150 function is a signal handler. The compiler will generate function
2151 entry and exit sequences suitable for use in a signal handler when this
2152 attribute is present. Interrupts will be disabled inside the function.
2155 Use this attribute on the SH to indicate an @code{interrupt_handler}
2156 function should switch to an alternate stack. It expects a string
2157 argument that names a global variable holding the address of the
2162 void f () __attribute__ ((interrupt_handler,
2163 sp_switch ("alt_stack")));
2167 @cindex functions that pop the argument stack on the 386
2168 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2169 assume that the called function will pop off the stack space used to
2170 pass arguments, unless it takes a variable number of arguments.
2173 @cindex tiny data section on the H8/300H and H8S
2174 Use this attribute on the H8/300H and H8S to indicate that the specified
2175 variable should be placed into the tiny data section.
2176 The compiler will generate more efficient code for loads and stores
2177 on data in the tiny data section. Note the tiny data area is limited to
2178 slightly under 32kbytes of data.
2181 Use this attribute on the SH for an @code{interrupt_handler} to return using
2182 @code{trapa} instead of @code{rte}. This attribute expects an integer
2183 argument specifying the trap number to be used.
2186 @cindex @code{unused} attribute.
2187 This attribute, attached to a function, means that the function is meant
2188 to be possibly unused. GCC will not produce a warning for this
2192 @cindex @code{used} attribute.
2193 This attribute, attached to a function, means that code must be emitted
2194 for the function even if it appears that the function is not referenced.
2195 This is useful, for example, when the function is referenced only in
2198 @item visibility ("@var{visibility_type}")
2199 @cindex @code{visibility} attribute
2200 The @code{visibility} attribute on ELF targets causes the declaration
2201 to be emitted with default, hidden, protected or internal visibility.
2204 void __attribute__ ((visibility ("protected")))
2205 f () @{ /* @r{Do something.} */; @}
2206 int i __attribute__ ((visibility ("hidden")));
2209 See the ELF gABI for complete details, but the short story is:
2212 @c keep this list of visibilities in alphabetical order.
2215 Default visibility is the normal case for ELF. This value is
2216 available for the visibility attribute to override other options
2217 that may change the assumed visibility of symbols.
2220 Hidden visibility indicates that the symbol will not be placed into
2221 the dynamic symbol table, so no other @dfn{module} (executable or
2222 shared library) can reference it directly.
2225 Internal visibility is like hidden visibility, but with additional
2226 processor specific semantics. Unless otherwise specified by the psABI,
2227 GCC defines internal visibility to mean that the function is @emph{never}
2228 called from another module. Note that hidden symbols, while they cannot
2229 be referenced directly by other modules, can be referenced indirectly via
2230 function pointers. By indicating that a symbol cannot be called from
2231 outside the module, GCC may for instance omit the load of a PIC register
2232 since it is known that the calling function loaded the correct value.
2235 Protected visibility indicates that the symbol will be placed in the
2236 dynamic symbol table, but that references within the defining module
2237 will bind to the local symbol. That is, the symbol cannot be overridden
2242 Not all ELF targets support this attribute.
2244 @item warn_unused_result
2245 @cindex @code{warn_unused_result} attribute
2246 The @code{warn_unused_result} attribute causes a warning to be emitted
2247 if a caller of the function with this attribute does not use its
2248 return value. This is useful for functions where not checking
2249 the result is either a security problem or always a bug, such as
2253 int fn () __attribute__ ((warn_unused_result));
2256 if (fn () < 0) return -1;
2262 results in warning on line 5.
2265 @cindex @code{weak} attribute
2266 The @code{weak} attribute causes the declaration to be emitted as a weak
2267 symbol rather than a global. This is primarily useful in defining
2268 library functions which can be overridden in user code, though it can
2269 also be used with non-function declarations. Weak symbols are supported
2270 for ELF targets, and also for a.out targets when using the GNU assembler
2275 You can specify multiple attributes in a declaration by separating them
2276 by commas within the double parentheses or by immediately following an
2277 attribute declaration with another attribute declaration.
2279 @cindex @code{#pragma}, reason for not using
2280 @cindex pragma, reason for not using
2281 Some people object to the @code{__attribute__} feature, suggesting that
2282 ISO C's @code{#pragma} should be used instead. At the time
2283 @code{__attribute__} was designed, there were two reasons for not doing
2288 It is impossible to generate @code{#pragma} commands from a macro.
2291 There is no telling what the same @code{#pragma} might mean in another
2295 These two reasons applied to almost any application that might have been
2296 proposed for @code{#pragma}. It was basically a mistake to use
2297 @code{#pragma} for @emph{anything}.
2299 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2300 to be generated from macros. In addition, a @code{#pragma GCC}
2301 namespace is now in use for GCC-specific pragmas. However, it has been
2302 found convenient to use @code{__attribute__} to achieve a natural
2303 attachment of attributes to their corresponding declarations, whereas
2304 @code{#pragma GCC} is of use for constructs that do not naturally form
2305 part of the grammar. @xref{Other Directives,,Miscellaneous
2306 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2308 @node Attribute Syntax
2309 @section Attribute Syntax
2310 @cindex attribute syntax
2312 This section describes the syntax with which @code{__attribute__} may be
2313 used, and the constructs to which attribute specifiers bind, for the C
2314 language. Some details may vary for C++ and Objective-C@. Because of
2315 infelicities in the grammar for attributes, some forms described here
2316 may not be successfully parsed in all cases.
2318 There are some problems with the semantics of attributes in C++. For
2319 example, there are no manglings for attributes, although they may affect
2320 code generation, so problems may arise when attributed types are used in
2321 conjunction with templates or overloading. Similarly, @code{typeid}
2322 does not distinguish between types with different attributes. Support
2323 for attributes in C++ may be restricted in future to attributes on
2324 declarations only, but not on nested declarators.
2326 @xref{Function Attributes}, for details of the semantics of attributes
2327 applying to functions. @xref{Variable Attributes}, for details of the
2328 semantics of attributes applying to variables. @xref{Type Attributes},
2329 for details of the semantics of attributes applying to structure, union
2330 and enumerated types.
2332 An @dfn{attribute specifier} is of the form
2333 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2334 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2335 each attribute is one of the following:
2339 Empty. Empty attributes are ignored.
2342 A word (which may be an identifier such as @code{unused}, or a reserved
2343 word such as @code{const}).
2346 A word, followed by, in parentheses, parameters for the attribute.
2347 These parameters take one of the following forms:
2351 An identifier. For example, @code{mode} attributes use this form.
2354 An identifier followed by a comma and a non-empty comma-separated list
2355 of expressions. For example, @code{format} attributes use this form.
2358 A possibly empty comma-separated list of expressions. For example,
2359 @code{format_arg} attributes use this form with the list being a single
2360 integer constant expression, and @code{alias} attributes use this form
2361 with the list being a single string constant.
2365 An @dfn{attribute specifier list} is a sequence of one or more attribute
2366 specifiers, not separated by any other tokens.
2368 In GNU C, an attribute specifier list may appear after the colon following a
2369 label, other than a @code{case} or @code{default} label. The only
2370 attribute it makes sense to use after a label is @code{unused}. This
2371 feature is intended for code generated by programs which contains labels
2372 that may be unused but which is compiled with @option{-Wall}. It would
2373 not normally be appropriate to use in it human-written code, though it
2374 could be useful in cases where the code that jumps to the label is
2375 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2376 such placement of attribute lists, as it is permissible for a
2377 declaration, which could begin with an attribute list, to be labelled in
2378 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2379 does not arise there.
2381 An attribute specifier list may appear as part of a @code{struct},
2382 @code{union} or @code{enum} specifier. It may go either immediately
2383 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2384 the closing brace. It is ignored if the content of the structure, union
2385 or enumerated type is not defined in the specifier in which the
2386 attribute specifier list is used---that is, in usages such as
2387 @code{struct __attribute__((foo)) bar} with no following opening brace.
2388 Where attribute specifiers follow the closing brace, they are considered
2389 to relate to the structure, union or enumerated type defined, not to any
2390 enclosing declaration the type specifier appears in, and the type
2391 defined is not complete until after the attribute specifiers.
2392 @c Otherwise, there would be the following problems: a shift/reduce
2393 @c conflict between attributes binding the struct/union/enum and
2394 @c binding to the list of specifiers/qualifiers; and "aligned"
2395 @c attributes could use sizeof for the structure, but the size could be
2396 @c changed later by "packed" attributes.
2398 Otherwise, an attribute specifier appears as part of a declaration,
2399 counting declarations of unnamed parameters and type names, and relates
2400 to that declaration (which may be nested in another declaration, for
2401 example in the case of a parameter declaration), or to a particular declarator
2402 within a declaration. Where an
2403 attribute specifier is applied to a parameter declared as a function or
2404 an array, it should apply to the function or array rather than the
2405 pointer to which the parameter is implicitly converted, but this is not
2406 yet correctly implemented.
2408 Any list of specifiers and qualifiers at the start of a declaration may
2409 contain attribute specifiers, whether or not such a list may in that
2410 context contain storage class specifiers. (Some attributes, however,
2411 are essentially in the nature of storage class specifiers, and only make
2412 sense where storage class specifiers may be used; for example,
2413 @code{section}.) There is one necessary limitation to this syntax: the
2414 first old-style parameter declaration in a function definition cannot
2415 begin with an attribute specifier, because such an attribute applies to
2416 the function instead by syntax described below (which, however, is not
2417 yet implemented in this case). In some other cases, attribute
2418 specifiers are permitted by this grammar but not yet supported by the
2419 compiler. All attribute specifiers in this place relate to the
2420 declaration as a whole. In the obsolescent usage where a type of
2421 @code{int} is implied by the absence of type specifiers, such a list of
2422 specifiers and qualifiers may be an attribute specifier list with no
2423 other specifiers or qualifiers.
2425 An attribute specifier list may appear immediately before a declarator
2426 (other than the first) in a comma-separated list of declarators in a
2427 declaration of more than one identifier using a single list of
2428 specifiers and qualifiers. Such attribute specifiers apply
2429 only to the identifier before whose declarator they appear. For
2433 __attribute__((noreturn)) void d0 (void),
2434 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2439 the @code{noreturn} attribute applies to all the functions
2440 declared; the @code{format} attribute only applies to @code{d1}.
2442 An attribute specifier list may appear immediately before the comma,
2443 @code{=} or semicolon terminating the declaration of an identifier other
2444 than a function definition. At present, such attribute specifiers apply
2445 to the declared object or function, but in future they may attach to the
2446 outermost adjacent declarator. In simple cases there is no difference,
2447 but, for example, in
2450 void (****f)(void) __attribute__((noreturn));
2454 at present the @code{noreturn} attribute applies to @code{f}, which
2455 causes a warning since @code{f} is not a function, but in future it may
2456 apply to the function @code{****f}. The precise semantics of what
2457 attributes in such cases will apply to are not yet specified. Where an
2458 assembler name for an object or function is specified (@pxref{Asm
2459 Labels}), at present the attribute must follow the @code{asm}
2460 specification; in future, attributes before the @code{asm} specification
2461 may apply to the adjacent declarator, and those after it to the declared
2464 An attribute specifier list may, in future, be permitted to appear after
2465 the declarator in a function definition (before any old-style parameter
2466 declarations or the function body).
2468 Attribute specifiers may be mixed with type qualifiers appearing inside
2469 the @code{[]} of a parameter array declarator, in the C99 construct by
2470 which such qualifiers are applied to the pointer to which the array is
2471 implicitly converted. Such attribute specifiers apply to the pointer,
2472 not to the array, but at present this is not implemented and they are
2475 An attribute specifier list may appear at the start of a nested
2476 declarator. At present, there are some limitations in this usage: the
2477 attributes correctly apply to the declarator, but for most individual
2478 attributes the semantics this implies are not implemented.
2479 When attribute specifiers follow the @code{*} of a pointer
2480 declarator, they may be mixed with any type qualifiers present.
2481 The following describes the formal semantics of this syntax. It will make the
2482 most sense if you are familiar with the formal specification of
2483 declarators in the ISO C standard.
2485 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2486 D1}, where @code{T} contains declaration specifiers that specify a type
2487 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2488 contains an identifier @var{ident}. The type specified for @var{ident}
2489 for derived declarators whose type does not include an attribute
2490 specifier is as in the ISO C standard.
2492 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2493 and the declaration @code{T D} specifies the type
2494 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2495 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2496 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2498 If @code{D1} has the form @code{*
2499 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2500 declaration @code{T D} specifies the type
2501 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2502 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2503 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2509 void (__attribute__((noreturn)) ****f) (void);
2513 specifies the type ``pointer to pointer to pointer to pointer to
2514 non-returning function returning @code{void}''. As another example,
2517 char *__attribute__((aligned(8))) *f;
2521 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2522 Note again that this does not work with most attributes; for example,
2523 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2524 is not yet supported.
2526 For compatibility with existing code written for compiler versions that
2527 did not implement attributes on nested declarators, some laxity is
2528 allowed in the placing of attributes. If an attribute that only applies
2529 to types is applied to a declaration, it will be treated as applying to
2530 the type of that declaration. If an attribute that only applies to
2531 declarations is applied to the type of a declaration, it will be treated
2532 as applying to that declaration; and, for compatibility with code
2533 placing the attributes immediately before the identifier declared, such
2534 an attribute applied to a function return type will be treated as
2535 applying to the function type, and such an attribute applied to an array
2536 element type will be treated as applying to the array type. If an
2537 attribute that only applies to function types is applied to a
2538 pointer-to-function type, it will be treated as applying to the pointer
2539 target type; if such an attribute is applied to a function return type
2540 that is not a pointer-to-function type, it will be treated as applying
2541 to the function type.
2543 @node Function Prototypes
2544 @section Prototypes and Old-Style Function Definitions
2545 @cindex function prototype declarations
2546 @cindex old-style function definitions
2547 @cindex promotion of formal parameters
2549 GNU C extends ISO C to allow a function prototype to override a later
2550 old-style non-prototype definition. Consider the following example:
2553 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2560 /* @r{Prototype function declaration.} */
2561 int isroot P((uid_t));
2563 /* @r{Old-style function definition.} */
2565 isroot (x) /* ??? lossage here ??? */
2572 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2573 not allow this example, because subword arguments in old-style
2574 non-prototype definitions are promoted. Therefore in this example the
2575 function definition's argument is really an @code{int}, which does not
2576 match the prototype argument type of @code{short}.
2578 This restriction of ISO C makes it hard to write code that is portable
2579 to traditional C compilers, because the programmer does not know
2580 whether the @code{uid_t} type is @code{short}, @code{int}, or
2581 @code{long}. Therefore, in cases like these GNU C allows a prototype
2582 to override a later old-style definition. More precisely, in GNU C, a
2583 function prototype argument type overrides the argument type specified
2584 by a later old-style definition if the former type is the same as the
2585 latter type before promotion. Thus in GNU C the above example is
2586 equivalent to the following:
2599 GNU C++ does not support old-style function definitions, so this
2600 extension is irrelevant.
2603 @section C++ Style Comments
2605 @cindex C++ comments
2606 @cindex comments, C++ style
2608 In GNU C, you may use C++ style comments, which start with @samp{//} and
2609 continue until the end of the line. Many other C implementations allow
2610 such comments, and they are included in the 1999 C standard. However,
2611 C++ style comments are not recognized if you specify an @option{-std}
2612 option specifying a version of ISO C before C99, or @option{-ansi}
2613 (equivalent to @option{-std=c89}).
2616 @section Dollar Signs in Identifier Names
2618 @cindex dollar signs in identifier names
2619 @cindex identifier names, dollar signs in
2621 In GNU C, you may normally use dollar signs in identifier names.
2622 This is because many traditional C implementations allow such identifiers.
2623 However, dollar signs in identifiers are not supported on a few target
2624 machines, typically because the target assembler does not allow them.
2626 @node Character Escapes
2627 @section The Character @key{ESC} in Constants
2629 You can use the sequence @samp{\e} in a string or character constant to
2630 stand for the ASCII character @key{ESC}.
2633 @section Inquiring on Alignment of Types or Variables
2635 @cindex type alignment
2636 @cindex variable alignment
2638 The keyword @code{__alignof__} allows you to inquire about how an object
2639 is aligned, or the minimum alignment usually required by a type. Its
2640 syntax is just like @code{sizeof}.
2642 For example, if the target machine requires a @code{double} value to be
2643 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2644 This is true on many RISC machines. On more traditional machine
2645 designs, @code{__alignof__ (double)} is 4 or even 2.
2647 Some machines never actually require alignment; they allow reference to any
2648 data type even at an odd address. For these machines, @code{__alignof__}
2649 reports the @emph{recommended} alignment of a type.
2651 If the operand of @code{__alignof__} is an lvalue rather than a type,
2652 its value is the required alignment for its type, taking into account
2653 any minimum alignment specified with GCC's @code{__attribute__}
2654 extension (@pxref{Variable Attributes}). For example, after this
2658 struct foo @{ int x; char y; @} foo1;
2662 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2663 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2665 It is an error to ask for the alignment of an incomplete type.
2667 @node Variable Attributes
2668 @section Specifying Attributes of Variables
2669 @cindex attribute of variables
2670 @cindex variable attributes
2672 The keyword @code{__attribute__} allows you to specify special
2673 attributes of variables or structure fields. This keyword is followed
2674 by an attribute specification inside double parentheses. Some
2675 attributes are currently defined generically for variables.
2676 Other attributes are defined for variables on particular target
2677 systems. Other attributes are available for functions
2678 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
2679 Other front ends might define more attributes
2680 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2682 You may also specify attributes with @samp{__} preceding and following
2683 each keyword. This allows you to use them in header files without
2684 being concerned about a possible macro of the same name. For example,
2685 you may use @code{__aligned__} instead of @code{aligned}.
2687 @xref{Attribute Syntax}, for details of the exact syntax for using
2691 @cindex @code{aligned} attribute
2692 @item aligned (@var{alignment})
2693 This attribute specifies a minimum alignment for the variable or
2694 structure field, measured in bytes. For example, the declaration:
2697 int x __attribute__ ((aligned (16))) = 0;
2701 causes the compiler to allocate the global variable @code{x} on a
2702 16-byte boundary. On a 68040, this could be used in conjunction with
2703 an @code{asm} expression to access the @code{move16} instruction which
2704 requires 16-byte aligned operands.
2706 You can also specify the alignment of structure fields. For example, to
2707 create a double-word aligned @code{int} pair, you could write:
2710 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2714 This is an alternative to creating a union with a @code{double} member
2715 that forces the union to be double-word aligned.
2717 As in the preceding examples, you can explicitly specify the alignment
2718 (in bytes) that you wish the compiler to use for a given variable or
2719 structure field. Alternatively, you can leave out the alignment factor
2720 and just ask the compiler to align a variable or field to the maximum
2721 useful alignment for the target machine you are compiling for. For
2722 example, you could write:
2725 short array[3] __attribute__ ((aligned));
2728 Whenever you leave out the alignment factor in an @code{aligned} attribute
2729 specification, the compiler automatically sets the alignment for the declared
2730 variable or field to the largest alignment which is ever used for any data
2731 type on the target machine you are compiling for. Doing this can often make
2732 copy operations more efficient, because the compiler can use whatever
2733 instructions copy the biggest chunks of memory when performing copies to
2734 or from the variables or fields that you have aligned this way.
2736 The @code{aligned} attribute can only increase the alignment; but you
2737 can decrease it by specifying @code{packed} as well. See below.
2739 Note that the effectiveness of @code{aligned} attributes may be limited
2740 by inherent limitations in your linker. On many systems, the linker is
2741 only able to arrange for variables to be aligned up to a certain maximum
2742 alignment. (For some linkers, the maximum supported alignment may
2743 be very very small.) If your linker is only able to align variables
2744 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
2745 in an @code{__attribute__} will still only provide you with 8 byte
2746 alignment. See your linker documentation for further information.
2748 @item cleanup (@var{cleanup_function})
2749 @cindex @code{cleanup} attribute
2750 The @code{cleanup} attribute runs a function when the variable goes
2751 out of scope. This attribute can only be applied to auto function
2752 scope variables; it may not be applied to parameters or variables
2753 with static storage duration. The function must take one parameter,
2754 a pointer to a type compatible with the variable. The return value
2755 of the function (if any) is ignored.
2757 If @option{-fexceptions} is enabled, then @var{cleanup_function}
2758 will be run during the stack unwinding that happens during the
2759 processing of the exception. Note that the @code{cleanup} attribute
2760 does not allow the exception to be caught, only to perform an action.
2761 It is undefined what happens if @var{cleanup_function} does not
2766 @cindex @code{common} attribute
2767 @cindex @code{nocommon} attribute
2770 The @code{common} attribute requests GCC to place a variable in
2771 ``common'' storage. The @code{nocommon} attribute requests the
2772 opposite -- to allocate space for it directly.
2774 These attributes override the default chosen by the
2775 @option{-fno-common} and @option{-fcommon} flags respectively.
2778 @cindex @code{deprecated} attribute
2779 The @code{deprecated} attribute results in a warning if the variable
2780 is used anywhere in the source file. This is useful when identifying
2781 variables that are expected to be removed in a future version of a
2782 program. The warning also includes the location of the declaration
2783 of the deprecated variable, to enable users to easily find further
2784 information about why the variable is deprecated, or what they should
2785 do instead. Note that the warning only occurs for uses:
2788 extern int old_var __attribute__ ((deprecated));
2790 int new_fn () @{ return old_var; @}
2793 results in a warning on line 3 but not line 2.
2795 The @code{deprecated} attribute can also be used for functions and
2796 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
2798 @item mode (@var{mode})
2799 @cindex @code{mode} attribute
2800 This attribute specifies the data type for the declaration---whichever
2801 type corresponds to the mode @var{mode}. This in effect lets you
2802 request an integer or floating point type according to its width.
2804 You may also specify a mode of @samp{byte} or @samp{__byte__} to
2805 indicate the mode corresponding to a one-byte integer, @samp{word} or
2806 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
2807 or @samp{__pointer__} for the mode used to represent pointers.
2810 @cindex @code{packed} attribute
2811 The @code{packed} attribute specifies that a variable or structure field
2812 should have the smallest possible alignment---one byte for a variable,
2813 and one bit for a field, unless you specify a larger value with the
2814 @code{aligned} attribute.
2816 Here is a structure in which the field @code{x} is packed, so that it
2817 immediately follows @code{a}:
2823 int x[2] __attribute__ ((packed));
2827 @item section ("@var{section-name}")
2828 @cindex @code{section} variable attribute
2829 Normally, the compiler places the objects it generates in sections like
2830 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
2831 or you need certain particular variables to appear in special sections,
2832 for example to map to special hardware. The @code{section}
2833 attribute specifies that a variable (or function) lives in a particular
2834 section. For example, this small program uses several specific section names:
2837 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
2838 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
2839 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
2840 int init_data __attribute__ ((section ("INITDATA"))) = 0;
2844 /* Initialize stack pointer */
2845 init_sp (stack + sizeof (stack));
2847 /* Initialize initialized data */
2848 memcpy (&init_data, &data, &edata - &data);
2850 /* Turn on the serial ports */
2857 Use the @code{section} attribute with an @emph{initialized} definition
2858 of a @emph{global} variable, as shown in the example. GCC issues
2859 a warning and otherwise ignores the @code{section} attribute in
2860 uninitialized variable declarations.
2862 You may only use the @code{section} attribute with a fully initialized
2863 global definition because of the way linkers work. The linker requires
2864 each object be defined once, with the exception that uninitialized
2865 variables tentatively go in the @code{common} (or @code{bss}) section
2866 and can be multiply ``defined''. You can force a variable to be
2867 initialized with the @option{-fno-common} flag or the @code{nocommon}
2870 Some file formats do not support arbitrary sections so the @code{section}
2871 attribute is not available on all platforms.
2872 If you need to map the entire contents of a module to a particular
2873 section, consider using the facilities of the linker instead.
2876 @cindex @code{shared} variable attribute
2877 On Microsoft Windows, in addition to putting variable definitions in a named
2878 section, the section can also be shared among all running copies of an
2879 executable or DLL@. For example, this small program defines shared data
2880 by putting it in a named section @code{shared} and marking the section
2884 int foo __attribute__((section ("shared"), shared)) = 0;
2889 /* Read and write foo. All running
2890 copies see the same value. */
2896 You may only use the @code{shared} attribute along with @code{section}
2897 attribute with a fully initialized global definition because of the way
2898 linkers work. See @code{section} attribute for more information.
2900 The @code{shared} attribute is only available on Microsoft Windows@.
2902 @item tls_model ("@var{tls_model}")
2903 @cindex @code{tls_model} attribute
2904 The @code{tls_model} attribute sets thread-local storage model
2905 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
2906 overriding @code{-ftls-model=} command line switch on a per-variable
2908 The @var{tls_model} argument should be one of @code{global-dynamic},
2909 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
2911 Not all targets support this attribute.
2913 @item transparent_union
2914 This attribute, attached to a function parameter which is a union, means
2915 that the corresponding argument may have the type of any union member,
2916 but the argument is passed as if its type were that of the first union
2917 member. For more details see @xref{Type Attributes}. You can also use
2918 this attribute on a @code{typedef} for a union data type; then it
2919 applies to all function parameters with that type.
2922 This attribute, attached to a variable, means that the variable is meant
2923 to be possibly unused. GCC will not produce a warning for this
2926 @item vector_size (@var{bytes})
2927 This attribute specifies the vector size for the variable, measured in
2928 bytes. For example, the declaration:
2931 int foo __attribute__ ((vector_size (16)));
2935 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
2936 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
2937 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
2939 This attribute is only applicable to integral and float scalars,
2940 although arrays, pointers, and function return values are allowed in
2941 conjunction with this construct.
2943 Aggregates with this attribute are invalid, even if they are of the same
2944 size as a corresponding scalar. For example, the declaration:
2947 struct S @{ int a; @};
2948 struct S __attribute__ ((vector_size (16))) foo;
2952 is invalid even if the size of the structure is the same as the size of
2956 The @code{weak} attribute is described in @xref{Function Attributes}.
2959 The @code{dllimport} attribute is described in @xref{Function Attributes}.
2962 The @code{dllexport} attribute is described in @xref{Function Attributes}.
2966 @subsection M32R/D Variable Attributes
2968 One attribute is currently defined for the M32R/D.
2971 @item model (@var{model-name})
2972 @cindex variable addressability on the M32R/D
2973 Use this attribute on the M32R/D to set the addressability of an object.
2974 The identifier @var{model-name} is one of @code{small}, @code{medium},
2975 or @code{large}, representing each of the code models.
2977 Small model objects live in the lower 16MB of memory (so that their
2978 addresses can be loaded with the @code{ld24} instruction).
2980 Medium and large model objects may live anywhere in the 32-bit address space
2981 (the compiler will generate @code{seth/add3} instructions to load their
2985 @subsection i386 Variable Attributes
2987 Two attributes are currently defined for i386 configurations:
2988 @code{ms_struct} and @code{gcc_struct}
2993 @cindex @code{ms_struct} attribute
2994 @cindex @code{gcc_struct} attribute
2996 If @code{packed} is used on a structure, or if bit-fields are used
2997 it may be that the Microsoft ABI packs them differently
2998 than GCC would normally pack them. Particularly when moving packed
2999 data between functions compiled with GCC and the native Microsoft compiler
3000 (either via function call or as data in a file), it may be necessary to access
3003 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3004 compilers to match the native Microsoft compiler.
3007 @subsection Xstormy16 Variable Attributes
3009 One attribute is currently defined for xstormy16 configurations:
3014 @cindex @code{below100} attribute
3016 If a variable has the @code{below100} attribute (@code{BELOW100} is
3017 allowed also), GCC will place the variable in the first 0x100 bytes of
3018 memory and use special opcodes to access it. Such variables will be
3019 placed in either the @code{.bss_below100} section or the
3020 @code{.data_below100} section.
3024 @node Type Attributes
3025 @section Specifying Attributes of Types
3026 @cindex attribute of types
3027 @cindex type attributes
3029 The keyword @code{__attribute__} allows you to specify special
3030 attributes of @code{struct} and @code{union} types when you define such
3031 types. This keyword is followed by an attribute specification inside
3032 double parentheses. Six attributes are currently defined for types:
3033 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3034 @code{deprecated} and @code{may_alias}. Other attributes are defined for
3035 functions (@pxref{Function Attributes}) and for variables
3036 (@pxref{Variable Attributes}).
3038 You may also specify any one of these attributes with @samp{__}
3039 preceding and following its keyword. This allows you to use these
3040 attributes in header files without being concerned about a possible
3041 macro of the same name. For example, you may use @code{__aligned__}
3042 instead of @code{aligned}.
3044 You may specify the @code{aligned} and @code{transparent_union}
3045 attributes either in a @code{typedef} declaration or just past the
3046 closing curly brace of a complete enum, struct or union type
3047 @emph{definition} and the @code{packed} attribute only past the closing
3048 brace of a definition.
3050 You may also specify attributes between the enum, struct or union
3051 tag and the name of the type rather than after the closing brace.
3053 @xref{Attribute Syntax}, for details of the exact syntax for using
3057 @cindex @code{aligned} attribute
3058 @item aligned (@var{alignment})
3059 This attribute specifies a minimum alignment (in bytes) for variables
3060 of the specified type. For example, the declarations:
3063 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3064 typedef int more_aligned_int __attribute__ ((aligned (8)));
3068 force the compiler to insure (as far as it can) that each variable whose
3069 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3070 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3071 variables of type @code{struct S} aligned to 8-byte boundaries allows
3072 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3073 store) instructions when copying one variable of type @code{struct S} to
3074 another, thus improving run-time efficiency.
3076 Note that the alignment of any given @code{struct} or @code{union} type
3077 is required by the ISO C standard to be at least a perfect multiple of
3078 the lowest common multiple of the alignments of all of the members of
3079 the @code{struct} or @code{union} in question. This means that you @emph{can}
3080 effectively adjust the alignment of a @code{struct} or @code{union}
3081 type by attaching an @code{aligned} attribute to any one of the members
3082 of such a type, but the notation illustrated in the example above is a
3083 more obvious, intuitive, and readable way to request the compiler to
3084 adjust the alignment of an entire @code{struct} or @code{union} type.
3086 As in the preceding example, you can explicitly specify the alignment
3087 (in bytes) that you wish the compiler to use for a given @code{struct}
3088 or @code{union} type. Alternatively, you can leave out the alignment factor
3089 and just ask the compiler to align a type to the maximum
3090 useful alignment for the target machine you are compiling for. For
3091 example, you could write:
3094 struct S @{ short f[3]; @} __attribute__ ((aligned));
3097 Whenever you leave out the alignment factor in an @code{aligned}
3098 attribute specification, the compiler automatically sets the alignment
3099 for the type to the largest alignment which is ever used for any data
3100 type on the target machine you are compiling for. Doing this can often
3101 make copy operations more efficient, because the compiler can use
3102 whatever instructions copy the biggest chunks of memory when performing
3103 copies to or from the variables which have types that you have aligned
3106 In the example above, if the size of each @code{short} is 2 bytes, then
3107 the size of the entire @code{struct S} type is 6 bytes. The smallest
3108 power of two which is greater than or equal to that is 8, so the
3109 compiler sets the alignment for the entire @code{struct S} type to 8
3112 Note that although you can ask the compiler to select a time-efficient
3113 alignment for a given type and then declare only individual stand-alone
3114 objects of that type, the compiler's ability to select a time-efficient
3115 alignment is primarily useful only when you plan to create arrays of
3116 variables having the relevant (efficiently aligned) type. If you
3117 declare or use arrays of variables of an efficiently-aligned type, then
3118 it is likely that your program will also be doing pointer arithmetic (or
3119 subscripting, which amounts to the same thing) on pointers to the
3120 relevant type, and the code that the compiler generates for these
3121 pointer arithmetic operations will often be more efficient for
3122 efficiently-aligned types than for other types.
3124 The @code{aligned} attribute can only increase the alignment; but you
3125 can decrease it by specifying @code{packed} as well. See below.
3127 Note that the effectiveness of @code{aligned} attributes may be limited
3128 by inherent limitations in your linker. On many systems, the linker is
3129 only able to arrange for variables to be aligned up to a certain maximum
3130 alignment. (For some linkers, the maximum supported alignment may
3131 be very very small.) If your linker is only able to align variables
3132 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3133 in an @code{__attribute__} will still only provide you with 8 byte
3134 alignment. See your linker documentation for further information.
3137 This attribute, attached to @code{struct} or @code{union} type
3138 definition, specifies that each member of the structure or union is
3139 placed to minimize the memory required. When attached to an @code{enum}
3140 definition, it indicates that the smallest integral type should be used.
3142 @opindex fshort-enums
3143 Specifying this attribute for @code{struct} and @code{union} types is
3144 equivalent to specifying the @code{packed} attribute on each of the
3145 structure or union members. Specifying the @option{-fshort-enums}
3146 flag on the line is equivalent to specifying the @code{packed}
3147 attribute on all @code{enum} definitions.
3149 In the following example @code{struct my_packed_struct}'s members are
3150 packed closely together, but the internal layout of its @code{s} member
3151 is not packed -- to do that, @code{struct my_unpacked_struct} would need to
3155 struct my_unpacked_struct
3161 struct my_packed_struct __attribute__ ((__packed__))
3165 struct my_unpacked_struct s;
3169 You may only specify this attribute on the definition of a @code{enum},
3170 @code{struct} or @code{union}, not on a @code{typedef} which does not
3171 also define the enumerated type, structure or union.
3173 @item transparent_union
3174 This attribute, attached to a @code{union} type definition, indicates
3175 that any function parameter having that union type causes calls to that
3176 function to be treated in a special way.
3178 First, the argument corresponding to a transparent union type can be of
3179 any type in the union; no cast is required. Also, if the union contains
3180 a pointer type, the corresponding argument can be a null pointer
3181 constant or a void pointer expression; and if the union contains a void
3182 pointer type, the corresponding argument can be any pointer expression.
3183 If the union member type is a pointer, qualifiers like @code{const} on
3184 the referenced type must be respected, just as with normal pointer
3187 Second, the argument is passed to the function using the calling
3188 conventions of the first member of the transparent union, not the calling
3189 conventions of the union itself. All members of the union must have the
3190 same machine representation; this is necessary for this argument passing
3193 Transparent unions are designed for library functions that have multiple
3194 interfaces for compatibility reasons. For example, suppose the
3195 @code{wait} function must accept either a value of type @code{int *} to
3196 comply with Posix, or a value of type @code{union wait *} to comply with
3197 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3198 @code{wait} would accept both kinds of arguments, but it would also
3199 accept any other pointer type and this would make argument type checking
3200 less useful. Instead, @code{<sys/wait.h>} might define the interface
3208 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3210 pid_t wait (wait_status_ptr_t);
3213 This interface allows either @code{int *} or @code{union wait *}
3214 arguments to be passed, using the @code{int *} calling convention.
3215 The program can call @code{wait} with arguments of either type:
3218 int w1 () @{ int w; return wait (&w); @}
3219 int w2 () @{ union wait w; return wait (&w); @}
3222 With this interface, @code{wait}'s implementation might look like this:
3225 pid_t wait (wait_status_ptr_t p)
3227 return waitpid (-1, p.__ip, 0);
3232 When attached to a type (including a @code{union} or a @code{struct}),
3233 this attribute means that variables of that type are meant to appear
3234 possibly unused. GCC will not produce a warning for any variables of
3235 that type, even if the variable appears to do nothing. This is often
3236 the case with lock or thread classes, which are usually defined and then
3237 not referenced, but contain constructors and destructors that have
3238 nontrivial bookkeeping functions.
3241 The @code{deprecated} attribute results in a warning if the type
3242 is used anywhere in the source file. This is useful when identifying
3243 types that are expected to be removed in a future version of a program.
3244 If possible, the warning also includes the location of the declaration
3245 of the deprecated type, to enable users to easily find further
3246 information about why the type is deprecated, or what they should do
3247 instead. Note that the warnings only occur for uses and then only
3248 if the type is being applied to an identifier that itself is not being
3249 declared as deprecated.
3252 typedef int T1 __attribute__ ((deprecated));
3256 typedef T1 T3 __attribute__ ((deprecated));
3257 T3 z __attribute__ ((deprecated));
3260 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3261 warning is issued for line 4 because T2 is not explicitly
3262 deprecated. Line 5 has no warning because T3 is explicitly
3263 deprecated. Similarly for line 6.
3265 The @code{deprecated} attribute can also be used for functions and
3266 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3269 Accesses to objects with types with this attribute are not subjected to
3270 type-based alias analysis, but are instead assumed to be able to alias
3271 any other type of objects, just like the @code{char} type. See
3272 @option{-fstrict-aliasing} for more information on aliasing issues.
3277 typedef short __attribute__((__may_alias__)) short_a;
3283 short_a *b = (short_a *) &a;
3287 if (a == 0x12345678)
3294 If you replaced @code{short_a} with @code{short} in the variable
3295 declaration, the above program would abort when compiled with
3296 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3297 above in recent GCC versions.
3299 @subsection i386 Type Attributes
3301 Two attributes are currently defined for i386 configurations:
3302 @code{ms_struct} and @code{gcc_struct}
3306 @cindex @code{ms_struct}
3307 @cindex @code{gcc_struct}
3309 If @code{packed} is used on a structure, or if bit-fields are used
3310 it may be that the Microsoft ABI packs them differently
3311 than GCC would normally pack them. Particularly when moving packed
3312 data between functions compiled with GCC and the native Microsoft compiler
3313 (either via function call or as data in a file), it may be necessary to access
3316 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3317 compilers to match the native Microsoft compiler.
3320 To specify multiple attributes, separate them by commas within the
3321 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3325 @section An Inline Function is As Fast As a Macro
3326 @cindex inline functions
3327 @cindex integrating function code
3329 @cindex macros, inline alternative
3331 By declaring a function @code{inline}, you can direct GCC to
3332 integrate that function's code into the code for its callers. This
3333 makes execution faster by eliminating the function-call overhead; in
3334 addition, if any of the actual argument values are constant, their known
3335 values may permit simplifications at compile time so that not all of the
3336 inline function's code needs to be included. The effect on code size is
3337 less predictable; object code may be larger or smaller with function
3338 inlining, depending on the particular case. Inlining of functions is an
3339 optimization and it really ``works'' only in optimizing compilation. If
3340 you don't use @option{-O}, no function is really inline.
3342 Inline functions are included in the ISO C99 standard, but there are
3343 currently substantial differences between what GCC implements and what
3344 the ISO C99 standard requires.
3346 To declare a function inline, use the @code{inline} keyword in its
3347 declaration, like this:
3357 (If you are writing a header file to be included in ISO C programs, write
3358 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3359 You can also make all ``simple enough'' functions inline with the option
3360 @option{-finline-functions}.
3363 Note that certain usages in a function definition can make it unsuitable
3364 for inline substitution. Among these usages are: use of varargs, use of
3365 alloca, use of variable sized data types (@pxref{Variable Length}),
3366 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3367 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3368 will warn when a function marked @code{inline} could not be substituted,
3369 and will give the reason for the failure.
3371 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3372 does not affect the linkage of the function.
3374 @cindex automatic @code{inline} for C++ member fns
3375 @cindex @code{inline} automatic for C++ member fns
3376 @cindex member fns, automatically @code{inline}
3377 @cindex C++ member fns, automatically @code{inline}
3378 @opindex fno-default-inline
3379 GCC automatically inlines member functions defined within the class
3380 body of C++ programs even if they are not explicitly declared
3381 @code{inline}. (You can override this with @option{-fno-default-inline};
3382 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3384 @cindex inline functions, omission of
3385 @opindex fkeep-inline-functions
3386 When a function is both inline and @code{static}, if all calls to the
3387 function are integrated into the caller, and the function's address is
3388 never used, then the function's own assembler code is never referenced.
3389 In this case, GCC does not actually output assembler code for the
3390 function, unless you specify the option @option{-fkeep-inline-functions}.
3391 Some calls cannot be integrated for various reasons (in particular,
3392 calls that precede the function's definition cannot be integrated, and
3393 neither can recursive calls within the definition). If there is a
3394 nonintegrated call, then the function is compiled to assembler code as
3395 usual. The function must also be compiled as usual if the program
3396 refers to its address, because that can't be inlined.
3398 @cindex non-static inline function
3399 When an inline function is not @code{static}, then the compiler must assume
3400 that there may be calls from other source files; since a global symbol can
3401 be defined only once in any program, the function must not be defined in
3402 the other source files, so the calls therein cannot be integrated.
3403 Therefore, a non-@code{static} inline function is always compiled on its
3404 own in the usual fashion.
3406 If you specify both @code{inline} and @code{extern} in the function
3407 definition, then the definition is used only for inlining. In no case
3408 is the function compiled on its own, not even if you refer to its
3409 address explicitly. Such an address becomes an external reference, as
3410 if you had only declared the function, and had not defined it.
3412 This combination of @code{inline} and @code{extern} has almost the
3413 effect of a macro. The way to use it is to put a function definition in
3414 a header file with these keywords, and put another copy of the
3415 definition (lacking @code{inline} and @code{extern}) in a library file.
3416 The definition in the header file will cause most calls to the function
3417 to be inlined. If any uses of the function remain, they will refer to
3418 the single copy in the library.
3420 Since GCC eventually will implement ISO C99 semantics for
3421 inline functions, it is best to use @code{static inline} only
3422 to guarantee compatibility. (The
3423 existing semantics will remain available when @option{-std=gnu89} is
3424 specified, but eventually the default will be @option{-std=gnu99} and
3425 that will implement the C99 semantics, though it does not do so yet.)
3427 GCC does not inline any functions when not optimizing unless you specify
3428 the @samp{always_inline} attribute for the function, like this:
3432 inline void foo (const char) __attribute__((always_inline));
3436 @section Assembler Instructions with C Expression Operands
3437 @cindex extended @code{asm}
3438 @cindex @code{asm} expressions
3439 @cindex assembler instructions
3442 In an assembler instruction using @code{asm}, you can specify the
3443 operands of the instruction using C expressions. This means you need not
3444 guess which registers or memory locations will contain the data you want
3447 You must specify an assembler instruction template much like what
3448 appears in a machine description, plus an operand constraint string for
3451 For example, here is how to use the 68881's @code{fsinx} instruction:
3454 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3458 Here @code{angle} is the C expression for the input operand while
3459 @code{result} is that of the output operand. Each has @samp{"f"} as its
3460 operand constraint, saying that a floating point register is required.
3461 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3462 output operands' constraints must use @samp{=}. The constraints use the
3463 same language used in the machine description (@pxref{Constraints}).
3465 Each operand is described by an operand-constraint string followed by
3466 the C expression in parentheses. A colon separates the assembler
3467 template from the first output operand and another separates the last
3468 output operand from the first input, if any. Commas separate the
3469 operands within each group. The total number of operands is currently
3470 limited to 30; this limitation may be lifted in some future version of
3473 If there are no output operands but there are input operands, you must
3474 place two consecutive colons surrounding the place where the output
3477 As of GCC version 3.1, it is also possible to specify input and output
3478 operands using symbolic names which can be referenced within the
3479 assembler code. These names are specified inside square brackets
3480 preceding the constraint string, and can be referenced inside the
3481 assembler code using @code{%[@var{name}]} instead of a percentage sign
3482 followed by the operand number. Using named operands the above example
3486 asm ("fsinx %[angle],%[output]"
3487 : [output] "=f" (result)
3488 : [angle] "f" (angle));
3492 Note that the symbolic operand names have no relation whatsoever to
3493 other C identifiers. You may use any name you like, even those of
3494 existing C symbols, but you must ensure that no two operands within the same
3495 assembler construct use the same symbolic name.
3497 Output operand expressions must be lvalues; the compiler can check this.
3498 The input operands need not be lvalues. The compiler cannot check
3499 whether the operands have data types that are reasonable for the
3500 instruction being executed. It does not parse the assembler instruction
3501 template and does not know what it means or even whether it is valid
3502 assembler input. The extended @code{asm} feature is most often used for
3503 machine instructions the compiler itself does not know exist. If
3504 the output expression cannot be directly addressed (for example, it is a
3505 bit-field), your constraint must allow a register. In that case, GCC
3506 will use the register as the output of the @code{asm}, and then store
3507 that register into the output.
3509 The ordinary output operands must be write-only; GCC will assume that
3510 the values in these operands before the instruction are dead and need
3511 not be generated. Extended asm supports input-output or read-write
3512 operands. Use the constraint character @samp{+} to indicate such an
3513 operand and list it with the output operands. You should only use
3514 read-write operands when the constraints for the operand (or the
3515 operand in which only some of the bits are to be changed) allow a
3518 You may, as an alternative, logically split its function into two
3519 separate operands, one input operand and one write-only output
3520 operand. The connection between them is expressed by constraints
3521 which say they need to be in the same location when the instruction
3522 executes. You can use the same C expression for both operands, or
3523 different expressions. For example, here we write the (fictitious)
3524 @samp{combine} instruction with @code{bar} as its read-only source
3525 operand and @code{foo} as its read-write destination:
3528 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3532 The constraint @samp{"0"} for operand 1 says that it must occupy the
3533 same location as operand 0. A number in constraint is allowed only in
3534 an input operand and it must refer to an output operand.
3536 Only a number in the constraint can guarantee that one operand will be in
3537 the same place as another. The mere fact that @code{foo} is the value
3538 of both operands is not enough to guarantee that they will be in the
3539 same place in the generated assembler code. The following would not
3543 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3546 Various optimizations or reloading could cause operands 0 and 1 to be in
3547 different registers; GCC knows no reason not to do so. For example, the
3548 compiler might find a copy of the value of @code{foo} in one register and
3549 use it for operand 1, but generate the output operand 0 in a different
3550 register (copying it afterward to @code{foo}'s own address). Of course,
3551 since the register for operand 1 is not even mentioned in the assembler
3552 code, the result will not work, but GCC can't tell that.
3554 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3555 the operand number for a matching constraint. For example:
3558 asm ("cmoveq %1,%2,%[result]"
3559 : [result] "=r"(result)
3560 : "r" (test), "r"(new), "[result]"(old));
3563 Some instructions clobber specific hard registers. To describe this,
3564 write a third colon after the input operands, followed by the names of
3565 the clobbered hard registers (given as strings). Here is a realistic
3566 example for the VAX:
3569 asm volatile ("movc3 %0,%1,%2"
3571 : "g" (from), "g" (to), "g" (count)
3572 : "r0", "r1", "r2", "r3", "r4", "r5");
3575 You may not write a clobber description in a way that overlaps with an
3576 input or output operand. For example, you may not have an operand
3577 describing a register class with one member if you mention that register
3578 in the clobber list. Variables declared to live in specific registers
3579 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
3580 have no part mentioned in the clobber description.
3581 There is no way for you to specify that an input
3582 operand is modified without also specifying it as an output
3583 operand. Note that if all the output operands you specify are for this
3584 purpose (and hence unused), you will then also need to specify
3585 @code{volatile} for the @code{asm} construct, as described below, to
3586 prevent GCC from deleting the @code{asm} statement as unused.
3588 If you refer to a particular hardware register from the assembler code,
3589 you will probably have to list the register after the third colon to
3590 tell the compiler the register's value is modified. In some assemblers,
3591 the register names begin with @samp{%}; to produce one @samp{%} in the
3592 assembler code, you must write @samp{%%} in the input.
3594 If your assembler instruction can alter the condition code register, add
3595 @samp{cc} to the list of clobbered registers. GCC on some machines
3596 represents the condition codes as a specific hardware register;
3597 @samp{cc} serves to name this register. On other machines, the
3598 condition code is handled differently, and specifying @samp{cc} has no
3599 effect. But it is valid no matter what the machine.
3601 If your assembler instructions access memory in an unpredictable
3602 fashion, add @samp{memory} to the list of clobbered registers. This
3603 will cause GCC to not keep memory values cached in registers across the
3604 assembler instruction and not optimize stores or loads to that memory.
3605 You will also want to add the @code{volatile} keyword if the memory
3606 affected is not listed in the inputs or outputs of the @code{asm}, as
3607 the @samp{memory} clobber does not count as a side-effect of the
3608 @code{asm}. If you know how large the accessed memory is, you can add
3609 it as input or output but if this is not known, you should add
3610 @samp{memory}. As an example, if you access ten bytes of a string, you
3611 can use a memory input like:
3614 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
3617 Note that in the following example the memory input is necessary,
3618 otherwise GCC might optimize the store to @code{x} away:
3625 asm ("magic stuff accessing an 'int' pointed to by '%1'"
3626 "=&d" (r) : "a" (y), "m" (*y));
3631 You can put multiple assembler instructions together in a single
3632 @code{asm} template, separated by the characters normally used in assembly
3633 code for the system. A combination that works in most places is a newline
3634 to break the line, plus a tab character to move to the instruction field
3635 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
3636 assembler allows semicolons as a line-breaking character. Note that some
3637 assembler dialects use semicolons to start a comment.
3638 The input operands are guaranteed not to use any of the clobbered
3639 registers, and neither will the output operands' addresses, so you can
3640 read and write the clobbered registers as many times as you like. Here
3641 is an example of multiple instructions in a template; it assumes the
3642 subroutine @code{_foo} accepts arguments in registers 9 and 10:
3645 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
3647 : "g" (from), "g" (to)
3651 Unless an output operand has the @samp{&} constraint modifier, GCC
3652 may allocate it in the same register as an unrelated input operand, on
3653 the assumption the inputs are consumed before the outputs are produced.
3654 This assumption may be false if the assembler code actually consists of
3655 more than one instruction. In such a case, use @samp{&} for each output
3656 operand that may not overlap an input. @xref{Modifiers}.
3658 If you want to test the condition code produced by an assembler
3659 instruction, you must include a branch and a label in the @code{asm}
3660 construct, as follows:
3663 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
3669 This assumes your assembler supports local labels, as the GNU assembler
3670 and most Unix assemblers do.
3672 Speaking of labels, jumps from one @code{asm} to another are not
3673 supported. The compiler's optimizers do not know about these jumps, and
3674 therefore they cannot take account of them when deciding how to
3677 @cindex macros containing @code{asm}
3678 Usually the most convenient way to use these @code{asm} instructions is to
3679 encapsulate them in macros that look like functions. For example,
3683 (@{ double __value, __arg = (x); \
3684 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
3689 Here the variable @code{__arg} is used to make sure that the instruction
3690 operates on a proper @code{double} value, and to accept only those
3691 arguments @code{x} which can convert automatically to a @code{double}.
3693 Another way to make sure the instruction operates on the correct data
3694 type is to use a cast in the @code{asm}. This is different from using a
3695 variable @code{__arg} in that it converts more different types. For
3696 example, if the desired type were @code{int}, casting the argument to
3697 @code{int} would accept a pointer with no complaint, while assigning the
3698 argument to an @code{int} variable named @code{__arg} would warn about
3699 using a pointer unless the caller explicitly casts it.
3701 If an @code{asm} has output operands, GCC assumes for optimization
3702 purposes the instruction has no side effects except to change the output
3703 operands. This does not mean instructions with a side effect cannot be
3704 used, but you must be careful, because the compiler may eliminate them
3705 if the output operands aren't used, or move them out of loops, or
3706 replace two with one if they constitute a common subexpression. Also,
3707 if your instruction does have a side effect on a variable that otherwise
3708 appears not to change, the old value of the variable may be reused later
3709 if it happens to be found in a register.
3711 You can prevent an @code{asm} instruction from being deleted, moved
3712 significantly, or combined, by writing the keyword @code{volatile} after
3713 the @code{asm}. For example:
3716 #define get_and_set_priority(new) \
3718 asm volatile ("get_and_set_priority %0, %1" \
3719 : "=g" (__old) : "g" (new)); \
3724 If you write an @code{asm} instruction with no outputs, GCC will know
3725 the instruction has side-effects and will not delete the instruction or
3726 move it outside of loops.
3728 The @code{volatile} keyword indicates that the instruction has
3729 important side-effects. GCC will not delete a volatile @code{asm} if
3730 it is reachable. (The instruction can still be deleted if GCC can
3731 prove that control-flow will never reach the location of the
3732 instruction.) In addition, GCC will not reschedule instructions
3733 across a volatile @code{asm} instruction. For example:
3736 *(volatile int *)addr = foo;
3737 asm volatile ("eieio" : : );
3741 Assume @code{addr} contains the address of a memory mapped device
3742 register. The PowerPC @code{eieio} instruction (Enforce In-order
3743 Execution of I/O) tells the CPU to make sure that the store to that
3744 device register happens before it issues any other I/O@.
3746 Note that even a volatile @code{asm} instruction can be moved in ways
3747 that appear insignificant to the compiler, such as across jump
3748 instructions. You can't expect a sequence of volatile @code{asm}
3749 instructions to remain perfectly consecutive. If you want consecutive
3750 output, use a single @code{asm}. Also, GCC will perform some
3751 optimizations across a volatile @code{asm} instruction; GCC does not
3752 ``forget everything'' when it encounters a volatile @code{asm}
3753 instruction the way some other compilers do.
3755 An @code{asm} instruction without any operands or clobbers (an ``old
3756 style'' @code{asm}) will be treated identically to a volatile
3757 @code{asm} instruction.
3759 It is a natural idea to look for a way to give access to the condition
3760 code left by the assembler instruction. However, when we attempted to
3761 implement this, we found no way to make it work reliably. The problem
3762 is that output operands might need reloading, which would result in
3763 additional following ``store'' instructions. On most machines, these
3764 instructions would alter the condition code before there was time to
3765 test it. This problem doesn't arise for ordinary ``test'' and
3766 ``compare'' instructions because they don't have any output operands.
3768 For reasons similar to those described above, it is not possible to give
3769 an assembler instruction access to the condition code left by previous
3772 If you are writing a header file that should be includable in ISO C
3773 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
3776 @subsection Size of an @code{asm}
3778 Some targets require that GCC track the size of each instruction used in
3779 order to generate correct code. Because the final length of an
3780 @code{asm} is only known by the assembler, GCC must make an estimate as
3781 to how big it will be. The estimate is formed by counting the number of
3782 statements in the pattern of the @code{asm} and multiplying that by the
3783 length of the longest instruction on that processor. Statements in the
3784 @code{asm} are identified by newline characters and whatever statement
3785 separator characters are supported by the assembler; on most processors
3786 this is the `@code{;}' character.
3788 Normally, GCC's estimate is perfectly adequate to ensure that correct
3789 code is generated, but it is possible to confuse the compiler if you use
3790 pseudo instructions or assembler macros that expand into multiple real
3791 instructions or if you use assembler directives that expand to more
3792 space in the object file than would be needed for a single instruction.
3793 If this happens then the assembler will produce a diagnostic saying that
3794 a label is unreachable.
3796 @subsection i386 floating point asm operands
3798 There are several rules on the usage of stack-like regs in
3799 asm_operands insns. These rules apply only to the operands that are
3804 Given a set of input regs that die in an asm_operands, it is
3805 necessary to know which are implicitly popped by the asm, and
3806 which must be explicitly popped by gcc.
3808 An input reg that is implicitly popped by the asm must be
3809 explicitly clobbered, unless it is constrained to match an
3813 For any input reg that is implicitly popped by an asm, it is
3814 necessary to know how to adjust the stack to compensate for the pop.
3815 If any non-popped input is closer to the top of the reg-stack than
3816 the implicitly popped reg, it would not be possible to know what the
3817 stack looked like---it's not clear how the rest of the stack ``slides
3820 All implicitly popped input regs must be closer to the top of
3821 the reg-stack than any input that is not implicitly popped.
3823 It is possible that if an input dies in an insn, reload might
3824 use the input reg for an output reload. Consider this example:
3827 asm ("foo" : "=t" (a) : "f" (b));
3830 This asm says that input B is not popped by the asm, and that
3831 the asm pushes a result onto the reg-stack, i.e., the stack is one
3832 deeper after the asm than it was before. But, it is possible that
3833 reload will think that it can use the same reg for both the input and
3834 the output, if input B dies in this insn.
3836 If any input operand uses the @code{f} constraint, all output reg
3837 constraints must use the @code{&} earlyclobber.
3839 The asm above would be written as
3842 asm ("foo" : "=&t" (a) : "f" (b));
3846 Some operands need to be in particular places on the stack. All
3847 output operands fall in this category---there is no other way to
3848 know which regs the outputs appear in unless the user indicates
3849 this in the constraints.
3851 Output operands must specifically indicate which reg an output
3852 appears in after an asm. @code{=f} is not allowed: the operand
3853 constraints must select a class with a single reg.
3856 Output operands may not be ``inserted'' between existing stack regs.
3857 Since no 387 opcode uses a read/write operand, all output operands
3858 are dead before the asm_operands, and are pushed by the asm_operands.
3859 It makes no sense to push anywhere but the top of the reg-stack.
3861 Output operands must start at the top of the reg-stack: output
3862 operands may not ``skip'' a reg.
3865 Some asm statements may need extra stack space for internal
3866 calculations. This can be guaranteed by clobbering stack registers
3867 unrelated to the inputs and outputs.
3871 Here are a couple of reasonable asms to want to write. This asm
3872 takes one input, which is internally popped, and produces two outputs.
3875 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
3878 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
3879 and replaces them with one output. The user must code the @code{st(1)}
3880 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
3883 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
3889 @section Controlling Names Used in Assembler Code
3890 @cindex assembler names for identifiers
3891 @cindex names used in assembler code
3892 @cindex identifiers, names in assembler code
3894 You can specify the name to be used in the assembler code for a C
3895 function or variable by writing the @code{asm} (or @code{__asm__})
3896 keyword after the declarator as follows:
3899 int foo asm ("myfoo") = 2;
3903 This specifies that the name to be used for the variable @code{foo} in
3904 the assembler code should be @samp{myfoo} rather than the usual
3907 On systems where an underscore is normally prepended to the name of a C
3908 function or variable, this feature allows you to define names for the
3909 linker that do not start with an underscore.
3911 It does not make sense to use this feature with a non-static local
3912 variable since such variables do not have assembler names. If you are
3913 trying to put the variable in a particular register, see @ref{Explicit
3914 Reg Vars}. GCC presently accepts such code with a warning, but will
3915 probably be changed to issue an error, rather than a warning, in the
3918 You cannot use @code{asm} in this way in a function @emph{definition}; but
3919 you can get the same effect by writing a declaration for the function
3920 before its definition and putting @code{asm} there, like this:
3923 extern func () asm ("FUNC");
3930 It is up to you to make sure that the assembler names you choose do not
3931 conflict with any other assembler symbols. Also, you must not use a
3932 register name; that would produce completely invalid assembler code. GCC
3933 does not as yet have the ability to store static variables in registers.
3934 Perhaps that will be added.
3936 @node Explicit Reg Vars
3937 @section Variables in Specified Registers
3938 @cindex explicit register variables
3939 @cindex variables in specified registers
3940 @cindex specified registers
3941 @cindex registers, global allocation
3943 GNU C allows you to put a few global variables into specified hardware
3944 registers. You can also specify the register in which an ordinary
3945 register variable should be allocated.
3949 Global register variables reserve registers throughout the program.
3950 This may be useful in programs such as programming language
3951 interpreters which have a couple of global variables that are accessed
3955 Local register variables in specific registers do not reserve the
3956 registers. The compiler's data flow analysis is capable of determining
3957 where the specified registers contain live values, and where they are
3958 available for other uses. Stores into local register variables may be deleted
3959 when they appear to be dead according to dataflow analysis. References
3960 to local register variables may be deleted or moved or simplified.
3962 These local variables are sometimes convenient for use with the extended
3963 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
3964 output of the assembler instruction directly into a particular register.
3965 (This will work provided the register you specify fits the constraints
3966 specified for that operand in the @code{asm}.)
3974 @node Global Reg Vars
3975 @subsection Defining Global Register Variables
3976 @cindex global register variables
3977 @cindex registers, global variables in
3979 You can define a global register variable in GNU C like this:
3982 register int *foo asm ("a5");
3986 Here @code{a5} is the name of the register which should be used. Choose a
3987 register which is normally saved and restored by function calls on your
3988 machine, so that library routines will not clobber it.
3990 Naturally the register name is cpu-dependent, so you would need to
3991 conditionalize your program according to cpu type. The register
3992 @code{a5} would be a good choice on a 68000 for a variable of pointer
3993 type. On machines with register windows, be sure to choose a ``global''
3994 register that is not affected magically by the function call mechanism.
3996 In addition, operating systems on one type of cpu may differ in how they
3997 name the registers; then you would need additional conditionals. For
3998 example, some 68000 operating systems call this register @code{%a5}.
4000 Eventually there may be a way of asking the compiler to choose a register
4001 automatically, but first we need to figure out how it should choose and
4002 how to enable you to guide the choice. No solution is evident.
4004 Defining a global register variable in a certain register reserves that
4005 register entirely for this use, at least within the current compilation.
4006 The register will not be allocated for any other purpose in the functions
4007 in the current compilation. The register will not be saved and restored by
4008 these functions. Stores into this register are never deleted even if they
4009 would appear to be dead, but references may be deleted or moved or
4012 It is not safe to access the global register variables from signal
4013 handlers, or from more than one thread of control, because the system
4014 library routines may temporarily use the register for other things (unless
4015 you recompile them specially for the task at hand).
4017 @cindex @code{qsort}, and global register variables
4018 It is not safe for one function that uses a global register variable to
4019 call another such function @code{foo} by way of a third function
4020 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4021 different source file in which the variable wasn't declared). This is
4022 because @code{lose} might save the register and put some other value there.
4023 For example, you can't expect a global register variable to be available in
4024 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4025 might have put something else in that register. (If you are prepared to
4026 recompile @code{qsort} with the same global register variable, you can
4027 solve this problem.)
4029 If you want to recompile @code{qsort} or other source files which do not
4030 actually use your global register variable, so that they will not use that
4031 register for any other purpose, then it suffices to specify the compiler
4032 option @option{-ffixed-@var{reg}}. You need not actually add a global
4033 register declaration to their source code.
4035 A function which can alter the value of a global register variable cannot
4036 safely be called from a function compiled without this variable, because it
4037 could clobber the value the caller expects to find there on return.
4038 Therefore, the function which is the entry point into the part of the
4039 program that uses the global register variable must explicitly save and
4040 restore the value which belongs to its caller.
4042 @cindex register variable after @code{longjmp}
4043 @cindex global register after @code{longjmp}
4044 @cindex value after @code{longjmp}
4047 On most machines, @code{longjmp} will restore to each global register
4048 variable the value it had at the time of the @code{setjmp}. On some
4049 machines, however, @code{longjmp} will not change the value of global
4050 register variables. To be portable, the function that called @code{setjmp}
4051 should make other arrangements to save the values of the global register
4052 variables, and to restore them in a @code{longjmp}. This way, the same
4053 thing will happen regardless of what @code{longjmp} does.
4055 All global register variable declarations must precede all function
4056 definitions. If such a declaration could appear after function
4057 definitions, the declaration would be too late to prevent the register from
4058 being used for other purposes in the preceding functions.
4060 Global register variables may not have initial values, because an
4061 executable file has no means to supply initial contents for a register.
4063 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4064 registers, but certain library functions, such as @code{getwd}, as well
4065 as the subroutines for division and remainder, modify g3 and g4. g1 and
4066 g2 are local temporaries.
4068 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4069 Of course, it will not do to use more than a few of those.
4071 @node Local Reg Vars
4072 @subsection Specifying Registers for Local Variables
4073 @cindex local variables, specifying registers
4074 @cindex specifying registers for local variables
4075 @cindex registers for local variables
4077 You can define a local register variable with a specified register
4081 register int *foo asm ("a5");
4085 Here @code{a5} is the name of the register which should be used. Note
4086 that this is the same syntax used for defining global register
4087 variables, but for a local variable it would appear within a function.
4089 Naturally the register name is cpu-dependent, but this is not a
4090 problem, since specific registers are most often useful with explicit
4091 assembler instructions (@pxref{Extended Asm}). Both of these things
4092 generally require that you conditionalize your program according to
4095 In addition, operating systems on one type of cpu may differ in how they
4096 name the registers; then you would need additional conditionals. For
4097 example, some 68000 operating systems call this register @code{%a5}.
4099 Defining such a register variable does not reserve the register; it
4100 remains available for other uses in places where flow control determines
4101 the variable's value is not live.
4103 This option does not guarantee that GCC will generate code that has
4104 this variable in the register you specify at all times. You may not
4105 code an explicit reference to this register in an @code{asm} statement
4106 and assume it will always refer to this variable.
4108 Stores into local register variables may be deleted when they appear to be dead
4109 according to dataflow analysis. References to local register variables may
4110 be deleted or moved or simplified.
4112 @node Alternate Keywords
4113 @section Alternate Keywords
4114 @cindex alternate keywords
4115 @cindex keywords, alternate
4117 @option{-ansi} and the various @option{-std} options disable certain
4118 keywords. This causes trouble when you want to use GNU C extensions, or
4119 a general-purpose header file that should be usable by all programs,
4120 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4121 @code{inline} are not available in programs compiled with
4122 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4123 program compiled with @option{-std=c99}). The ISO C99 keyword
4124 @code{restrict} is only available when @option{-std=gnu99} (which will
4125 eventually be the default) or @option{-std=c99} (or the equivalent
4126 @option{-std=iso9899:1999}) is used.
4128 The way to solve these problems is to put @samp{__} at the beginning and
4129 end of each problematical keyword. For example, use @code{__asm__}
4130 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4132 Other C compilers won't accept these alternative keywords; if you want to
4133 compile with another compiler, you can define the alternate keywords as
4134 macros to replace them with the customary keywords. It looks like this:
4142 @findex __extension__
4144 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4146 prevent such warnings within one expression by writing
4147 @code{__extension__} before the expression. @code{__extension__} has no
4148 effect aside from this.
4150 @node Incomplete Enums
4151 @section Incomplete @code{enum} Types
4153 You can define an @code{enum} tag without specifying its possible values.
4154 This results in an incomplete type, much like what you get if you write
4155 @code{struct foo} without describing the elements. A later declaration
4156 which does specify the possible values completes the type.
4158 You can't allocate variables or storage using the type while it is
4159 incomplete. However, you can work with pointers to that type.
4161 This extension may not be very useful, but it makes the handling of
4162 @code{enum} more consistent with the way @code{struct} and @code{union}
4165 This extension is not supported by GNU C++.
4167 @node Function Names
4168 @section Function Names as Strings
4169 @cindex @code{__func__} identifier
4170 @cindex @code{__FUNCTION__} identifier
4171 @cindex @code{__PRETTY_FUNCTION__} identifier
4173 GCC provides three magic variables which hold the name of the current
4174 function, as a string. The first of these is @code{__func__}, which
4175 is part of the C99 standard:
4178 The identifier @code{__func__} is implicitly declared by the translator
4179 as if, immediately following the opening brace of each function
4180 definition, the declaration
4183 static const char __func__[] = "function-name";
4186 appeared, where function-name is the name of the lexically-enclosing
4187 function. This name is the unadorned name of the function.
4190 @code{__FUNCTION__} is another name for @code{__func__}. Older
4191 versions of GCC recognize only this name. However, it is not
4192 standardized. For maximum portability, we recommend you use
4193 @code{__func__}, but provide a fallback definition with the
4197 #if __STDC_VERSION__ < 199901L
4199 # define __func__ __FUNCTION__
4201 # define __func__ "<unknown>"
4206 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4207 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4208 the type signature of the function as well as its bare name. For
4209 example, this program:
4213 extern int printf (char *, ...);
4220 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4221 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4239 __PRETTY_FUNCTION__ = void a::sub(int)
4242 These identifiers are not preprocessor macros. In GCC 3.3 and
4243 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4244 were treated as string literals; they could be used to initialize
4245 @code{char} arrays, and they could be concatenated with other string
4246 literals. GCC 3.4 and later treat them as variables, like
4247 @code{__func__}. In C++, @code{__FUNCTION__} and
4248 @code{__PRETTY_FUNCTION__} have always been variables.
4250 @node Return Address
4251 @section Getting the Return or Frame Address of a Function
4253 These functions may be used to get information about the callers of a
4256 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4257 This function returns the return address of the current function, or of
4258 one of its callers. The @var{level} argument is number of frames to
4259 scan up the call stack. A value of @code{0} yields the return address
4260 of the current function, a value of @code{1} yields the return address
4261 of the caller of the current function, and so forth. When inlining
4262 the expected behavior is that the function will return the address of
4263 the function that will be returned to. To work around this behavior use
4264 the @code{noinline} function attribute.
4266 The @var{level} argument must be a constant integer.
4268 On some machines it may be impossible to determine the return address of
4269 any function other than the current one; in such cases, or when the top
4270 of the stack has been reached, this function will return @code{0} or a
4271 random value. In addition, @code{__builtin_frame_address} may be used
4272 to determine if the top of the stack has been reached.
4274 This function should only be used with a nonzero argument for debugging
4278 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4279 This function is similar to @code{__builtin_return_address}, but it
4280 returns the address of the function frame rather than the return address
4281 of the function. Calling @code{__builtin_frame_address} with a value of
4282 @code{0} yields the frame address of the current function, a value of
4283 @code{1} yields the frame address of the caller of the current function,
4286 The frame is the area on the stack which holds local variables and saved
4287 registers. The frame address is normally the address of the first word
4288 pushed on to the stack by the function. However, the exact definition
4289 depends upon the processor and the calling convention. If the processor
4290 has a dedicated frame pointer register, and the function has a frame,
4291 then @code{__builtin_frame_address} will return the value of the frame
4294 On some machines it may be impossible to determine the frame address of
4295 any function other than the current one; in such cases, or when the top
4296 of the stack has been reached, this function will return @code{0} if
4297 the first frame pointer is properly initialized by the startup code.
4299 This function should only be used with a nonzero argument for debugging
4303 @node Vector Extensions
4304 @section Using vector instructions through built-in functions
4306 On some targets, the instruction set contains SIMD vector instructions that
4307 operate on multiple values contained in one large register at the same time.
4308 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4311 The first step in using these extensions is to provide the necessary data
4312 types. This should be done using an appropriate @code{typedef}:
4315 typedef int v4si __attribute__ ((vector_size (16)));
4318 The @code{int} type specifies the base type, while the attribute specifies
4319 the vector size for the variable, measured in bytes. For example, the
4320 declaration above causes the compiler to set the mode for the @code{v4si}
4321 type to be 16 bytes wide and divided into @code{int} sized units. For
4322 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4323 corresponding mode of @code{foo} will be @acronym{V4SI}.
4325 The @code{vector_size} attribute is only applicable to integral and
4326 float scalars, although arrays, pointers, and function return values
4327 are allowed in conjunction with this construct.
4329 All the basic integer types can be used as base types, both as signed
4330 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4331 @code{long long}. In addition, @code{float} and @code{double} can be
4332 used to build floating-point vector types.
4334 Specifying a combination that is not valid for the current architecture
4335 will cause GCC to synthesize the instructions using a narrower mode.
4336 For example, if you specify a variable of type @code{V4SI} and your
4337 architecture does not allow for this specific SIMD type, GCC will
4338 produce code that uses 4 @code{SIs}.
4340 The types defined in this manner can be used with a subset of normal C
4341 operations. Currently, GCC will allow using the following operators
4342 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4344 The operations behave like C++ @code{valarrays}. Addition is defined as
4345 the addition of the corresponding elements of the operands. For
4346 example, in the code below, each of the 4 elements in @var{a} will be
4347 added to the corresponding 4 elements in @var{b} and the resulting
4348 vector will be stored in @var{c}.
4351 typedef int v4si __attribute__ ((vector_size (16)));
4358 Subtraction, multiplication, division, and the logical operations
4359 operate in a similar manner. Likewise, the result of using the unary
4360 minus or complement operators on a vector type is a vector whose
4361 elements are the negative or complemented values of the corresponding
4362 elements in the operand.
4364 You can declare variables and use them in function calls and returns, as
4365 well as in assignments and some casts. You can specify a vector type as
4366 a return type for a function. Vector types can also be used as function
4367 arguments. It is possible to cast from one vector type to another,
4368 provided they are of the same size (in fact, you can also cast vectors
4369 to and from other datatypes of the same size).
4371 You cannot operate between vectors of different lengths or different
4372 signedness without a cast.
4374 A port that supports hardware vector operations, usually provides a set
4375 of built-in functions that can be used to operate on vectors. For
4376 example, a function to add two vectors and multiply the result by a
4377 third could look like this:
4380 v4si f (v4si a, v4si b, v4si c)
4382 v4si tmp = __builtin_addv4si (a, b);
4383 return __builtin_mulv4si (tmp, c);
4390 @findex __builtin_offsetof
4392 GCC implements for both C and C++ a syntactic extension to implement
4393 the @code{offsetof} macro.
4397 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
4399 offsetof_member_designator:
4401 | offsetof_member_designator "." @code{identifier}
4402 | offsetof_member_designator "[" @code{expr} "]"
4405 This extension is sufficient such that
4408 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
4411 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
4412 may be dependent. In either case, @var{member} may consist of a single
4413 identifier, or a sequence of member accesses and array references.
4415 @node Other Builtins
4416 @section Other built-in functions provided by GCC
4417 @cindex built-in functions
4418 @findex __builtin_isgreater
4419 @findex __builtin_isgreaterequal
4420 @findex __builtin_isless
4421 @findex __builtin_islessequal
4422 @findex __builtin_islessgreater
4423 @findex __builtin_isunordered
4578 @findex fprintf_unlocked
4580 @findex fputs_unlocked
4690 @findex printf_unlocked
4719 @findex significandf
4720 @findex significandl
4787 GCC provides a large number of built-in functions other than the ones
4788 mentioned above. Some of these are for internal use in the processing
4789 of exceptions or variable-length argument lists and will not be
4790 documented here because they may change from time to time; we do not
4791 recommend general use of these functions.
4793 The remaining functions are provided for optimization purposes.
4795 @opindex fno-builtin
4796 GCC includes built-in versions of many of the functions in the standard
4797 C library. The versions prefixed with @code{__builtin_} will always be
4798 treated as having the same meaning as the C library function even if you
4799 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
4800 Many of these functions are only optimized in certain cases; if they are
4801 not optimized in a particular case, a call to the library function will
4806 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
4807 @option{-std=c99}), the functions
4808 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
4809 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
4810 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
4811 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
4812 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
4813 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
4814 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
4815 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
4816 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
4817 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
4818 @code{significandf}, @code{significandl}, @code{significand},
4819 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
4820 @code{strdup}, @code{strfmon}, @code{toascii}, @code{y0f}, @code{y0l},
4821 @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
4823 may be handled as built-in functions.
4824 All these functions have corresponding versions
4825 prefixed with @code{__builtin_}, which may be used even in strict C89
4828 The ISO C99 functions
4829 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
4830 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
4831 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
4832 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
4833 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
4834 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
4835 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
4836 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
4837 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
4838 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
4839 @code{cimagl}, @code{cimag}, @code{conjf}, @code{conjl}, @code{conj},
4840 @code{copysignf}, @code{copysignl}, @code{copysign}, @code{cpowf},
4841 @code{cpowl}, @code{cpow}, @code{cprojf}, @code{cprojl}, @code{cproj},
4842 @code{crealf}, @code{creall}, @code{creal}, @code{csinf}, @code{csinhf},
4843 @code{csinhl}, @code{csinh}, @code{csinl}, @code{csin}, @code{csqrtf},
4844 @code{csqrtl}, @code{csqrt}, @code{ctanf}, @code{ctanhf}, @code{ctanhl},
4845 @code{ctanh}, @code{ctanl}, @code{ctan}, @code{erfcf}, @code{erfcl},
4846 @code{erfc}, @code{erff}, @code{erfl}, @code{erf}, @code{exp2f},
4847 @code{exp2l}, @code{exp2}, @code{expm1f}, @code{expm1l}, @code{expm1},
4848 @code{fdimf}, @code{fdiml}, @code{fdim}, @code{fmaf}, @code{fmal},
4849 @code{fmaxf}, @code{fmaxl}, @code{fmax}, @code{fma}, @code{fminf},
4850 @code{fminl}, @code{fmin}, @code{hypotf}, @code{hypotl}, @code{hypot},
4851 @code{ilogbf}, @code{ilogbl}, @code{ilogb}, @code{imaxabs},
4852 @code{isblank}, @code{iswblank}, @code{lgammaf}, @code{lgammal},
4853 @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
4854 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
4855 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
4856 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
4857 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
4858 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
4859 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
4860 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
4861 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
4862 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
4863 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
4864 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
4865 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
4866 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
4867 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
4868 are handled as built-in functions
4869 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
4871 There are also built-in versions of the ISO C99 functions
4872 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
4873 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
4874 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
4875 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
4876 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
4877 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
4878 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
4879 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
4880 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
4881 that are recognized in any mode since ISO C90 reserves these names for
4882 the purpose to which ISO C99 puts them. All these functions have
4883 corresponding versions prefixed with @code{__builtin_}.
4885 The ISO C94 functions
4886 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
4887 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
4888 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
4890 are handled as built-in functions
4891 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
4893 The ISO C90 functions
4894 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
4895 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
4896 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
4897 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
4898 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
4899 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
4900 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
4901 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
4902 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
4903 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
4904 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
4905 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
4906 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
4907 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
4908 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
4909 @code{vprintf} and @code{vsprintf}
4910 are all recognized as built-in functions unless
4911 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
4912 is specified for an individual function). All of these functions have
4913 corresponding versions prefixed with @code{__builtin_}.
4915 GCC provides built-in versions of the ISO C99 floating point comparison
4916 macros that avoid raising exceptions for unordered operands. They have
4917 the same names as the standard macros ( @code{isgreater},
4918 @code{isgreaterequal}, @code{isless}, @code{islessequal},
4919 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
4920 prefixed. We intend for a library implementor to be able to simply
4921 @code{#define} each standard macro to its built-in equivalent.
4923 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
4925 You can use the built-in function @code{__builtin_types_compatible_p} to
4926 determine whether two types are the same.
4928 This built-in function returns 1 if the unqualified versions of the
4929 types @var{type1} and @var{type2} (which are types, not expressions) are
4930 compatible, 0 otherwise. The result of this built-in function can be
4931 used in integer constant expressions.
4933 This built-in function ignores top level qualifiers (e.g., @code{const},
4934 @code{volatile}). For example, @code{int} is equivalent to @code{const
4937 The type @code{int[]} and @code{int[5]} are compatible. On the other
4938 hand, @code{int} and @code{char *} are not compatible, even if the size
4939 of their types, on the particular architecture are the same. Also, the
4940 amount of pointer indirection is taken into account when determining
4941 similarity. Consequently, @code{short *} is not similar to
4942 @code{short **}. Furthermore, two types that are typedefed are
4943 considered compatible if their underlying types are compatible.
4945 An @code{enum} type is not considered to be compatible with another
4946 @code{enum} type even if both are compatible with the same integer
4947 type; this is what the C standard specifies.
4948 For example, @code{enum @{foo, bar@}} is not similar to
4949 @code{enum @{hot, dog@}}.
4951 You would typically use this function in code whose execution varies
4952 depending on the arguments' types. For example:
4958 if (__builtin_types_compatible_p (typeof (x), long double)) \
4959 tmp = foo_long_double (tmp); \
4960 else if (__builtin_types_compatible_p (typeof (x), double)) \
4961 tmp = foo_double (tmp); \
4962 else if (__builtin_types_compatible_p (typeof (x), float)) \
4963 tmp = foo_float (tmp); \
4970 @emph{Note:} This construct is only available for C.
4974 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
4976 You can use the built-in function @code{__builtin_choose_expr} to
4977 evaluate code depending on the value of a constant expression. This
4978 built-in function returns @var{exp1} if @var{const_exp}, which is a
4979 constant expression that must be able to be determined at compile time,
4980 is nonzero. Otherwise it returns 0.
4982 This built-in function is analogous to the @samp{? :} operator in C,
4983 except that the expression returned has its type unaltered by promotion
4984 rules. Also, the built-in function does not evaluate the expression
4985 that was not chosen. For example, if @var{const_exp} evaluates to true,
4986 @var{exp2} is not evaluated even if it has side-effects.
4988 This built-in function can return an lvalue if the chosen argument is an
4991 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
4992 type. Similarly, if @var{exp2} is returned, its return type is the same
4999 __builtin_choose_expr ( \
5000 __builtin_types_compatible_p (typeof (x), double), \
5002 __builtin_choose_expr ( \
5003 __builtin_types_compatible_p (typeof (x), float), \
5005 /* @r{The void expression results in a compile-time error} \
5006 @r{when assigning the result to something.} */ \
5010 @emph{Note:} This construct is only available for C. Furthermore, the
5011 unused expression (@var{exp1} or @var{exp2} depending on the value of
5012 @var{const_exp}) may still generate syntax errors. This may change in
5017 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5018 You can use the built-in function @code{__builtin_constant_p} to
5019 determine if a value is known to be constant at compile-time and hence
5020 that GCC can perform constant-folding on expressions involving that
5021 value. The argument of the function is the value to test. The function
5022 returns the integer 1 if the argument is known to be a compile-time
5023 constant and 0 if it is not known to be a compile-time constant. A
5024 return of 0 does not indicate that the value is @emph{not} a constant,
5025 but merely that GCC cannot prove it is a constant with the specified
5026 value of the @option{-O} option.
5028 You would typically use this function in an embedded application where
5029 memory was a critical resource. If you have some complex calculation,
5030 you may want it to be folded if it involves constants, but need to call
5031 a function if it does not. For example:
5034 #define Scale_Value(X) \
5035 (__builtin_constant_p (X) \
5036 ? ((X) * SCALE + OFFSET) : Scale (X))
5039 You may use this built-in function in either a macro or an inline
5040 function. However, if you use it in an inlined function and pass an
5041 argument of the function as the argument to the built-in, GCC will
5042 never return 1 when you call the inline function with a string constant
5043 or compound literal (@pxref{Compound Literals}) and will not return 1
5044 when you pass a constant numeric value to the inline function unless you
5045 specify the @option{-O} option.
5047 You may also use @code{__builtin_constant_p} in initializers for static
5048 data. For instance, you can write
5051 static const int table[] = @{
5052 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5058 This is an acceptable initializer even if @var{EXPRESSION} is not a
5059 constant expression. GCC must be more conservative about evaluating the
5060 built-in in this case, because it has no opportunity to perform
5063 Previous versions of GCC did not accept this built-in in data
5064 initializers. The earliest version where it is completely safe is
5068 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5069 @opindex fprofile-arcs
5070 You may use @code{__builtin_expect} to provide the compiler with
5071 branch prediction information. In general, you should prefer to
5072 use actual profile feedback for this (@option{-fprofile-arcs}), as
5073 programmers are notoriously bad at predicting how their programs
5074 actually perform. However, there are applications in which this
5075 data is hard to collect.
5077 The return value is the value of @var{exp}, which should be an
5078 integral expression. The value of @var{c} must be a compile-time
5079 constant. The semantics of the built-in are that it is expected
5080 that @var{exp} == @var{c}. For example:
5083 if (__builtin_expect (x, 0))
5088 would indicate that we do not expect to call @code{foo}, since
5089 we expect @code{x} to be zero. Since you are limited to integral
5090 expressions for @var{exp}, you should use constructions such as
5093 if (__builtin_expect (ptr != NULL, 1))
5098 when testing pointer or floating-point values.
5101 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5102 This function is used to minimize cache-miss latency by moving data into
5103 a cache before it is accessed.
5104 You can insert calls to @code{__builtin_prefetch} into code for which
5105 you know addresses of data in memory that is likely to be accessed soon.
5106 If the target supports them, data prefetch instructions will be generated.
5107 If the prefetch is done early enough before the access then the data will
5108 be in the cache by the time it is accessed.
5110 The value of @var{addr} is the address of the memory to prefetch.
5111 There are two optional arguments, @var{rw} and @var{locality}.
5112 The value of @var{rw} is a compile-time constant one or zero; one
5113 means that the prefetch is preparing for a write to the memory address
5114 and zero, the default, means that the prefetch is preparing for a read.
5115 The value @var{locality} must be a compile-time constant integer between
5116 zero and three. A value of zero means that the data has no temporal
5117 locality, so it need not be left in the cache after the access. A value
5118 of three means that the data has a high degree of temporal locality and
5119 should be left in all levels of cache possible. Values of one and two
5120 mean, respectively, a low or moderate degree of temporal locality. The
5124 for (i = 0; i < n; i++)
5127 __builtin_prefetch (&a[i+j], 1, 1);
5128 __builtin_prefetch (&b[i+j], 0, 1);
5133 Data prefetch does not generate faults if @var{addr} is invalid, but
5134 the address expression itself must be valid. For example, a prefetch
5135 of @code{p->next} will not fault if @code{p->next} is not a valid
5136 address, but evaluation will fault if @code{p} is not a valid address.
5138 If the target does not support data prefetch, the address expression
5139 is evaluated if it includes side effects but no other code is generated
5140 and GCC does not issue a warning.
5143 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5144 Returns a positive infinity, if supported by the floating-point format,
5145 else @code{DBL_MAX}. This function is suitable for implementing the
5146 ISO C macro @code{HUGE_VAL}.
5149 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5150 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5153 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5154 Similar to @code{__builtin_huge_val}, except the return
5155 type is @code{long double}.
5158 @deftypefn {Built-in Function} double __builtin_inf (void)
5159 Similar to @code{__builtin_huge_val}, except a warning is generated
5160 if the target floating-point format does not support infinities.
5161 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5164 @deftypefn {Built-in Function} float __builtin_inff (void)
5165 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5168 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5169 Similar to @code{__builtin_inf}, except the return
5170 type is @code{long double}.
5173 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5174 This is an implementation of the ISO C99 function @code{nan}.
5176 Since ISO C99 defines this function in terms of @code{strtod}, which we
5177 do not implement, a description of the parsing is in order. The string
5178 is parsed as by @code{strtol}; that is, the base is recognized by
5179 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5180 in the significand such that the least significant bit of the number
5181 is at the least significant bit of the significand. The number is
5182 truncated to fit the significand field provided. The significand is
5183 forced to be a quiet NaN.
5185 This function, if given a string literal, is evaluated early enough
5186 that it is considered a compile-time constant.
5189 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
5190 Similar to @code{__builtin_nan}, except the return type is @code{float}.
5193 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
5194 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
5197 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
5198 Similar to @code{__builtin_nan}, except the significand is forced
5199 to be a signaling NaN. The @code{nans} function is proposed by
5200 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
5203 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
5204 Similar to @code{__builtin_nans}, except the return type is @code{float}.
5207 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
5208 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
5211 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
5212 Returns one plus the index of the least significant 1-bit of @var{x}, or
5213 if @var{x} is zero, returns zero.
5216 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
5217 Returns the number of leading 0-bits in @var{x}, starting at the most
5218 significant bit position. If @var{x} is 0, the result is undefined.
5221 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
5222 Returns the number of trailing 0-bits in @var{x}, starting at the least
5223 significant bit position. If @var{x} is 0, the result is undefined.
5226 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
5227 Returns the number of 1-bits in @var{x}.
5230 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
5231 Returns the parity of @var{x}, i.@:e. the number of 1-bits in @var{x}
5235 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
5236 Similar to @code{__builtin_ffs}, except the argument type is
5237 @code{unsigned long}.
5240 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
5241 Similar to @code{__builtin_clz}, except the argument type is
5242 @code{unsigned long}.
5245 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
5246 Similar to @code{__builtin_ctz}, except the argument type is
5247 @code{unsigned long}.
5250 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
5251 Similar to @code{__builtin_popcount}, except the argument type is
5252 @code{unsigned long}.
5255 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
5256 Similar to @code{__builtin_parity}, except the argument type is
5257 @code{unsigned long}.
5260 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
5261 Similar to @code{__builtin_ffs}, except the argument type is
5262 @code{unsigned long long}.
5265 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
5266 Similar to @code{__builtin_clz}, except the argument type is
5267 @code{unsigned long long}.
5270 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
5271 Similar to @code{__builtin_ctz}, except the argument type is
5272 @code{unsigned long long}.
5275 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
5276 Similar to @code{__builtin_popcount}, except the argument type is
5277 @code{unsigned long long}.
5280 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
5281 Similar to @code{__builtin_parity}, except the argument type is
5282 @code{unsigned long long}.
5286 @node Target Builtins
5287 @section Built-in Functions Specific to Particular Target Machines
5289 On some target machines, GCC supports many built-in functions specific
5290 to those machines. Generally these generate calls to specific machine
5291 instructions, but allow the compiler to schedule those calls.
5294 * Alpha Built-in Functions::
5295 * ARM Built-in Functions::
5296 * X86 Built-in Functions::
5297 * PowerPC AltiVec Built-in Functions::
5300 @node Alpha Built-in Functions
5301 @subsection Alpha Built-in Functions
5303 These built-in functions are available for the Alpha family of
5304 processors, depending on the command-line switches used.
5306 The following built-in functions are always available. They
5307 all generate the machine instruction that is part of the name.
5310 long __builtin_alpha_implver (void)
5311 long __builtin_alpha_rpcc (void)
5312 long __builtin_alpha_amask (long)
5313 long __builtin_alpha_cmpbge (long, long)
5314 long __builtin_alpha_extbl (long, long)
5315 long __builtin_alpha_extwl (long, long)
5316 long __builtin_alpha_extll (long, long)
5317 long __builtin_alpha_extql (long, long)
5318 long __builtin_alpha_extwh (long, long)
5319 long __builtin_alpha_extlh (long, long)
5320 long __builtin_alpha_extqh (long, long)
5321 long __builtin_alpha_insbl (long, long)
5322 long __builtin_alpha_inswl (long, long)
5323 long __builtin_alpha_insll (long, long)
5324 long __builtin_alpha_insql (long, long)
5325 long __builtin_alpha_inswh (long, long)
5326 long __builtin_alpha_inslh (long, long)
5327 long __builtin_alpha_insqh (long, long)
5328 long __builtin_alpha_mskbl (long, long)
5329 long __builtin_alpha_mskwl (long, long)
5330 long __builtin_alpha_mskll (long, long)
5331 long __builtin_alpha_mskql (long, long)
5332 long __builtin_alpha_mskwh (long, long)
5333 long __builtin_alpha_msklh (long, long)
5334 long __builtin_alpha_mskqh (long, long)
5335 long __builtin_alpha_umulh (long, long)
5336 long __builtin_alpha_zap (long, long)
5337 long __builtin_alpha_zapnot (long, long)
5340 The following built-in functions are always with @option{-mmax}
5341 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
5342 later. They all generate the machine instruction that is part
5346 long __builtin_alpha_pklb (long)
5347 long __builtin_alpha_pkwb (long)
5348 long __builtin_alpha_unpkbl (long)
5349 long __builtin_alpha_unpkbw (long)
5350 long __builtin_alpha_minub8 (long, long)
5351 long __builtin_alpha_minsb8 (long, long)
5352 long __builtin_alpha_minuw4 (long, long)
5353 long __builtin_alpha_minsw4 (long, long)
5354 long __builtin_alpha_maxub8 (long, long)
5355 long __builtin_alpha_maxsb8 (long, long)
5356 long __builtin_alpha_maxuw4 (long, long)
5357 long __builtin_alpha_maxsw4 (long, long)
5358 long __builtin_alpha_perr (long, long)
5361 The following built-in functions are always with @option{-mcix}
5362 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
5363 later. They all generate the machine instruction that is part
5367 long __builtin_alpha_cttz (long)
5368 long __builtin_alpha_ctlz (long)
5369 long __builtin_alpha_ctpop (long)
5372 The following builtins are available on systems that use the OSF/1
5373 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
5374 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
5375 @code{rdval} and @code{wrval}.
5378 void *__builtin_thread_pointer (void)
5379 void __builtin_set_thread_pointer (void *)
5382 @node ARM Built-in Functions
5383 @subsection ARM Built-in Functions
5385 These built-in functions are available for the ARM family of
5386 processors, when the @option{-mcpu=iwmmxt} switch is used:
5389 typedef int v2si __attribute__ ((vector_size (8)));
5390 typedef short v4hi __attribute__ ((vector_size (8)));
5391 typedef char v8qi __attribute__ ((vector_size (8)));
5393 int __builtin_arm_getwcx (int)
5394 void __builtin_arm_setwcx (int, int)
5395 int __builtin_arm_textrmsb (v8qi, int)
5396 int __builtin_arm_textrmsh (v4hi, int)
5397 int __builtin_arm_textrmsw (v2si, int)
5398 int __builtin_arm_textrmub (v8qi, int)
5399 int __builtin_arm_textrmuh (v4hi, int)
5400 int __builtin_arm_textrmuw (v2si, int)
5401 v8qi __builtin_arm_tinsrb (v8qi, int)
5402 v4hi __builtin_arm_tinsrh (v4hi, int)
5403 v2si __builtin_arm_tinsrw (v2si, int)
5404 long long __builtin_arm_tmia (long long, int, int)
5405 long long __builtin_arm_tmiabb (long long, int, int)
5406 long long __builtin_arm_tmiabt (long long, int, int)
5407 long long __builtin_arm_tmiaph (long long, int, int)
5408 long long __builtin_arm_tmiatb (long long, int, int)
5409 long long __builtin_arm_tmiatt (long long, int, int)
5410 int __builtin_arm_tmovmskb (v8qi)
5411 int __builtin_arm_tmovmskh (v4hi)
5412 int __builtin_arm_tmovmskw (v2si)
5413 long long __builtin_arm_waccb (v8qi)
5414 long long __builtin_arm_wacch (v4hi)
5415 long long __builtin_arm_waccw (v2si)
5416 v8qi __builtin_arm_waddb (v8qi, v8qi)
5417 v8qi __builtin_arm_waddbss (v8qi, v8qi)
5418 v8qi __builtin_arm_waddbus (v8qi, v8qi)
5419 v4hi __builtin_arm_waddh (v4hi, v4hi)
5420 v4hi __builtin_arm_waddhss (v4hi, v4hi)
5421 v4hi __builtin_arm_waddhus (v4hi, v4hi)
5422 v2si __builtin_arm_waddw (v2si, v2si)
5423 v2si __builtin_arm_waddwss (v2si, v2si)
5424 v2si __builtin_arm_waddwus (v2si, v2si)
5425 v8qi __builtin_arm_walign (v8qi, v8qi, int)
5426 long long __builtin_arm_wand(long long, long long)
5427 long long __builtin_arm_wandn (long long, long long)
5428 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
5429 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
5430 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
5431 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
5432 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
5433 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
5434 v2si __builtin_arm_wcmpeqw (v2si, v2si)
5435 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
5436 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
5437 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
5438 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
5439 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
5440 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
5441 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
5442 long long __builtin_arm_wmacsz (v4hi, v4hi)
5443 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
5444 long long __builtin_arm_wmacuz (v4hi, v4hi)
5445 v4hi __builtin_arm_wmadds (v4hi, v4hi)
5446 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
5447 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
5448 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
5449 v2si __builtin_arm_wmaxsw (v2si, v2si)
5450 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
5451 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
5452 v2si __builtin_arm_wmaxuw (v2si, v2si)
5453 v8qi __builtin_arm_wminsb (v8qi, v8qi)
5454 v4hi __builtin_arm_wminsh (v4hi, v4hi)
5455 v2si __builtin_arm_wminsw (v2si, v2si)
5456 v8qi __builtin_arm_wminub (v8qi, v8qi)
5457 v4hi __builtin_arm_wminuh (v4hi, v4hi)
5458 v2si __builtin_arm_wminuw (v2si, v2si)
5459 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
5460 v4hi __builtin_arm_wmulul (v4hi, v4hi)
5461 v4hi __builtin_arm_wmulum (v4hi, v4hi)
5462 long long __builtin_arm_wor (long long, long long)
5463 v2si __builtin_arm_wpackdss (long long, long long)
5464 v2si __builtin_arm_wpackdus (long long, long long)
5465 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
5466 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
5467 v4hi __builtin_arm_wpackwss (v2si, v2si)
5468 v4hi __builtin_arm_wpackwus (v2si, v2si)
5469 long long __builtin_arm_wrord (long long, long long)
5470 long long __builtin_arm_wrordi (long long, int)
5471 v4hi __builtin_arm_wrorh (v4hi, long long)
5472 v4hi __builtin_arm_wrorhi (v4hi, int)
5473 v2si __builtin_arm_wrorw (v2si, long long)
5474 v2si __builtin_arm_wrorwi (v2si, int)
5475 v2si __builtin_arm_wsadb (v8qi, v8qi)
5476 v2si __builtin_arm_wsadbz (v8qi, v8qi)
5477 v2si __builtin_arm_wsadh (v4hi, v4hi)
5478 v2si __builtin_arm_wsadhz (v4hi, v4hi)
5479 v4hi __builtin_arm_wshufh (v4hi, int)
5480 long long __builtin_arm_wslld (long long, long long)
5481 long long __builtin_arm_wslldi (long long, int)
5482 v4hi __builtin_arm_wsllh (v4hi, long long)
5483 v4hi __builtin_arm_wsllhi (v4hi, int)
5484 v2si __builtin_arm_wsllw (v2si, long long)
5485 v2si __builtin_arm_wsllwi (v2si, int)
5486 long long __builtin_arm_wsrad (long long, long long)
5487 long long __builtin_arm_wsradi (long long, int)
5488 v4hi __builtin_arm_wsrah (v4hi, long long)
5489 v4hi __builtin_arm_wsrahi (v4hi, int)
5490 v2si __builtin_arm_wsraw (v2si, long long)
5491 v2si __builtin_arm_wsrawi (v2si, int)
5492 long long __builtin_arm_wsrld (long long, long long)
5493 long long __builtin_arm_wsrldi (long long, int)
5494 v4hi __builtin_arm_wsrlh (v4hi, long long)
5495 v4hi __builtin_arm_wsrlhi (v4hi, int)
5496 v2si __builtin_arm_wsrlw (v2si, long long)
5497 v2si __builtin_arm_wsrlwi (v2si, int)
5498 v8qi __builtin_arm_wsubb (v8qi, v8qi)
5499 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
5500 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
5501 v4hi __builtin_arm_wsubh (v4hi, v4hi)
5502 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
5503 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
5504 v2si __builtin_arm_wsubw (v2si, v2si)
5505 v2si __builtin_arm_wsubwss (v2si, v2si)
5506 v2si __builtin_arm_wsubwus (v2si, v2si)
5507 v4hi __builtin_arm_wunpckehsb (v8qi)
5508 v2si __builtin_arm_wunpckehsh (v4hi)
5509 long long __builtin_arm_wunpckehsw (v2si)
5510 v4hi __builtin_arm_wunpckehub (v8qi)
5511 v2si __builtin_arm_wunpckehuh (v4hi)
5512 long long __builtin_arm_wunpckehuw (v2si)
5513 v4hi __builtin_arm_wunpckelsb (v8qi)
5514 v2si __builtin_arm_wunpckelsh (v4hi)
5515 long long __builtin_arm_wunpckelsw (v2si)
5516 v4hi __builtin_arm_wunpckelub (v8qi)
5517 v2si __builtin_arm_wunpckeluh (v4hi)
5518 long long __builtin_arm_wunpckeluw (v2si)
5519 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
5520 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
5521 v2si __builtin_arm_wunpckihw (v2si, v2si)
5522 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
5523 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
5524 v2si __builtin_arm_wunpckilw (v2si, v2si)
5525 long long __builtin_arm_wxor (long long, long long)
5526 long long __builtin_arm_wzero ()
5529 @node X86 Built-in Functions
5530 @subsection X86 Built-in Functions
5532 These built-in functions are available for the i386 and x86-64 family
5533 of computers, depending on the command-line switches used.
5535 The following machine modes are available for use with MMX built-in functions
5536 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
5537 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
5538 vector of eight 8-bit integers. Some of the built-in functions operate on
5539 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
5541 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
5542 of two 32-bit floating point values.
5544 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
5545 floating point values. Some instructions use a vector of four 32-bit
5546 integers, these use @code{V4SI}. Finally, some instructions operate on an
5547 entire vector register, interpreting it as a 128-bit integer, these use mode
5550 The following built-in functions are made available by @option{-mmmx}.
5551 All of them generate the machine instruction that is part of the name.
5554 v8qi __builtin_ia32_paddb (v8qi, v8qi)
5555 v4hi __builtin_ia32_paddw (v4hi, v4hi)
5556 v2si __builtin_ia32_paddd (v2si, v2si)
5557 v8qi __builtin_ia32_psubb (v8qi, v8qi)
5558 v4hi __builtin_ia32_psubw (v4hi, v4hi)
5559 v2si __builtin_ia32_psubd (v2si, v2si)
5560 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
5561 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
5562 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
5563 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
5564 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
5565 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
5566 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
5567 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
5568 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
5569 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
5570 di __builtin_ia32_pand (di, di)
5571 di __builtin_ia32_pandn (di,di)
5572 di __builtin_ia32_por (di, di)
5573 di __builtin_ia32_pxor (di, di)
5574 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
5575 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
5576 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
5577 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
5578 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
5579 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
5580 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
5581 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
5582 v2si __builtin_ia32_punpckhdq (v2si, v2si)
5583 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
5584 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
5585 v2si __builtin_ia32_punpckldq (v2si, v2si)
5586 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
5587 v4hi __builtin_ia32_packssdw (v2si, v2si)
5588 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
5591 The following built-in functions are made available either with
5592 @option{-msse}, or with a combination of @option{-m3dnow} and
5593 @option{-march=athlon}. All of them generate the machine
5594 instruction that is part of the name.
5597 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
5598 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
5599 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
5600 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
5601 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
5602 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
5603 v8qi __builtin_ia32_pminub (v8qi, v8qi)
5604 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
5605 int __builtin_ia32_pextrw (v4hi, int)
5606 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
5607 int __builtin_ia32_pmovmskb (v8qi)
5608 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
5609 void __builtin_ia32_movntq (di *, di)
5610 void __builtin_ia32_sfence (void)
5613 The following built-in functions are available when @option{-msse} is used.
5614 All of them generate the machine instruction that is part of the name.
5617 int __builtin_ia32_comieq (v4sf, v4sf)
5618 int __builtin_ia32_comineq (v4sf, v4sf)
5619 int __builtin_ia32_comilt (v4sf, v4sf)
5620 int __builtin_ia32_comile (v4sf, v4sf)
5621 int __builtin_ia32_comigt (v4sf, v4sf)
5622 int __builtin_ia32_comige (v4sf, v4sf)
5623 int __builtin_ia32_ucomieq (v4sf, v4sf)
5624 int __builtin_ia32_ucomineq (v4sf, v4sf)
5625 int __builtin_ia32_ucomilt (v4sf, v4sf)
5626 int __builtin_ia32_ucomile (v4sf, v4sf)
5627 int __builtin_ia32_ucomigt (v4sf, v4sf)
5628 int __builtin_ia32_ucomige (v4sf, v4sf)
5629 v4sf __builtin_ia32_addps (v4sf, v4sf)
5630 v4sf __builtin_ia32_subps (v4sf, v4sf)
5631 v4sf __builtin_ia32_mulps (v4sf, v4sf)
5632 v4sf __builtin_ia32_divps (v4sf, v4sf)
5633 v4sf __builtin_ia32_addss (v4sf, v4sf)
5634 v4sf __builtin_ia32_subss (v4sf, v4sf)
5635 v4sf __builtin_ia32_mulss (v4sf, v4sf)
5636 v4sf __builtin_ia32_divss (v4sf, v4sf)
5637 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
5638 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
5639 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
5640 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
5641 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
5642 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
5643 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
5644 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
5645 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
5646 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
5647 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
5648 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
5649 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
5650 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
5651 v4si __builtin_ia32_cmpless (v4sf, v4sf)
5652 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
5653 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
5654 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
5655 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
5656 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
5657 v4sf __builtin_ia32_maxps (v4sf, v4sf)
5658 v4sf __builtin_ia32_maxss (v4sf, v4sf)
5659 v4sf __builtin_ia32_minps (v4sf, v4sf)
5660 v4sf __builtin_ia32_minss (v4sf, v4sf)
5661 v4sf __builtin_ia32_andps (v4sf, v4sf)
5662 v4sf __builtin_ia32_andnps (v4sf, v4sf)
5663 v4sf __builtin_ia32_orps (v4sf, v4sf)
5664 v4sf __builtin_ia32_xorps (v4sf, v4sf)
5665 v4sf __builtin_ia32_movss (v4sf, v4sf)
5666 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
5667 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
5668 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
5669 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
5670 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
5671 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
5672 v2si __builtin_ia32_cvtps2pi (v4sf)
5673 int __builtin_ia32_cvtss2si (v4sf)
5674 v2si __builtin_ia32_cvttps2pi (v4sf)
5675 int __builtin_ia32_cvttss2si (v4sf)
5676 v4sf __builtin_ia32_rcpps (v4sf)
5677 v4sf __builtin_ia32_rsqrtps (v4sf)
5678 v4sf __builtin_ia32_sqrtps (v4sf)
5679 v4sf __builtin_ia32_rcpss (v4sf)
5680 v4sf __builtin_ia32_rsqrtss (v4sf)
5681 v4sf __builtin_ia32_sqrtss (v4sf)
5682 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
5683 void __builtin_ia32_movntps (float *, v4sf)
5684 int __builtin_ia32_movmskps (v4sf)
5687 The following built-in functions are available when @option{-msse} is used.
5690 @item v4sf __builtin_ia32_loadaps (float *)
5691 Generates the @code{movaps} machine instruction as a load from memory.
5692 @item void __builtin_ia32_storeaps (float *, v4sf)
5693 Generates the @code{movaps} machine instruction as a store to memory.
5694 @item v4sf __builtin_ia32_loadups (float *)
5695 Generates the @code{movups} machine instruction as a load from memory.
5696 @item void __builtin_ia32_storeups (float *, v4sf)
5697 Generates the @code{movups} machine instruction as a store to memory.
5698 @item v4sf __builtin_ia32_loadsss (float *)
5699 Generates the @code{movss} machine instruction as a load from memory.
5700 @item void __builtin_ia32_storess (float *, v4sf)
5701 Generates the @code{movss} machine instruction as a store to memory.
5702 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
5703 Generates the @code{movhps} machine instruction as a load from memory.
5704 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
5705 Generates the @code{movlps} machine instruction as a load from memory
5706 @item void __builtin_ia32_storehps (v4sf, v2si *)
5707 Generates the @code{movhps} machine instruction as a store to memory.
5708 @item void __builtin_ia32_storelps (v4sf, v2si *)
5709 Generates the @code{movlps} machine instruction as a store to memory.
5712 The following built-in functions are available when @option{-msse3} is used.
5713 All of them generate the machine instruction that is part of the name.
5716 v2df __builtin_ia32_addsubpd (v2df, v2df)
5717 v2df __builtin_ia32_addsubps (v2df, v2df)
5718 v2df __builtin_ia32_haddpd (v2df, v2df)
5719 v2df __builtin_ia32_haddps (v2df, v2df)
5720 v2df __builtin_ia32_hsubpd (v2df, v2df)
5721 v2df __builtin_ia32_hsubps (v2df, v2df)
5722 v16qi __builtin_ia32_lddqu (char const *)
5723 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
5724 v2df __builtin_ia32_movddup (v2df)
5725 v4sf __builtin_ia32_movshdup (v4sf)
5726 v4sf __builtin_ia32_movsldup (v4sf)
5727 void __builtin_ia32_mwait (unsigned int, unsigned int)
5730 The following built-in functions are available when @option{-msse3} is used.
5733 @item v2df __builtin_ia32_loadddup (double const *)
5734 Generates the @code{movddup} machine instruction as a load from memory.
5737 The following built-in functions are available when @option{-m3dnow} is used.
5738 All of them generate the machine instruction that is part of the name.
5741 void __builtin_ia32_femms (void)
5742 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
5743 v2si __builtin_ia32_pf2id (v2sf)
5744 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
5745 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
5746 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
5747 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
5748 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
5749 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
5750 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
5751 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
5752 v2sf __builtin_ia32_pfrcp (v2sf)
5753 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
5754 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
5755 v2sf __builtin_ia32_pfrsqrt (v2sf)
5756 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
5757 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
5758 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
5759 v2sf __builtin_ia32_pi2fd (v2si)
5760 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
5763 The following built-in functions are available when both @option{-m3dnow}
5764 and @option{-march=athlon} are used. All of them generate the machine
5765 instruction that is part of the name.
5768 v2si __builtin_ia32_pf2iw (v2sf)
5769 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
5770 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
5771 v2sf __builtin_ia32_pi2fw (v2si)
5772 v2sf __builtin_ia32_pswapdsf (v2sf)
5773 v2si __builtin_ia32_pswapdsi (v2si)
5776 @node PowerPC AltiVec Built-in Functions
5777 @subsection PowerPC AltiVec Built-in Functions
5779 GCC provides an interface for the PowerPC family of processors to access
5780 the AltiVec operations described in Motorola's AltiVec Programming
5781 Interface Manual. The interface is made available by including
5782 @code{<altivec.h>} and using @option{-maltivec} and
5783 @option{-mabi=altivec}. The interface supports the following vector
5787 vector unsigned char
5791 vector unsigned short
5802 GCC's implementation of the high-level language interface available from
5803 C and C++ code differs from Motorola's documentation in several ways.
5808 A vector constant is a list of constant expressions within curly braces.
5811 A vector initializer requires no cast if the vector constant is of the
5812 same type as the variable it is initializing.
5815 If @code{signed} or @code{unsigned} is omitted, the vector type defaults
5816 to @code{signed} for @code{vector int} or @code{vector short} and to
5817 @code{unsigned} for @code{vector char}.
5820 Compiling with @option{-maltivec} adds keywords @code{__vector},
5821 @code{__pixel}, and @code{__bool}. Macros @option{vector},
5822 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
5826 GCC allows using a @code{typedef} name as the type specifier for a
5830 For C, overloaded functions are implemented with macros so the following
5834 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
5837 Since @code{vec_add} is a macro, the vector constant in the example
5838 is treated as four separate arguments. Wrap the entire argument in
5839 parentheses for this to work.
5842 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
5843 Internally, GCC uses built-in functions to achieve the functionality in
5844 the aforementioned header file, but they are not supported and are
5845 subject to change without notice.
5847 The following interfaces are supported for the generic and specific
5848 AltiVec operations and the AltiVec predicates. In cases where there
5849 is a direct mapping between generic and specific operations, only the
5850 generic names are shown here, although the specific operations can also
5853 Arguments that are documented as @code{const int} require literal
5854 integral values within the range required for that operation.
5857 vector signed char vec_abs (vector signed char);
5858 vector signed short vec_abs (vector signed short);
5859 vector signed int vec_abs (vector signed int);
5860 vector float vec_abs (vector float);
5862 vector signed char vec_abss (vector signed char);
5863 vector signed short vec_abss (vector signed short);
5864 vector signed int vec_abss (vector signed int);
5866 vector signed char vec_add (vector bool char, vector signed char);
5867 vector signed char vec_add (vector signed char, vector bool char);
5868 vector signed char vec_add (vector signed char, vector signed char);
5869 vector unsigned char vec_add (vector bool char, vector unsigned char);
5870 vector unsigned char vec_add (vector unsigned char, vector bool char);
5871 vector unsigned char vec_add (vector unsigned char,
5872 vector unsigned char);
5873 vector signed short vec_add (vector bool short, vector signed short);
5874 vector signed short vec_add (vector signed short, vector bool short);
5875 vector signed short vec_add (vector signed short, vector signed short);
5876 vector unsigned short vec_add (vector bool short,
5877 vector unsigned short);
5878 vector unsigned short vec_add (vector unsigned short,
5880 vector unsigned short vec_add (vector unsigned short,
5881 vector unsigned short);
5882 vector signed int vec_add (vector bool int, vector signed int);
5883 vector signed int vec_add (vector signed int, vector bool int);
5884 vector signed int vec_add (vector signed int, vector signed int);
5885 vector unsigned int vec_add (vector bool int, vector unsigned int);
5886 vector unsigned int vec_add (vector unsigned int, vector bool int);
5887 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
5888 vector float vec_add (vector float, vector float);
5890 vector float vec_vaddfp (vector float, vector float);
5892 vector signed int vec_vadduwm (vector bool int, vector signed int);
5893 vector signed int vec_vadduwm (vector signed int, vector bool int);
5894 vector signed int vec_vadduwm (vector signed int, vector signed int);
5895 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
5896 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
5897 vector unsigned int vec_vadduwm (vector unsigned int,
5898 vector unsigned int);
5900 vector signed short vec_vadduhm (vector bool short,
5901 vector signed short);
5902 vector signed short vec_vadduhm (vector signed short,
5904 vector signed short vec_vadduhm (vector signed short,
5905 vector signed short);
5906 vector unsigned short vec_vadduhm (vector bool short,
5907 vector unsigned short);
5908 vector unsigned short vec_vadduhm (vector unsigned short,
5910 vector unsigned short vec_vadduhm (vector unsigned short,
5911 vector unsigned short);
5913 vector signed char vec_vaddubm (vector bool char, vector signed char);
5914 vector signed char vec_vaddubm (vector signed char, vector bool char);
5915 vector signed char vec_vaddubm (vector signed char, vector signed char);
5916 vector unsigned char vec_vaddubm (vector bool char,
5917 vector unsigned char);
5918 vector unsigned char vec_vaddubm (vector unsigned char,
5920 vector unsigned char vec_vaddubm (vector unsigned char,
5921 vector unsigned char);
5923 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
5925 vector unsigned char vec_adds (vector bool char, vector unsigned char);
5926 vector unsigned char vec_adds (vector unsigned char, vector bool char);
5927 vector unsigned char vec_adds (vector unsigned char,
5928 vector unsigned char);
5929 vector signed char vec_adds (vector bool char, vector signed char);
5930 vector signed char vec_adds (vector signed char, vector bool char);
5931 vector signed char vec_adds (vector signed char, vector signed char);
5932 vector unsigned short vec_adds (vector bool short,
5933 vector unsigned short);
5934 vector unsigned short vec_adds (vector unsigned short,
5936 vector unsigned short vec_adds (vector unsigned short,
5937 vector unsigned short);
5938 vector signed short vec_adds (vector bool short, vector signed short);
5939 vector signed short vec_adds (vector signed short, vector bool short);
5940 vector signed short vec_adds (vector signed short, vector signed short);
5941 vector unsigned int vec_adds (vector bool int, vector unsigned int);
5942 vector unsigned int vec_adds (vector unsigned int, vector bool int);
5943 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
5944 vector signed int vec_adds (vector bool int, vector signed int);
5945 vector signed int vec_adds (vector signed int, vector bool int);
5946 vector signed int vec_adds (vector signed int, vector signed int);
5948 vector signed int vec_vaddsws (vector bool int, vector signed int);
5949 vector signed int vec_vaddsws (vector signed int, vector bool int);
5950 vector signed int vec_vaddsws (vector signed int, vector signed int);
5952 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
5953 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
5954 vector unsigned int vec_vadduws (vector unsigned int,
5955 vector unsigned int);
5957 vector signed short vec_vaddshs (vector bool short,
5958 vector signed short);
5959 vector signed short vec_vaddshs (vector signed short,
5961 vector signed short vec_vaddshs (vector signed short,
5962 vector signed short);
5964 vector unsigned short vec_vadduhs (vector bool short,
5965 vector unsigned short);
5966 vector unsigned short vec_vadduhs (vector unsigned short,
5968 vector unsigned short vec_vadduhs (vector unsigned short,
5969 vector unsigned short);
5971 vector signed char vec_vaddsbs (vector bool char, vector signed char);
5972 vector signed char vec_vaddsbs (vector signed char, vector bool char);
5973 vector signed char vec_vaddsbs (vector signed char, vector signed char);
5975 vector unsigned char vec_vaddubs (vector bool char,
5976 vector unsigned char);
5977 vector unsigned char vec_vaddubs (vector unsigned char,
5979 vector unsigned char vec_vaddubs (vector unsigned char,
5980 vector unsigned char);
5982 vector float vec_and (vector float, vector float);
5983 vector float vec_and (vector float, vector bool int);
5984 vector float vec_and (vector bool int, vector float);
5985 vector bool int vec_and (vector bool int, vector bool int);
5986 vector signed int vec_and (vector bool int, vector signed int);
5987 vector signed int vec_and (vector signed int, vector bool int);
5988 vector signed int vec_and (vector signed int, vector signed int);
5989 vector unsigned int vec_and (vector bool int, vector unsigned int);
5990 vector unsigned int vec_and (vector unsigned int, vector bool int);
5991 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
5992 vector bool short vec_and (vector bool short, vector bool short);
5993 vector signed short vec_and (vector bool short, vector signed short);
5994 vector signed short vec_and (vector signed short, vector bool short);
5995 vector signed short vec_and (vector signed short, vector signed short);
5996 vector unsigned short vec_and (vector bool short,
5997 vector unsigned short);
5998 vector unsigned short vec_and (vector unsigned short,
6000 vector unsigned short vec_and (vector unsigned short,
6001 vector unsigned short);
6002 vector signed char vec_and (vector bool char, vector signed char);
6003 vector bool char vec_and (vector bool char, vector bool char);
6004 vector signed char vec_and (vector signed char, vector bool char);
6005 vector signed char vec_and (vector signed char, vector signed char);
6006 vector unsigned char vec_and (vector bool char, vector unsigned char);
6007 vector unsigned char vec_and (vector unsigned char, vector bool char);
6008 vector unsigned char vec_and (vector unsigned char,
6009 vector unsigned char);
6011 vector float vec_andc (vector float, vector float);
6012 vector float vec_andc (vector float, vector bool int);
6013 vector float vec_andc (vector bool int, vector float);
6014 vector bool int vec_andc (vector bool int, vector bool int);
6015 vector signed int vec_andc (vector bool int, vector signed int);
6016 vector signed int vec_andc (vector signed int, vector bool int);
6017 vector signed int vec_andc (vector signed int, vector signed int);
6018 vector unsigned int vec_andc (vector bool int, vector unsigned int);
6019 vector unsigned int vec_andc (vector unsigned int, vector bool int);
6020 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
6021 vector bool short vec_andc (vector bool short, vector bool short);
6022 vector signed short vec_andc (vector bool short, vector signed short);
6023 vector signed short vec_andc (vector signed short, vector bool short);
6024 vector signed short vec_andc (vector signed short, vector signed short);
6025 vector unsigned short vec_andc (vector bool short,
6026 vector unsigned short);
6027 vector unsigned short vec_andc (vector unsigned short,
6029 vector unsigned short vec_andc (vector unsigned short,
6030 vector unsigned short);
6031 vector signed char vec_andc (vector bool char, vector signed char);
6032 vector bool char vec_andc (vector bool char, vector bool char);
6033 vector signed char vec_andc (vector signed char, vector bool char);
6034 vector signed char vec_andc (vector signed char, vector signed char);
6035 vector unsigned char vec_andc (vector bool char, vector unsigned char);
6036 vector unsigned char vec_andc (vector unsigned char, vector bool char);
6037 vector unsigned char vec_andc (vector unsigned char,
6038 vector unsigned char);
6040 vector unsigned char vec_avg (vector unsigned char,
6041 vector unsigned char);
6042 vector signed char vec_avg (vector signed char, vector signed char);
6043 vector unsigned short vec_avg (vector unsigned short,
6044 vector unsigned short);
6045 vector signed short vec_avg (vector signed short, vector signed short);
6046 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
6047 vector signed int vec_avg (vector signed int, vector signed int);
6049 vector signed int vec_vavgsw (vector signed int, vector signed int);
6051 vector unsigned int vec_vavguw (vector unsigned int,
6052 vector unsigned int);
6054 vector signed short vec_vavgsh (vector signed short,
6055 vector signed short);
6057 vector unsigned short vec_vavguh (vector unsigned short,
6058 vector unsigned short);
6060 vector signed char vec_vavgsb (vector signed char, vector signed char);
6062 vector unsigned char vec_vavgub (vector unsigned char,
6063 vector unsigned char);
6065 vector float vec_ceil (vector float);
6067 vector signed int vec_cmpb (vector float, vector float);
6069 vector bool char vec_cmpeq (vector signed char, vector signed char);
6070 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
6071 vector bool short vec_cmpeq (vector signed short, vector signed short);
6072 vector bool short vec_cmpeq (vector unsigned short,
6073 vector unsigned short);
6074 vector bool int vec_cmpeq (vector signed int, vector signed int);
6075 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
6076 vector bool int vec_cmpeq (vector float, vector float);
6078 vector bool int vec_vcmpeqfp (vector float, vector float);
6080 vector bool int vec_vcmpequw (vector signed int, vector signed int);
6081 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
6083 vector bool short vec_vcmpequh (vector signed short,
6084 vector signed short);
6085 vector bool short vec_vcmpequh (vector unsigned short,
6086 vector unsigned short);
6088 vector bool char vec_vcmpequb (vector signed char, vector signed char);
6089 vector bool char vec_vcmpequb (vector unsigned char,
6090 vector unsigned char);
6092 vector bool int vec_cmpge (vector float, vector float);
6094 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
6095 vector bool char vec_cmpgt (vector signed char, vector signed char);
6096 vector bool short vec_cmpgt (vector unsigned short,
6097 vector unsigned short);
6098 vector bool short vec_cmpgt (vector signed short, vector signed short);
6099 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
6100 vector bool int vec_cmpgt (vector signed int, vector signed int);
6101 vector bool int vec_cmpgt (vector float, vector float);
6103 vector bool int vec_vcmpgtfp (vector float, vector float);
6105 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
6107 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
6109 vector bool short vec_vcmpgtsh (vector signed short,
6110 vector signed short);
6112 vector bool short vec_vcmpgtuh (vector unsigned short,
6113 vector unsigned short);
6115 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
6117 vector bool char vec_vcmpgtub (vector unsigned char,
6118 vector unsigned char);
6120 vector bool int vec_cmple (vector float, vector float);
6122 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
6123 vector bool char vec_cmplt (vector signed char, vector signed char);
6124 vector bool short vec_cmplt (vector unsigned short,
6125 vector unsigned short);
6126 vector bool short vec_cmplt (vector signed short, vector signed short);
6127 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
6128 vector bool int vec_cmplt (vector signed int, vector signed int);
6129 vector bool int vec_cmplt (vector float, vector float);
6131 vector float vec_ctf (vector unsigned int, const int);
6132 vector float vec_ctf (vector signed int, const int);
6134 vector float vec_vcfsx (vector signed int, const int);
6136 vector float vec_vcfux (vector unsigned int, const int);
6138 vector signed int vec_cts (vector float, const int);
6140 vector unsigned int vec_ctu (vector float, const int);
6142 void vec_dss (const int);
6144 void vec_dssall (void);
6146 void vec_dst (const vector unsigned char *, int, const int);
6147 void vec_dst (const vector signed char *, int, const int);
6148 void vec_dst (const vector bool char *, int, const int);
6149 void vec_dst (const vector unsigned short *, int, const int);
6150 void vec_dst (const vector signed short *, int, const int);
6151 void vec_dst (const vector bool short *, int, const int);
6152 void vec_dst (const vector pixel *, int, const int);
6153 void vec_dst (const vector unsigned int *, int, const int);
6154 void vec_dst (const vector signed int *, int, const int);
6155 void vec_dst (const vector bool int *, int, const int);
6156 void vec_dst (const vector float *, int, const int);
6157 void vec_dst (const unsigned char *, int, const int);
6158 void vec_dst (const signed char *, int, const int);
6159 void vec_dst (const unsigned short *, int, const int);
6160 void vec_dst (const short *, int, const int);
6161 void vec_dst (const unsigned int *, int, const int);
6162 void vec_dst (const int *, int, const int);
6163 void vec_dst (const unsigned long *, int, const int);
6164 void vec_dst (const long *, int, const int);
6165 void vec_dst (const float *, int, const int);
6167 void vec_dstst (const vector unsigned char *, int, const int);
6168 void vec_dstst (const vector signed char *, int, const int);
6169 void vec_dstst (const vector bool char *, int, const int);
6170 void vec_dstst (const vector unsigned short *, int, const int);
6171 void vec_dstst (const vector signed short *, int, const int);
6172 void vec_dstst (const vector bool short *, int, const int);
6173 void vec_dstst (const vector pixel *, int, const int);
6174 void vec_dstst (const vector unsigned int *, int, const int);
6175 void vec_dstst (const vector signed int *, int, const int);
6176 void vec_dstst (const vector bool int *, int, const int);
6177 void vec_dstst (const vector float *, int, const int);
6178 void vec_dstst (const unsigned char *, int, const int);
6179 void vec_dstst (const signed char *, int, const int);
6180 void vec_dstst (const unsigned short *, int, const int);
6181 void vec_dstst (const short *, int, const int);
6182 void vec_dstst (const unsigned int *, int, const int);
6183 void vec_dstst (const int *, int, const int);
6184 void vec_dstst (const unsigned long *, int, const int);
6185 void vec_dstst (const long *, int, const int);
6186 void vec_dstst (const float *, int, const int);
6188 void vec_dststt (const vector unsigned char *, int, const int);
6189 void vec_dststt (const vector signed char *, int, const int);
6190 void vec_dststt (const vector bool char *, int, const int);
6191 void vec_dststt (const vector unsigned short *, int, const int);
6192 void vec_dststt (const vector signed short *, int, const int);
6193 void vec_dststt (const vector bool short *, int, const int);
6194 void vec_dststt (const vector pixel *, int, const int);
6195 void vec_dststt (const vector unsigned int *, int, const int);
6196 void vec_dststt (const vector signed int *, int, const int);
6197 void vec_dststt (const vector bool int *, int, const int);
6198 void vec_dststt (const vector float *, int, const int);
6199 void vec_dststt (const unsigned char *, int, const int);
6200 void vec_dststt (const signed char *, int, const int);
6201 void vec_dststt (const unsigned short *, int, const int);
6202 void vec_dststt (const short *, int, const int);
6203 void vec_dststt (const unsigned int *, int, const int);
6204 void vec_dststt (const int *, int, const int);
6205 void vec_dststt (const unsigned long *, int, const int);
6206 void vec_dststt (const long *, int, const int);
6207 void vec_dststt (const float *, int, const int);
6209 void vec_dstt (const vector unsigned char *, int, const int);
6210 void vec_dstt (const vector signed char *, int, const int);
6211 void vec_dstt (const vector bool char *, int, const int);
6212 void vec_dstt (const vector unsigned short *, int, const int);
6213 void vec_dstt (const vector signed short *, int, const int);
6214 void vec_dstt (const vector bool short *, int, const int);
6215 void vec_dstt (const vector pixel *, int, const int);
6216 void vec_dstt (const vector unsigned int *, int, const int);
6217 void vec_dstt (const vector signed int *, int, const int);
6218 void vec_dstt (const vector bool int *, int, const int);
6219 void vec_dstt (const vector float *, int, const int);
6220 void vec_dstt (const unsigned char *, int, const int);
6221 void vec_dstt (const signed char *, int, const int);
6222 void vec_dstt (const unsigned short *, int, const int);
6223 void vec_dstt (const short *, int, const int);
6224 void vec_dstt (const unsigned int *, int, const int);
6225 void vec_dstt (const int *, int, const int);
6226 void vec_dstt (const unsigned long *, int, const int);
6227 void vec_dstt (const long *, int, const int);
6228 void vec_dstt (const float *, int, const int);
6230 vector float vec_expte (vector float);
6232 vector float vec_floor (vector float);
6234 vector float vec_ld (int, const vector float *);
6235 vector float vec_ld (int, const float *);
6236 vector bool int vec_ld (int, const vector bool int *);
6237 vector signed int vec_ld (int, const vector signed int *);
6238 vector signed int vec_ld (int, const int *);
6239 vector signed int vec_ld (int, const long *);
6240 vector unsigned int vec_ld (int, const vector unsigned int *);
6241 vector unsigned int vec_ld (int, const unsigned int *);
6242 vector unsigned int vec_ld (int, const unsigned long *);
6243 vector bool short vec_ld (int, const vector bool short *);
6244 vector pixel vec_ld (int, const vector pixel *);
6245 vector signed short vec_ld (int, const vector signed short *);
6246 vector signed short vec_ld (int, const short *);
6247 vector unsigned short vec_ld (int, const vector unsigned short *);
6248 vector unsigned short vec_ld (int, const unsigned short *);
6249 vector bool char vec_ld (int, const vector bool char *);
6250 vector signed char vec_ld (int, const vector signed char *);
6251 vector signed char vec_ld (int, const signed char *);
6252 vector unsigned char vec_ld (int, const vector unsigned char *);
6253 vector unsigned char vec_ld (int, const unsigned char *);
6255 vector signed char vec_lde (int, const signed char *);
6256 vector unsigned char vec_lde (int, const unsigned char *);
6257 vector signed short vec_lde (int, const short *);
6258 vector unsigned short vec_lde (int, const unsigned short *);
6259 vector float vec_lde (int, const float *);
6260 vector signed int vec_lde (int, const int *);
6261 vector unsigned int vec_lde (int, const unsigned int *);
6262 vector signed int vec_lde (int, const long *);
6263 vector unsigned int vec_lde (int, const unsigned long *);
6265 vector float vec_lvewx (int, float *);
6266 vector signed int vec_lvewx (int, int *);
6267 vector unsigned int vec_lvewx (int, unsigned int *);
6268 vector signed int vec_lvewx (int, long *);
6269 vector unsigned int vec_lvewx (int, unsigned long *);
6271 vector signed short vec_lvehx (int, short *);
6272 vector unsigned short vec_lvehx (int, unsigned short *);
6274 vector signed char vec_lvebx (int, char *);
6275 vector unsigned char vec_lvebx (int, unsigned char *);
6277 vector float vec_ldl (int, const vector float *);
6278 vector float vec_ldl (int, const float *);
6279 vector bool int vec_ldl (int, const vector bool int *);
6280 vector signed int vec_ldl (int, const vector signed int *);
6281 vector signed int vec_ldl (int, const int *);
6282 vector signed int vec_ldl (int, const long *);
6283 vector unsigned int vec_ldl (int, const vector unsigned int *);
6284 vector unsigned int vec_ldl (int, const unsigned int *);
6285 vector unsigned int vec_ldl (int, const unsigned long *);
6286 vector bool short vec_ldl (int, const vector bool short *);
6287 vector pixel vec_ldl (int, const vector pixel *);
6288 vector signed short vec_ldl (int, const vector signed short *);
6289 vector signed short vec_ldl (int, const short *);
6290 vector unsigned short vec_ldl (int, const vector unsigned short *);
6291 vector unsigned short vec_ldl (int, const unsigned short *);
6292 vector bool char vec_ldl (int, const vector bool char *);
6293 vector signed char vec_ldl (int, const vector signed char *);
6294 vector signed char vec_ldl (int, const signed char *);
6295 vector unsigned char vec_ldl (int, const vector unsigned char *);
6296 vector unsigned char vec_ldl (int, const unsigned char *);
6298 vector float vec_loge (vector float);
6300 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
6301 vector unsigned char vec_lvsl (int, const volatile signed char *);
6302 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
6303 vector unsigned char vec_lvsl (int, const volatile short *);
6304 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
6305 vector unsigned char vec_lvsl (int, const volatile int *);
6306 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
6307 vector unsigned char vec_lvsl (int, const volatile long *);
6308 vector unsigned char vec_lvsl (int, const volatile float *);
6310 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
6311 vector unsigned char vec_lvsr (int, const volatile signed char *);
6312 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
6313 vector unsigned char vec_lvsr (int, const volatile short *);
6314 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
6315 vector unsigned char vec_lvsr (int, const volatile int *);
6316 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
6317 vector unsigned char vec_lvsr (int, const volatile long *);
6318 vector unsigned char vec_lvsr (int, const volatile float *);
6320 vector float vec_madd (vector float, vector float, vector float);
6322 vector signed short vec_madds (vector signed short,
6323 vector signed short,
6324 vector signed short);
6326 vector unsigned char vec_max (vector bool char, vector unsigned char);
6327 vector unsigned char vec_max (vector unsigned char, vector bool char);
6328 vector unsigned char vec_max (vector unsigned char,
6329 vector unsigned char);
6330 vector signed char vec_max (vector bool char, vector signed char);
6331 vector signed char vec_max (vector signed char, vector bool char);
6332 vector signed char vec_max (vector signed char, vector signed char);
6333 vector unsigned short vec_max (vector bool short,
6334 vector unsigned short);
6335 vector unsigned short vec_max (vector unsigned short,
6337 vector unsigned short vec_max (vector unsigned short,
6338 vector unsigned short);
6339 vector signed short vec_max (vector bool short, vector signed short);
6340 vector signed short vec_max (vector signed short, vector bool short);
6341 vector signed short vec_max (vector signed short, vector signed short);
6342 vector unsigned int vec_max (vector bool int, vector unsigned int);
6343 vector unsigned int vec_max (vector unsigned int, vector bool int);
6344 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
6345 vector signed int vec_max (vector bool int, vector signed int);
6346 vector signed int vec_max (vector signed int, vector bool int);
6347 vector signed int vec_max (vector signed int, vector signed int);
6348 vector float vec_max (vector float, vector float);
6350 vector float vec_vmaxfp (vector float, vector float);
6352 vector signed int vec_vmaxsw (vector bool int, vector signed int);
6353 vector signed int vec_vmaxsw (vector signed int, vector bool int);
6354 vector signed int vec_vmaxsw (vector signed int, vector signed int);
6356 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
6357 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
6358 vector unsigned int vec_vmaxuw (vector unsigned int,
6359 vector unsigned int);
6361 vector signed short vec_vmaxsh (vector bool short, vector signed short);
6362 vector signed short vec_vmaxsh (vector signed short, vector bool short);
6363 vector signed short vec_vmaxsh (vector signed short,
6364 vector signed short);
6366 vector unsigned short vec_vmaxuh (vector bool short,
6367 vector unsigned short);
6368 vector unsigned short vec_vmaxuh (vector unsigned short,
6370 vector unsigned short vec_vmaxuh (vector unsigned short,
6371 vector unsigned short);
6373 vector signed char vec_vmaxsb (vector bool char, vector signed char);
6374 vector signed char vec_vmaxsb (vector signed char, vector bool char);
6375 vector signed char vec_vmaxsb (vector signed char, vector signed char);
6377 vector unsigned char vec_vmaxub (vector bool char,
6378 vector unsigned char);
6379 vector unsigned char vec_vmaxub (vector unsigned char,
6381 vector unsigned char vec_vmaxub (vector unsigned char,
6382 vector unsigned char);
6384 vector bool char vec_mergeh (vector bool char, vector bool char);
6385 vector signed char vec_mergeh (vector signed char, vector signed char);
6386 vector unsigned char vec_mergeh (vector unsigned char,
6387 vector unsigned char);
6388 vector bool short vec_mergeh (vector bool short, vector bool short);
6389 vector pixel vec_mergeh (vector pixel, vector pixel);
6390 vector signed short vec_mergeh (vector signed short,
6391 vector signed short);
6392 vector unsigned short vec_mergeh (vector unsigned short,
6393 vector unsigned short);
6394 vector float vec_mergeh (vector float, vector float);
6395 vector bool int vec_mergeh (vector bool int, vector bool int);
6396 vector signed int vec_mergeh (vector signed int, vector signed int);
6397 vector unsigned int vec_mergeh (vector unsigned int,
6398 vector unsigned int);
6400 vector float vec_vmrghw (vector float, vector float);
6401 vector bool int vec_vmrghw (vector bool int, vector bool int);
6402 vector signed int vec_vmrghw (vector signed int, vector signed int);
6403 vector unsigned int vec_vmrghw (vector unsigned int,
6404 vector unsigned int);
6406 vector bool short vec_vmrghh (vector bool short, vector bool short);
6407 vector signed short vec_vmrghh (vector signed short,
6408 vector signed short);
6409 vector unsigned short vec_vmrghh (vector unsigned short,
6410 vector unsigned short);
6411 vector pixel vec_vmrghh (vector pixel, vector pixel);
6413 vector bool char vec_vmrghb (vector bool char, vector bool char);
6414 vector signed char vec_vmrghb (vector signed char, vector signed char);
6415 vector unsigned char vec_vmrghb (vector unsigned char,
6416 vector unsigned char);
6418 vector bool char vec_mergel (vector bool char, vector bool char);
6419 vector signed char vec_mergel (vector signed char, vector signed char);
6420 vector unsigned char vec_mergel (vector unsigned char,
6421 vector unsigned char);
6422 vector bool short vec_mergel (vector bool short, vector bool short);
6423 vector pixel vec_mergel (vector pixel, vector pixel);
6424 vector signed short vec_mergel (vector signed short,
6425 vector signed short);
6426 vector unsigned short vec_mergel (vector unsigned short,
6427 vector unsigned short);
6428 vector float vec_mergel (vector float, vector float);
6429 vector bool int vec_mergel (vector bool int, vector bool int);
6430 vector signed int vec_mergel (vector signed int, vector signed int);
6431 vector unsigned int vec_mergel (vector unsigned int,
6432 vector unsigned int);
6434 vector float vec_vmrglw (vector float, vector float);
6435 vector signed int vec_vmrglw (vector signed int, vector signed int);
6436 vector unsigned int vec_vmrglw (vector unsigned int,
6437 vector unsigned int);
6438 vector bool int vec_vmrglw (vector bool int, vector bool int);
6440 vector bool short vec_vmrglh (vector bool short, vector bool short);
6441 vector signed short vec_vmrglh (vector signed short,
6442 vector signed short);
6443 vector unsigned short vec_vmrglh (vector unsigned short,
6444 vector unsigned short);
6445 vector pixel vec_vmrglh (vector pixel, vector pixel);
6447 vector bool char vec_vmrglb (vector bool char, vector bool char);
6448 vector signed char vec_vmrglb (vector signed char, vector signed char);
6449 vector unsigned char vec_vmrglb (vector unsigned char,
6450 vector unsigned char);
6452 vector unsigned short vec_mfvscr (void);
6454 vector unsigned char vec_min (vector bool char, vector unsigned char);
6455 vector unsigned char vec_min (vector unsigned char, vector bool char);
6456 vector unsigned char vec_min (vector unsigned char,
6457 vector unsigned char);
6458 vector signed char vec_min (vector bool char, vector signed char);
6459 vector signed char vec_min (vector signed char, vector bool char);
6460 vector signed char vec_min (vector signed char, vector signed char);
6461 vector unsigned short vec_min (vector bool short,
6462 vector unsigned short);
6463 vector unsigned short vec_min (vector unsigned short,
6465 vector unsigned short vec_min (vector unsigned short,
6466 vector unsigned short);
6467 vector signed short vec_min (vector bool short, vector signed short);
6468 vector signed short vec_min (vector signed short, vector bool short);
6469 vector signed short vec_min (vector signed short, vector signed short);
6470 vector unsigned int vec_min (vector bool int, vector unsigned int);
6471 vector unsigned int vec_min (vector unsigned int, vector bool int);
6472 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
6473 vector signed int vec_min (vector bool int, vector signed int);
6474 vector signed int vec_min (vector signed int, vector bool int);
6475 vector signed int vec_min (vector signed int, vector signed int);
6476 vector float vec_min (vector float, vector float);
6478 vector float vec_vminfp (vector float, vector float);
6480 vector signed int vec_vminsw (vector bool int, vector signed int);
6481 vector signed int vec_vminsw (vector signed int, vector bool int);
6482 vector signed int vec_vminsw (vector signed int, vector signed int);
6484 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
6485 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
6486 vector unsigned int vec_vminuw (vector unsigned int,
6487 vector unsigned int);
6489 vector signed short vec_vminsh (vector bool short, vector signed short);
6490 vector signed short vec_vminsh (vector signed short, vector bool short);
6491 vector signed short vec_vminsh (vector signed short,
6492 vector signed short);
6494 vector unsigned short vec_vminuh (vector bool short,
6495 vector unsigned short);
6496 vector unsigned short vec_vminuh (vector unsigned short,
6498 vector unsigned short vec_vminuh (vector unsigned short,
6499 vector unsigned short);
6501 vector signed char vec_vminsb (vector bool char, vector signed char);
6502 vector signed char vec_vminsb (vector signed char, vector bool char);
6503 vector signed char vec_vminsb (vector signed char, vector signed char);
6505 vector unsigned char vec_vminub (vector bool char,
6506 vector unsigned char);
6507 vector unsigned char vec_vminub (vector unsigned char,
6509 vector unsigned char vec_vminub (vector unsigned char,
6510 vector unsigned char);
6512 vector signed short vec_mladd (vector signed short,
6513 vector signed short,
6514 vector signed short);
6515 vector signed short vec_mladd (vector signed short,
6516 vector unsigned short,
6517 vector unsigned short);
6518 vector signed short vec_mladd (vector unsigned short,
6519 vector signed short,
6520 vector signed short);
6521 vector unsigned short vec_mladd (vector unsigned short,
6522 vector unsigned short,
6523 vector unsigned short);
6525 vector signed short vec_mradds (vector signed short,
6526 vector signed short,
6527 vector signed short);
6529 vector unsigned int vec_msum (vector unsigned char,
6530 vector unsigned char,
6531 vector unsigned int);
6532 vector signed int vec_msum (vector signed char,
6533 vector unsigned char,
6535 vector unsigned int vec_msum (vector unsigned short,
6536 vector unsigned short,
6537 vector unsigned int);
6538 vector signed int vec_msum (vector signed short,
6539 vector signed short,
6542 vector signed int vec_vmsumshm (vector signed short,
6543 vector signed short,
6546 vector unsigned int vec_vmsumuhm (vector unsigned short,
6547 vector unsigned short,
6548 vector unsigned int);
6550 vector signed int vec_vmsummbm (vector signed char,
6551 vector unsigned char,
6554 vector unsigned int vec_vmsumubm (vector unsigned char,
6555 vector unsigned char,
6556 vector unsigned int);
6558 vector unsigned int vec_msums (vector unsigned short,
6559 vector unsigned short,
6560 vector unsigned int);
6561 vector signed int vec_msums (vector signed short,
6562 vector signed short,
6565 vector signed int vec_vmsumshs (vector signed short,
6566 vector signed short,
6569 vector unsigned int vec_vmsumuhs (vector unsigned short,
6570 vector unsigned short,
6571 vector unsigned int);
6573 void vec_mtvscr (vector signed int);
6574 void vec_mtvscr (vector unsigned int);
6575 void vec_mtvscr (vector bool int);
6576 void vec_mtvscr (vector signed short);
6577 void vec_mtvscr (vector unsigned short);
6578 void vec_mtvscr (vector bool short);
6579 void vec_mtvscr (vector pixel);
6580 void vec_mtvscr (vector signed char);
6581 void vec_mtvscr (vector unsigned char);
6582 void vec_mtvscr (vector bool char);
6584 vector unsigned short vec_mule (vector unsigned char,
6585 vector unsigned char);
6586 vector signed short vec_mule (vector signed char,
6587 vector signed char);
6588 vector unsigned int vec_mule (vector unsigned short,
6589 vector unsigned short);
6590 vector signed int vec_mule (vector signed short, vector signed short);
6592 vector signed int vec_vmulesh (vector signed short,
6593 vector signed short);
6595 vector unsigned int vec_vmuleuh (vector unsigned short,
6596 vector unsigned short);
6598 vector signed short vec_vmulesb (vector signed char,
6599 vector signed char);
6601 vector unsigned short vec_vmuleub (vector unsigned char,
6602 vector unsigned char);
6604 vector unsigned short vec_mulo (vector unsigned char,
6605 vector unsigned char);
6606 vector signed short vec_mulo (vector signed char, vector signed char);
6607 vector unsigned int vec_mulo (vector unsigned short,
6608 vector unsigned short);
6609 vector signed int vec_mulo (vector signed short, vector signed short);
6611 vector signed int vec_vmulosh (vector signed short,
6612 vector signed short);
6614 vector unsigned int vec_vmulouh (vector unsigned short,
6615 vector unsigned short);
6617 vector signed short vec_vmulosb (vector signed char,
6618 vector signed char);
6620 vector unsigned short vec_vmuloub (vector unsigned char,
6621 vector unsigned char);
6623 vector float vec_nmsub (vector float, vector float, vector float);
6625 vector float vec_nor (vector float, vector float);
6626 vector signed int vec_nor (vector signed int, vector signed int);
6627 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
6628 vector bool int vec_nor (vector bool int, vector bool int);
6629 vector signed short vec_nor (vector signed short, vector signed short);
6630 vector unsigned short vec_nor (vector unsigned short,
6631 vector unsigned short);
6632 vector bool short vec_nor (vector bool short, vector bool short);
6633 vector signed char vec_nor (vector signed char, vector signed char);
6634 vector unsigned char vec_nor (vector unsigned char,
6635 vector unsigned char);
6636 vector bool char vec_nor (vector bool char, vector bool char);
6638 vector float vec_or (vector float, vector float);
6639 vector float vec_or (vector float, vector bool int);
6640 vector float vec_or (vector bool int, vector float);
6641 vector bool int vec_or (vector bool int, vector bool int);
6642 vector signed int vec_or (vector bool int, vector signed int);
6643 vector signed int vec_or (vector signed int, vector bool int);
6644 vector signed int vec_or (vector signed int, vector signed int);
6645 vector unsigned int vec_or (vector bool int, vector unsigned int);
6646 vector unsigned int vec_or (vector unsigned int, vector bool int);
6647 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
6648 vector bool short vec_or (vector bool short, vector bool short);
6649 vector signed short vec_or (vector bool short, vector signed short);
6650 vector signed short vec_or (vector signed short, vector bool short);
6651 vector signed short vec_or (vector signed short, vector signed short);
6652 vector unsigned short vec_or (vector bool short, vector unsigned short);
6653 vector unsigned short vec_or (vector unsigned short, vector bool short);
6654 vector unsigned short vec_or (vector unsigned short,
6655 vector unsigned short);
6656 vector signed char vec_or (vector bool char, vector signed char);
6657 vector bool char vec_or (vector bool char, vector bool char);
6658 vector signed char vec_or (vector signed char, vector bool char);
6659 vector signed char vec_or (vector signed char, vector signed char);
6660 vector unsigned char vec_or (vector bool char, vector unsigned char);
6661 vector unsigned char vec_or (vector unsigned char, vector bool char);
6662 vector unsigned char vec_or (vector unsigned char,
6663 vector unsigned char);
6665 vector signed char vec_pack (vector signed short, vector signed short);
6666 vector unsigned char vec_pack (vector unsigned short,
6667 vector unsigned short);
6668 vector bool char vec_pack (vector bool short, vector bool short);
6669 vector signed short vec_pack (vector signed int, vector signed int);
6670 vector unsigned short vec_pack (vector unsigned int,
6671 vector unsigned int);
6672 vector bool short vec_pack (vector bool int, vector bool int);
6674 vector bool short vec_vpkuwum (vector bool int, vector bool int);
6675 vector signed short vec_vpkuwum (vector signed int, vector signed int);
6676 vector unsigned short vec_vpkuwum (vector unsigned int,
6677 vector unsigned int);
6679 vector bool char vec_vpkuhum (vector bool short, vector bool short);
6680 vector signed char vec_vpkuhum (vector signed short,
6681 vector signed short);
6682 vector unsigned char vec_vpkuhum (vector unsigned short,
6683 vector unsigned short);
6685 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
6687 vector unsigned char vec_packs (vector unsigned short,
6688 vector unsigned short);
6689 vector signed char vec_packs (vector signed short, vector signed short);
6690 vector unsigned short vec_packs (vector unsigned int,
6691 vector unsigned int);
6692 vector signed short vec_packs (vector signed int, vector signed int);
6694 vector signed short vec_vpkswss (vector signed int, vector signed int);
6696 vector unsigned short vec_vpkuwus (vector unsigned int,
6697 vector unsigned int);
6699 vector signed char vec_vpkshss (vector signed short,
6700 vector signed short);
6702 vector unsigned char vec_vpkuhus (vector unsigned short,
6703 vector unsigned short);
6705 vector unsigned char vec_packsu (vector unsigned short,
6706 vector unsigned short);
6707 vector unsigned char vec_packsu (vector signed short,
6708 vector signed short);
6709 vector unsigned short vec_packsu (vector unsigned int,
6710 vector unsigned int);
6711 vector unsigned short vec_packsu (vector signed int, vector signed int);
6713 vector unsigned short vec_vpkswus (vector signed int,
6716 vector unsigned char vec_vpkshus (vector signed short,
6717 vector signed short);
6719 vector float vec_perm (vector float,
6721 vector unsigned char);
6722 vector signed int vec_perm (vector signed int,
6724 vector unsigned char);
6725 vector unsigned int vec_perm (vector unsigned int,
6726 vector unsigned int,
6727 vector unsigned char);
6728 vector bool int vec_perm (vector bool int,
6730 vector unsigned char);
6731 vector signed short vec_perm (vector signed short,
6732 vector signed short,
6733 vector unsigned char);
6734 vector unsigned short vec_perm (vector unsigned short,
6735 vector unsigned short,
6736 vector unsigned char);
6737 vector bool short vec_perm (vector bool short,
6739 vector unsigned char);
6740 vector pixel vec_perm (vector pixel,
6742 vector unsigned char);
6743 vector signed char vec_perm (vector signed char,
6745 vector unsigned char);
6746 vector unsigned char vec_perm (vector unsigned char,
6747 vector unsigned char,
6748 vector unsigned char);
6749 vector bool char vec_perm (vector bool char,
6751 vector unsigned char);
6753 vector float vec_re (vector float);
6755 vector signed char vec_rl (vector signed char,
6756 vector unsigned char);
6757 vector unsigned char vec_rl (vector unsigned char,
6758 vector unsigned char);
6759 vector signed short vec_rl (vector signed short, vector unsigned short);
6760 vector unsigned short vec_rl (vector unsigned short,
6761 vector unsigned short);
6762 vector signed int vec_rl (vector signed int, vector unsigned int);
6763 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
6765 vector signed int vec_vrlw (vector signed int, vector unsigned int);
6766 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
6768 vector signed short vec_vrlh (vector signed short,
6769 vector unsigned short);
6770 vector unsigned short vec_vrlh (vector unsigned short,
6771 vector unsigned short);
6773 vector signed char vec_vrlb (vector signed char, vector unsigned char);
6774 vector unsigned char vec_vrlb (vector unsigned char,
6775 vector unsigned char);
6777 vector float vec_round (vector float);
6779 vector float vec_rsqrte (vector float);
6781 vector float vec_sel (vector float, vector float, vector bool int);
6782 vector float vec_sel (vector float, vector float, vector unsigned int);
6783 vector signed int vec_sel (vector signed int,
6786 vector signed int vec_sel (vector signed int,
6788 vector unsigned int);
6789 vector unsigned int vec_sel (vector unsigned int,
6790 vector unsigned int,
6792 vector unsigned int vec_sel (vector unsigned int,
6793 vector unsigned int,
6794 vector unsigned int);
6795 vector bool int vec_sel (vector bool int,
6798 vector bool int vec_sel (vector bool int,
6800 vector unsigned int);
6801 vector signed short vec_sel (vector signed short,
6802 vector signed short,
6804 vector signed short vec_sel (vector signed short,
6805 vector signed short,
6806 vector unsigned short);
6807 vector unsigned short vec_sel (vector unsigned short,
6808 vector unsigned short,
6810 vector unsigned short vec_sel (vector unsigned short,
6811 vector unsigned short,
6812 vector unsigned short);
6813 vector bool short vec_sel (vector bool short,
6816 vector bool short vec_sel (vector bool short,
6818 vector unsigned short);
6819 vector signed char vec_sel (vector signed char,
6822 vector signed char vec_sel (vector signed char,
6824 vector unsigned char);
6825 vector unsigned char vec_sel (vector unsigned char,
6826 vector unsigned char,
6828 vector unsigned char vec_sel (vector unsigned char,
6829 vector unsigned char,
6830 vector unsigned char);
6831 vector bool char vec_sel (vector bool char,
6834 vector bool char vec_sel (vector bool char,
6836 vector unsigned char);
6838 vector signed char vec_sl (vector signed char,
6839 vector unsigned char);
6840 vector unsigned char vec_sl (vector unsigned char,
6841 vector unsigned char);
6842 vector signed short vec_sl (vector signed short, vector unsigned short);
6843 vector unsigned short vec_sl (vector unsigned short,
6844 vector unsigned short);
6845 vector signed int vec_sl (vector signed int, vector unsigned int);
6846 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
6848 vector signed int vec_vslw (vector signed int, vector unsigned int);
6849 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
6851 vector signed short vec_vslh (vector signed short,
6852 vector unsigned short);
6853 vector unsigned short vec_vslh (vector unsigned short,
6854 vector unsigned short);
6856 vector signed char vec_vslb (vector signed char, vector unsigned char);
6857 vector unsigned char vec_vslb (vector unsigned char,
6858 vector unsigned char);
6860 vector float vec_sld (vector float, vector float, const int);
6861 vector signed int vec_sld (vector signed int,
6864 vector unsigned int vec_sld (vector unsigned int,
6865 vector unsigned int,
6867 vector bool int vec_sld (vector bool int,
6870 vector signed short vec_sld (vector signed short,
6871 vector signed short,
6873 vector unsigned short vec_sld (vector unsigned short,
6874 vector unsigned short,
6876 vector bool short vec_sld (vector bool short,
6879 vector pixel vec_sld (vector pixel,
6882 vector signed char vec_sld (vector signed char,
6885 vector unsigned char vec_sld (vector unsigned char,
6886 vector unsigned char,
6888 vector bool char vec_sld (vector bool char,
6892 vector signed int vec_sll (vector signed int,
6893 vector unsigned int);
6894 vector signed int vec_sll (vector signed int,
6895 vector unsigned short);
6896 vector signed int vec_sll (vector signed int,
6897 vector unsigned char);
6898 vector unsigned int vec_sll (vector unsigned int,
6899 vector unsigned int);
6900 vector unsigned int vec_sll (vector unsigned int,
6901 vector unsigned short);
6902 vector unsigned int vec_sll (vector unsigned int,
6903 vector unsigned char);
6904 vector bool int vec_sll (vector bool int,
6905 vector unsigned int);
6906 vector bool int vec_sll (vector bool int,
6907 vector unsigned short);
6908 vector bool int vec_sll (vector bool int,
6909 vector unsigned char);
6910 vector signed short vec_sll (vector signed short,
6911 vector unsigned int);
6912 vector signed short vec_sll (vector signed short,
6913 vector unsigned short);
6914 vector signed short vec_sll (vector signed short,
6915 vector unsigned char);
6916 vector unsigned short vec_sll (vector unsigned short,
6917 vector unsigned int);
6918 vector unsigned short vec_sll (vector unsigned short,
6919 vector unsigned short);
6920 vector unsigned short vec_sll (vector unsigned short,
6921 vector unsigned char);
6922 vector bool short vec_sll (vector bool short, vector unsigned int);
6923 vector bool short vec_sll (vector bool short, vector unsigned short);
6924 vector bool short vec_sll (vector bool short, vector unsigned char);
6925 vector pixel vec_sll (vector pixel, vector unsigned int);
6926 vector pixel vec_sll (vector pixel, vector unsigned short);
6927 vector pixel vec_sll (vector pixel, vector unsigned char);
6928 vector signed char vec_sll (vector signed char, vector unsigned int);
6929 vector signed char vec_sll (vector signed char, vector unsigned short);
6930 vector signed char vec_sll (vector signed char, vector unsigned char);
6931 vector unsigned char vec_sll (vector unsigned char,
6932 vector unsigned int);
6933 vector unsigned char vec_sll (vector unsigned char,
6934 vector unsigned short);
6935 vector unsigned char vec_sll (vector unsigned char,
6936 vector unsigned char);
6937 vector bool char vec_sll (vector bool char, vector unsigned int);
6938 vector bool char vec_sll (vector bool char, vector unsigned short);
6939 vector bool char vec_sll (vector bool char, vector unsigned char);
6941 vector float vec_slo (vector float, vector signed char);
6942 vector float vec_slo (vector float, vector unsigned char);
6943 vector signed int vec_slo (vector signed int, vector signed char);
6944 vector signed int vec_slo (vector signed int, vector unsigned char);
6945 vector unsigned int vec_slo (vector unsigned int, vector signed char);
6946 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
6947 vector signed short vec_slo (vector signed short, vector signed char);
6948 vector signed short vec_slo (vector signed short, vector unsigned char);
6949 vector unsigned short vec_slo (vector unsigned short,
6950 vector signed char);
6951 vector unsigned short vec_slo (vector unsigned short,
6952 vector unsigned char);
6953 vector pixel vec_slo (vector pixel, vector signed char);
6954 vector pixel vec_slo (vector pixel, vector unsigned char);
6955 vector signed char vec_slo (vector signed char, vector signed char);
6956 vector signed char vec_slo (vector signed char, vector unsigned char);
6957 vector unsigned char vec_slo (vector unsigned char, vector signed char);
6958 vector unsigned char vec_slo (vector unsigned char,
6959 vector unsigned char);
6961 vector signed char vec_splat (vector signed char, const int);
6962 vector unsigned char vec_splat (vector unsigned char, const int);
6963 vector bool char vec_splat (vector bool char, const int);
6964 vector signed short vec_splat (vector signed short, const int);
6965 vector unsigned short vec_splat (vector unsigned short, const int);
6966 vector bool short vec_splat (vector bool short, const int);
6967 vector pixel vec_splat (vector pixel, const int);
6968 vector float vec_splat (vector float, const int);
6969 vector signed int vec_splat (vector signed int, const int);
6970 vector unsigned int vec_splat (vector unsigned int, const int);
6971 vector bool int vec_splat (vector bool int, const int);
6973 vector float vec_vspltw (vector float, const int);
6974 vector signed int vec_vspltw (vector signed int, const int);
6975 vector unsigned int vec_vspltw (vector unsigned int, const int);
6976 vector bool int vec_vspltw (vector bool int, const int);
6978 vector bool short vec_vsplth (vector bool short, const int);
6979 vector signed short vec_vsplth (vector signed short, const int);
6980 vector unsigned short vec_vsplth (vector unsigned short, const int);
6981 vector pixel vec_vsplth (vector pixel, const int);
6983 vector signed char vec_vspltb (vector signed char, const int);
6984 vector unsigned char vec_vspltb (vector unsigned char, const int);
6985 vector bool char vec_vspltb (vector bool char, const int);
6987 vector signed char vec_splat_s8 (const int);
6989 vector signed short vec_splat_s16 (const int);
6991 vector signed int vec_splat_s32 (const int);
6993 vector unsigned char vec_splat_u8 (const int);
6995 vector unsigned short vec_splat_u16 (const int);
6997 vector unsigned int vec_splat_u32 (const int);
6999 vector signed char vec_sr (vector signed char, vector unsigned char);
7000 vector unsigned char vec_sr (vector unsigned char,
7001 vector unsigned char);
7002 vector signed short vec_sr (vector signed short,
7003 vector unsigned short);
7004 vector unsigned short vec_sr (vector unsigned short,
7005 vector unsigned short);
7006 vector signed int vec_sr (vector signed int, vector unsigned int);
7007 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
7009 vector signed int vec_vsrw (vector signed int, vector unsigned int);
7010 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
7012 vector signed short vec_vsrh (vector signed short,
7013 vector unsigned short);
7014 vector unsigned short vec_vsrh (vector unsigned short,
7015 vector unsigned short);
7017 vector signed char vec_vsrb (vector signed char, vector unsigned char);
7018 vector unsigned char vec_vsrb (vector unsigned char,
7019 vector unsigned char);
7021 vector signed char vec_sra (vector signed char, vector unsigned char);
7022 vector unsigned char vec_sra (vector unsigned char,
7023 vector unsigned char);
7024 vector signed short vec_sra (vector signed short,
7025 vector unsigned short);
7026 vector unsigned short vec_sra (vector unsigned short,
7027 vector unsigned short);
7028 vector signed int vec_sra (vector signed int, vector unsigned int);
7029 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
7031 vector signed int vec_vsraw (vector signed int, vector unsigned int);
7032 vector unsigned int vec_vsraw (vector unsigned int,
7033 vector unsigned int);
7035 vector signed short vec_vsrah (vector signed short,
7036 vector unsigned short);
7037 vector unsigned short vec_vsrah (vector unsigned short,
7038 vector unsigned short);
7040 vector signed char vec_vsrab (vector signed char, vector unsigned char);
7041 vector unsigned char vec_vsrab (vector unsigned char,
7042 vector unsigned char);
7044 vector signed int vec_srl (vector signed int, vector unsigned int);
7045 vector signed int vec_srl (vector signed int, vector unsigned short);
7046 vector signed int vec_srl (vector signed int, vector unsigned char);
7047 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
7048 vector unsigned int vec_srl (vector unsigned int,
7049 vector unsigned short);
7050 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
7051 vector bool int vec_srl (vector bool int, vector unsigned int);
7052 vector bool int vec_srl (vector bool int, vector unsigned short);
7053 vector bool int vec_srl (vector bool int, vector unsigned char);
7054 vector signed short vec_srl (vector signed short, vector unsigned int);
7055 vector signed short vec_srl (vector signed short,
7056 vector unsigned short);
7057 vector signed short vec_srl (vector signed short, vector unsigned char);
7058 vector unsigned short vec_srl (vector unsigned short,
7059 vector unsigned int);
7060 vector unsigned short vec_srl (vector unsigned short,
7061 vector unsigned short);
7062 vector unsigned short vec_srl (vector unsigned short,
7063 vector unsigned char);
7064 vector bool short vec_srl (vector bool short, vector unsigned int);
7065 vector bool short vec_srl (vector bool short, vector unsigned short);
7066 vector bool short vec_srl (vector bool short, vector unsigned char);
7067 vector pixel vec_srl (vector pixel, vector unsigned int);
7068 vector pixel vec_srl (vector pixel, vector unsigned short);
7069 vector pixel vec_srl (vector pixel, vector unsigned char);
7070 vector signed char vec_srl (vector signed char, vector unsigned int);
7071 vector signed char vec_srl (vector signed char, vector unsigned short);
7072 vector signed char vec_srl (vector signed char, vector unsigned char);
7073 vector unsigned char vec_srl (vector unsigned char,
7074 vector unsigned int);
7075 vector unsigned char vec_srl (vector unsigned char,
7076 vector unsigned short);
7077 vector unsigned char vec_srl (vector unsigned char,
7078 vector unsigned char);
7079 vector bool char vec_srl (vector bool char, vector unsigned int);
7080 vector bool char vec_srl (vector bool char, vector unsigned short);
7081 vector bool char vec_srl (vector bool char, vector unsigned char);
7083 vector float vec_sro (vector float, vector signed char);
7084 vector float vec_sro (vector float, vector unsigned char);
7085 vector signed int vec_sro (vector signed int, vector signed char);
7086 vector signed int vec_sro (vector signed int, vector unsigned char);
7087 vector unsigned int vec_sro (vector unsigned int, vector signed char);
7088 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
7089 vector signed short vec_sro (vector signed short, vector signed char);
7090 vector signed short vec_sro (vector signed short, vector unsigned char);
7091 vector unsigned short vec_sro (vector unsigned short,
7092 vector signed char);
7093 vector unsigned short vec_sro (vector unsigned short,
7094 vector unsigned char);
7095 vector pixel vec_sro (vector pixel, vector signed char);
7096 vector pixel vec_sro (vector pixel, vector unsigned char);
7097 vector signed char vec_sro (vector signed char, vector signed char);
7098 vector signed char vec_sro (vector signed char, vector unsigned char);
7099 vector unsigned char vec_sro (vector unsigned char, vector signed char);
7100 vector unsigned char vec_sro (vector unsigned char,
7101 vector unsigned char);
7103 void vec_st (vector float, int, vector float *);
7104 void vec_st (vector float, int, float *);
7105 void vec_st (vector signed int, int, vector signed int *);
7106 void vec_st (vector signed int, int, int *);
7107 void vec_st (vector unsigned int, int, vector unsigned int *);
7108 void vec_st (vector unsigned int, int, unsigned int *);
7109 void vec_st (vector bool int, int, vector bool int *);
7110 void vec_st (vector bool int, int, unsigned int *);
7111 void vec_st (vector bool int, int, int *);
7112 void vec_st (vector signed short, int, vector signed short *);
7113 void vec_st (vector signed short, int, short *);
7114 void vec_st (vector unsigned short, int, vector unsigned short *);
7115 void vec_st (vector unsigned short, int, unsigned short *);
7116 void vec_st (vector bool short, int, vector bool short *);
7117 void vec_st (vector bool short, int, unsigned short *);
7118 void vec_st (vector pixel, int, vector pixel *);
7119 void vec_st (vector pixel, int, unsigned short *);
7120 void vec_st (vector pixel, int, short *);
7121 void vec_st (vector bool short, int, short *);
7122 void vec_st (vector signed char, int, vector signed char *);
7123 void vec_st (vector signed char, int, signed char *);
7124 void vec_st (vector unsigned char, int, vector unsigned char *);
7125 void vec_st (vector unsigned char, int, unsigned char *);
7126 void vec_st (vector bool char, int, vector bool char *);
7127 void vec_st (vector bool char, int, unsigned char *);
7128 void vec_st (vector bool char, int, signed char *);
7130 void vec_ste (vector signed char, int, signed char *);
7131 void vec_ste (vector unsigned char, int, unsigned char *);
7132 void vec_ste (vector bool char, int, signed char *);
7133 void vec_ste (vector bool char, int, unsigned char *);
7134 void vec_ste (vector signed short, int, short *);
7135 void vec_ste (vector unsigned short, int, unsigned short *);
7136 void vec_ste (vector bool short, int, short *);
7137 void vec_ste (vector bool short, int, unsigned short *);
7138 void vec_ste (vector pixel, int, short *);
7139 void vec_ste (vector pixel, int, unsigned short *);
7140 void vec_ste (vector float, int, float *);
7141 void vec_ste (vector signed int, int, int *);
7142 void vec_ste (vector unsigned int, int, unsigned int *);
7143 void vec_ste (vector bool int, int, int *);
7144 void vec_ste (vector bool int, int, unsigned int *);
7146 void vec_stvewx (vector float, int, float *);
7147 void vec_stvewx (vector signed int, int, int *);
7148 void vec_stvewx (vector unsigned int, int, unsigned int *);
7149 void vec_stvewx (vector bool int, int, int *);
7150 void vec_stvewx (vector bool int, int, unsigned int *);
7152 void vec_stvehx (vector signed short, int, short *);
7153 void vec_stvehx (vector unsigned short, int, unsigned short *);
7154 void vec_stvehx (vector bool short, int, short *);
7155 void vec_stvehx (vector bool short, int, unsigned short *);
7156 void vec_stvehx (vector pixel, int, short *);
7157 void vec_stvehx (vector pixel, int, unsigned short *);
7159 void vec_stvebx (vector signed char, int, signed char *);
7160 void vec_stvebx (vector unsigned char, int, unsigned char *);
7161 void vec_stvebx (vector bool char, int, signed char *);
7162 void vec_stvebx (vector bool char, int, unsigned char *);
7164 void vec_stl (vector float, int, vector float *);
7165 void vec_stl (vector float, int, float *);
7166 void vec_stl (vector signed int, int, vector signed int *);
7167 void vec_stl (vector signed int, int, int *);
7168 void vec_stl (vector unsigned int, int, vector unsigned int *);
7169 void vec_stl (vector unsigned int, int, unsigned int *);
7170 void vec_stl (vector bool int, int, vector bool int *);
7171 void vec_stl (vector bool int, int, unsigned int *);
7172 void vec_stl (vector bool int, int, int *);
7173 void vec_stl (vector signed short, int, vector signed short *);
7174 void vec_stl (vector signed short, int, short *);
7175 void vec_stl (vector unsigned short, int, vector unsigned short *);
7176 void vec_stl (vector unsigned short, int, unsigned short *);
7177 void vec_stl (vector bool short, int, vector bool short *);
7178 void vec_stl (vector bool short, int, unsigned short *);
7179 void vec_stl (vector bool short, int, short *);
7180 void vec_stl (vector pixel, int, vector pixel *);
7181 void vec_stl (vector pixel, int, unsigned short *);
7182 void vec_stl (vector pixel, int, short *);
7183 void vec_stl (vector signed char, int, vector signed char *);
7184 void vec_stl (vector signed char, int, signed char *);
7185 void vec_stl (vector unsigned char, int, vector unsigned char *);
7186 void vec_stl (vector unsigned char, int, unsigned char *);
7187 void vec_stl (vector bool char, int, vector bool char *);
7188 void vec_stl (vector bool char, int, unsigned char *);
7189 void vec_stl (vector bool char, int, signed char *);
7191 vector signed char vec_sub (vector bool char, vector signed char);
7192 vector signed char vec_sub (vector signed char, vector bool char);
7193 vector signed char vec_sub (vector signed char, vector signed char);
7194 vector unsigned char vec_sub (vector bool char, vector unsigned char);
7195 vector unsigned char vec_sub (vector unsigned char, vector bool char);
7196 vector unsigned char vec_sub (vector unsigned char,
7197 vector unsigned char);
7198 vector signed short vec_sub (vector bool short, vector signed short);
7199 vector signed short vec_sub (vector signed short, vector bool short);
7200 vector signed short vec_sub (vector signed short, vector signed short);
7201 vector unsigned short vec_sub (vector bool short,
7202 vector unsigned short);
7203 vector unsigned short vec_sub (vector unsigned short,
7205 vector unsigned short vec_sub (vector unsigned short,
7206 vector unsigned short);
7207 vector signed int vec_sub (vector bool int, vector signed int);
7208 vector signed int vec_sub (vector signed int, vector bool int);
7209 vector signed int vec_sub (vector signed int, vector signed int);
7210 vector unsigned int vec_sub (vector bool int, vector unsigned int);
7211 vector unsigned int vec_sub (vector unsigned int, vector bool int);
7212 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
7213 vector float vec_sub (vector float, vector float);
7215 vector float vec_vsubfp (vector float, vector float);
7217 vector signed int vec_vsubuwm (vector bool int, vector signed int);
7218 vector signed int vec_vsubuwm (vector signed int, vector bool int);
7219 vector signed int vec_vsubuwm (vector signed int, vector signed int);
7220 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
7221 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
7222 vector unsigned int vec_vsubuwm (vector unsigned int,
7223 vector unsigned int);
7225 vector signed short vec_vsubuhm (vector bool short,
7226 vector signed short);
7227 vector signed short vec_vsubuhm (vector signed short,
7229 vector signed short vec_vsubuhm (vector signed short,
7230 vector signed short);
7231 vector unsigned short vec_vsubuhm (vector bool short,
7232 vector unsigned short);
7233 vector unsigned short vec_vsubuhm (vector unsigned short,
7235 vector unsigned short vec_vsubuhm (vector unsigned short,
7236 vector unsigned short);
7238 vector signed char vec_vsububm (vector bool char, vector signed char);
7239 vector signed char vec_vsububm (vector signed char, vector bool char);
7240 vector signed char vec_vsububm (vector signed char, vector signed char);
7241 vector unsigned char vec_vsububm (vector bool char,
7242 vector unsigned char);
7243 vector unsigned char vec_vsububm (vector unsigned char,
7245 vector unsigned char vec_vsububm (vector unsigned char,
7246 vector unsigned char);
7248 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
7250 vector unsigned char vec_subs (vector bool char, vector unsigned char);
7251 vector unsigned char vec_subs (vector unsigned char, vector bool char);
7252 vector unsigned char vec_subs (vector unsigned char,
7253 vector unsigned char);
7254 vector signed char vec_subs (vector bool char, vector signed char);
7255 vector signed char vec_subs (vector signed char, vector bool char);
7256 vector signed char vec_subs (vector signed char, vector signed char);
7257 vector unsigned short vec_subs (vector bool short,
7258 vector unsigned short);
7259 vector unsigned short vec_subs (vector unsigned short,
7261 vector unsigned short vec_subs (vector unsigned short,
7262 vector unsigned short);
7263 vector signed short vec_subs (vector bool short, vector signed short);
7264 vector signed short vec_subs (vector signed short, vector bool short);
7265 vector signed short vec_subs (vector signed short, vector signed short);
7266 vector unsigned int vec_subs (vector bool int, vector unsigned int);
7267 vector unsigned int vec_subs (vector unsigned int, vector bool int);
7268 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
7269 vector signed int vec_subs (vector bool int, vector signed int);
7270 vector signed int vec_subs (vector signed int, vector bool int);
7271 vector signed int vec_subs (vector signed int, vector signed int);
7273 vector signed int vec_vsubsws (vector bool int, vector signed int);
7274 vector signed int vec_vsubsws (vector signed int, vector bool int);
7275 vector signed int vec_vsubsws (vector signed int, vector signed int);
7277 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
7278 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
7279 vector unsigned int vec_vsubuws (vector unsigned int,
7280 vector unsigned int);
7282 vector signed short vec_vsubshs (vector bool short,
7283 vector signed short);
7284 vector signed short vec_vsubshs (vector signed short,
7286 vector signed short vec_vsubshs (vector signed short,
7287 vector signed short);
7289 vector unsigned short vec_vsubuhs (vector bool short,
7290 vector unsigned short);
7291 vector unsigned short vec_vsubuhs (vector unsigned short,
7293 vector unsigned short vec_vsubuhs (vector unsigned short,
7294 vector unsigned short);
7296 vector signed char vec_vsubsbs (vector bool char, vector signed char);
7297 vector signed char vec_vsubsbs (vector signed char, vector bool char);
7298 vector signed char vec_vsubsbs (vector signed char, vector signed char);
7300 vector unsigned char vec_vsububs (vector bool char,
7301 vector unsigned char);
7302 vector unsigned char vec_vsububs (vector unsigned char,
7304 vector unsigned char vec_vsububs (vector unsigned char,
7305 vector unsigned char);
7307 vector unsigned int vec_sum4s (vector unsigned char,
7308 vector unsigned int);
7309 vector signed int vec_sum4s (vector signed char, vector signed int);
7310 vector signed int vec_sum4s (vector signed short, vector signed int);
7312 vector signed int vec_vsum4shs (vector signed short, vector signed int);
7314 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
7316 vector unsigned int vec_vsum4ubs (vector unsigned char,
7317 vector unsigned int);
7319 vector signed int vec_sum2s (vector signed int, vector signed int);
7321 vector signed int vec_sums (vector signed int, vector signed int);
7323 vector float vec_trunc (vector float);
7325 vector signed short vec_unpackh (vector signed char);
7326 vector bool short vec_unpackh (vector bool char);
7327 vector signed int vec_unpackh (vector signed short);
7328 vector bool int vec_unpackh (vector bool short);
7329 vector unsigned int vec_unpackh (vector pixel);
7331 vector bool int vec_vupkhsh (vector bool short);
7332 vector signed int vec_vupkhsh (vector signed short);
7334 vector unsigned int vec_vupkhpx (vector pixel);
7336 vector bool short vec_vupkhsb (vector bool char);
7337 vector signed short vec_vupkhsb (vector signed char);
7339 vector signed short vec_unpackl (vector signed char);
7340 vector bool short vec_unpackl (vector bool char);
7341 vector unsigned int vec_unpackl (vector pixel);
7342 vector signed int vec_unpackl (vector signed short);
7343 vector bool int vec_unpackl (vector bool short);
7345 vector unsigned int vec_vupklpx (vector pixel);
7347 vector bool int vec_vupklsh (vector bool short);
7348 vector signed int vec_vupklsh (vector signed short);
7350 vector bool short vec_vupklsb (vector bool char);
7351 vector signed short vec_vupklsb (vector signed char);
7353 vector float vec_xor (vector float, vector float);
7354 vector float vec_xor (vector float, vector bool int);
7355 vector float vec_xor (vector bool int, vector float);
7356 vector bool int vec_xor (vector bool int, vector bool int);
7357 vector signed int vec_xor (vector bool int, vector signed int);
7358 vector signed int vec_xor (vector signed int, vector bool int);
7359 vector signed int vec_xor (vector signed int, vector signed int);
7360 vector unsigned int vec_xor (vector bool int, vector unsigned int);
7361 vector unsigned int vec_xor (vector unsigned int, vector bool int);
7362 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
7363 vector bool short vec_xor (vector bool short, vector bool short);
7364 vector signed short vec_xor (vector bool short, vector signed short);
7365 vector signed short vec_xor (vector signed short, vector bool short);
7366 vector signed short vec_xor (vector signed short, vector signed short);
7367 vector unsigned short vec_xor (vector bool short,
7368 vector unsigned short);
7369 vector unsigned short vec_xor (vector unsigned short,
7371 vector unsigned short vec_xor (vector unsigned short,
7372 vector unsigned short);
7373 vector signed char vec_xor (vector bool char, vector signed char);
7374 vector bool char vec_xor (vector bool char, vector bool char);
7375 vector signed char vec_xor (vector signed char, vector bool char);
7376 vector signed char vec_xor (vector signed char, vector signed char);
7377 vector unsigned char vec_xor (vector bool char, vector unsigned char);
7378 vector unsigned char vec_xor (vector unsigned char, vector bool char);
7379 vector unsigned char vec_xor (vector unsigned char,
7380 vector unsigned char);
7382 int vec_all_eq (vector signed char, vector bool char);
7383 int vec_all_eq (vector signed char, vector signed char);
7384 int vec_all_eq (vector unsigned char, vector bool char);
7385 int vec_all_eq (vector unsigned char, vector unsigned char);
7386 int vec_all_eq (vector bool char, vector bool char);
7387 int vec_all_eq (vector bool char, vector unsigned char);
7388 int vec_all_eq (vector bool char, vector signed char);
7389 int vec_all_eq (vector signed short, vector bool short);
7390 int vec_all_eq (vector signed short, vector signed short);
7391 int vec_all_eq (vector unsigned short, vector bool short);
7392 int vec_all_eq (vector unsigned short, vector unsigned short);
7393 int vec_all_eq (vector bool short, vector bool short);
7394 int vec_all_eq (vector bool short, vector unsigned short);
7395 int vec_all_eq (vector bool short, vector signed short);
7396 int vec_all_eq (vector pixel, vector pixel);
7397 int vec_all_eq (vector signed int, vector bool int);
7398 int vec_all_eq (vector signed int, vector signed int);
7399 int vec_all_eq (vector unsigned int, vector bool int);
7400 int vec_all_eq (vector unsigned int, vector unsigned int);
7401 int vec_all_eq (vector bool int, vector bool int);
7402 int vec_all_eq (vector bool int, vector unsigned int);
7403 int vec_all_eq (vector bool int, vector signed int);
7404 int vec_all_eq (vector float, vector float);
7406 int vec_all_ge (vector bool char, vector unsigned char);
7407 int vec_all_ge (vector unsigned char, vector bool char);
7408 int vec_all_ge (vector unsigned char, vector unsigned char);
7409 int vec_all_ge (vector bool char, vector signed char);
7410 int vec_all_ge (vector signed char, vector bool char);
7411 int vec_all_ge (vector signed char, vector signed char);
7412 int vec_all_ge (vector bool short, vector unsigned short);
7413 int vec_all_ge (vector unsigned short, vector bool short);
7414 int vec_all_ge (vector unsigned short, vector unsigned short);
7415 int vec_all_ge (vector signed short, vector signed short);
7416 int vec_all_ge (vector bool short, vector signed short);
7417 int vec_all_ge (vector signed short, vector bool short);
7418 int vec_all_ge (vector bool int, vector unsigned int);
7419 int vec_all_ge (vector unsigned int, vector bool int);
7420 int vec_all_ge (vector unsigned int, vector unsigned int);
7421 int vec_all_ge (vector bool int, vector signed int);
7422 int vec_all_ge (vector signed int, vector bool int);
7423 int vec_all_ge (vector signed int, vector signed int);
7424 int vec_all_ge (vector float, vector float);
7426 int vec_all_gt (vector bool char, vector unsigned char);
7427 int vec_all_gt (vector unsigned char, vector bool char);
7428 int vec_all_gt (vector unsigned char, vector unsigned char);
7429 int vec_all_gt (vector bool char, vector signed char);
7430 int vec_all_gt (vector signed char, vector bool char);
7431 int vec_all_gt (vector signed char, vector signed char);
7432 int vec_all_gt (vector bool short, vector unsigned short);
7433 int vec_all_gt (vector unsigned short, vector bool short);
7434 int vec_all_gt (vector unsigned short, vector unsigned short);
7435 int vec_all_gt (vector bool short, vector signed short);
7436 int vec_all_gt (vector signed short, vector bool short);
7437 int vec_all_gt (vector signed short, vector signed short);
7438 int vec_all_gt (vector bool int, vector unsigned int);
7439 int vec_all_gt (vector unsigned int, vector bool int);
7440 int vec_all_gt (vector unsigned int, vector unsigned int);
7441 int vec_all_gt (vector bool int, vector signed int);
7442 int vec_all_gt (vector signed int, vector bool int);
7443 int vec_all_gt (vector signed int, vector signed int);
7444 int vec_all_gt (vector float, vector float);
7446 int vec_all_in (vector float, vector float);
7448 int vec_all_le (vector bool char, vector unsigned char);
7449 int vec_all_le (vector unsigned char, vector bool char);
7450 int vec_all_le (vector unsigned char, vector unsigned char);
7451 int vec_all_le (vector bool char, vector signed char);
7452 int vec_all_le (vector signed char, vector bool char);
7453 int vec_all_le (vector signed char, vector signed char);
7454 int vec_all_le (vector bool short, vector unsigned short);
7455 int vec_all_le (vector unsigned short, vector bool short);
7456 int vec_all_le (vector unsigned short, vector unsigned short);
7457 int vec_all_le (vector bool short, vector signed short);
7458 int vec_all_le (vector signed short, vector bool short);
7459 int vec_all_le (vector signed short, vector signed short);
7460 int vec_all_le (vector bool int, vector unsigned int);
7461 int vec_all_le (vector unsigned int, vector bool int);
7462 int vec_all_le (vector unsigned int, vector unsigned int);
7463 int vec_all_le (vector bool int, vector signed int);
7464 int vec_all_le (vector signed int, vector bool int);
7465 int vec_all_le (vector signed int, vector signed int);
7466 int vec_all_le (vector float, vector float);
7468 int vec_all_lt (vector bool char, vector unsigned char);
7469 int vec_all_lt (vector unsigned char, vector bool char);
7470 int vec_all_lt (vector unsigned char, vector unsigned char);
7471 int vec_all_lt (vector bool char, vector signed char);
7472 int vec_all_lt (vector signed char, vector bool char);
7473 int vec_all_lt (vector signed char, vector signed char);
7474 int vec_all_lt (vector bool short, vector unsigned short);
7475 int vec_all_lt (vector unsigned short, vector bool short);
7476 int vec_all_lt (vector unsigned short, vector unsigned short);
7477 int vec_all_lt (vector bool short, vector signed short);
7478 int vec_all_lt (vector signed short, vector bool short);
7479 int vec_all_lt (vector signed short, vector signed short);
7480 int vec_all_lt (vector bool int, vector unsigned int);
7481 int vec_all_lt (vector unsigned int, vector bool int);
7482 int vec_all_lt (vector unsigned int, vector unsigned int);
7483 int vec_all_lt (vector bool int, vector signed int);
7484 int vec_all_lt (vector signed int, vector bool int);
7485 int vec_all_lt (vector signed int, vector signed int);
7486 int vec_all_lt (vector float, vector float);
7488 int vec_all_nan (vector float);
7490 int vec_all_ne (vector signed char, vector bool char);
7491 int vec_all_ne (vector signed char, vector signed char);
7492 int vec_all_ne (vector unsigned char, vector bool char);
7493 int vec_all_ne (vector unsigned char, vector unsigned char);
7494 int vec_all_ne (vector bool char, vector bool char);
7495 int vec_all_ne (vector bool char, vector unsigned char);
7496 int vec_all_ne (vector bool char, vector signed char);
7497 int vec_all_ne (vector signed short, vector bool short);
7498 int vec_all_ne (vector signed short, vector signed short);
7499 int vec_all_ne (vector unsigned short, vector bool short);
7500 int vec_all_ne (vector unsigned short, vector unsigned short);
7501 int vec_all_ne (vector bool short, vector bool short);
7502 int vec_all_ne (vector bool short, vector unsigned short);
7503 int vec_all_ne (vector bool short, vector signed short);
7504 int vec_all_ne (vector pixel, vector pixel);
7505 int vec_all_ne (vector signed int, vector bool int);
7506 int vec_all_ne (vector signed int, vector signed int);
7507 int vec_all_ne (vector unsigned int, vector bool int);
7508 int vec_all_ne (vector unsigned int, vector unsigned int);
7509 int vec_all_ne (vector bool int, vector bool int);
7510 int vec_all_ne (vector bool int, vector unsigned int);
7511 int vec_all_ne (vector bool int, vector signed int);
7512 int vec_all_ne (vector float, vector float);
7514 int vec_all_nge (vector float, vector float);
7516 int vec_all_ngt (vector float, vector float);
7518 int vec_all_nle (vector float, vector float);
7520 int vec_all_nlt (vector float, vector float);
7522 int vec_all_numeric (vector float);
7524 int vec_any_eq (vector signed char, vector bool char);
7525 int vec_any_eq (vector signed char, vector signed char);
7526 int vec_any_eq (vector unsigned char, vector bool char);
7527 int vec_any_eq (vector unsigned char, vector unsigned char);
7528 int vec_any_eq (vector bool char, vector bool char);
7529 int vec_any_eq (vector bool char, vector unsigned char);
7530 int vec_any_eq (vector bool char, vector signed char);
7531 int vec_any_eq (vector signed short, vector bool short);
7532 int vec_any_eq (vector signed short, vector signed short);
7533 int vec_any_eq (vector unsigned short, vector bool short);
7534 int vec_any_eq (vector unsigned short, vector unsigned short);
7535 int vec_any_eq (vector bool short, vector bool short);
7536 int vec_any_eq (vector bool short, vector unsigned short);
7537 int vec_any_eq (vector bool short, vector signed short);
7538 int vec_any_eq (vector pixel, vector pixel);
7539 int vec_any_eq (vector signed int, vector bool int);
7540 int vec_any_eq (vector signed int, vector signed int);
7541 int vec_any_eq (vector unsigned int, vector bool int);
7542 int vec_any_eq (vector unsigned int, vector unsigned int);
7543 int vec_any_eq (vector bool int, vector bool int);
7544 int vec_any_eq (vector bool int, vector unsigned int);
7545 int vec_any_eq (vector bool int, vector signed int);
7546 int vec_any_eq (vector float, vector float);
7548 int vec_any_ge (vector signed char, vector bool char);
7549 int vec_any_ge (vector unsigned char, vector bool char);
7550 int vec_any_ge (vector unsigned char, vector unsigned char);
7551 int vec_any_ge (vector signed char, vector signed char);
7552 int vec_any_ge (vector bool char, vector unsigned char);
7553 int vec_any_ge (vector bool char, vector signed char);
7554 int vec_any_ge (vector unsigned short, vector bool short);
7555 int vec_any_ge (vector unsigned short, vector unsigned short);
7556 int vec_any_ge (vector signed short, vector signed short);
7557 int vec_any_ge (vector signed short, vector bool short);
7558 int vec_any_ge (vector bool short, vector unsigned short);
7559 int vec_any_ge (vector bool short, vector signed short);
7560 int vec_any_ge (vector signed int, vector bool int);
7561 int vec_any_ge (vector unsigned int, vector bool int);
7562 int vec_any_ge (vector unsigned int, vector unsigned int);
7563 int vec_any_ge (vector signed int, vector signed int);
7564 int vec_any_ge (vector bool int, vector unsigned int);
7565 int vec_any_ge (vector bool int, vector signed int);
7566 int vec_any_ge (vector float, vector float);
7568 int vec_any_gt (vector bool char, vector unsigned char);
7569 int vec_any_gt (vector unsigned char, vector bool char);
7570 int vec_any_gt (vector unsigned char, vector unsigned char);
7571 int vec_any_gt (vector bool char, vector signed char);
7572 int vec_any_gt (vector signed char, vector bool char);
7573 int vec_any_gt (vector signed char, vector signed char);
7574 int vec_any_gt (vector bool short, vector unsigned short);
7575 int vec_any_gt (vector unsigned short, vector bool short);
7576 int vec_any_gt (vector unsigned short, vector unsigned short);
7577 int vec_any_gt (vector bool short, vector signed short);
7578 int vec_any_gt (vector signed short, vector bool short);
7579 int vec_any_gt (vector signed short, vector signed short);
7580 int vec_any_gt (vector bool int, vector unsigned int);
7581 int vec_any_gt (vector unsigned int, vector bool int);
7582 int vec_any_gt (vector unsigned int, vector unsigned int);
7583 int vec_any_gt (vector bool int, vector signed int);
7584 int vec_any_gt (vector signed int, vector bool int);
7585 int vec_any_gt (vector signed int, vector signed int);
7586 int vec_any_gt (vector float, vector float);
7588 int vec_any_le (vector bool char, vector unsigned char);
7589 int vec_any_le (vector unsigned char, vector bool char);
7590 int vec_any_le (vector unsigned char, vector unsigned char);
7591 int vec_any_le (vector bool char, vector signed char);
7592 int vec_any_le (vector signed char, vector bool char);
7593 int vec_any_le (vector signed char, vector signed char);
7594 int vec_any_le (vector bool short, vector unsigned short);
7595 int vec_any_le (vector unsigned short, vector bool short);
7596 int vec_any_le (vector unsigned short, vector unsigned short);
7597 int vec_any_le (vector bool short, vector signed short);
7598 int vec_any_le (vector signed short, vector bool short);
7599 int vec_any_le (vector signed short, vector signed short);
7600 int vec_any_le (vector bool int, vector unsigned int);
7601 int vec_any_le (vector unsigned int, vector bool int);
7602 int vec_any_le (vector unsigned int, vector unsigned int);
7603 int vec_any_le (vector bool int, vector signed int);
7604 int vec_any_le (vector signed int, vector bool int);
7605 int vec_any_le (vector signed int, vector signed int);
7606 int vec_any_le (vector float, vector float);
7608 int vec_any_lt (vector bool char, vector unsigned char);
7609 int vec_any_lt (vector unsigned char, vector bool char);
7610 int vec_any_lt (vector unsigned char, vector unsigned char);
7611 int vec_any_lt (vector bool char, vector signed char);
7612 int vec_any_lt (vector signed char, vector bool char);
7613 int vec_any_lt (vector signed char, vector signed char);
7614 int vec_any_lt (vector bool short, vector unsigned short);
7615 int vec_any_lt (vector unsigned short, vector bool short);
7616 int vec_any_lt (vector unsigned short, vector unsigned short);
7617 int vec_any_lt (vector bool short, vector signed short);
7618 int vec_any_lt (vector signed short, vector bool short);
7619 int vec_any_lt (vector signed short, vector signed short);
7620 int vec_any_lt (vector bool int, vector unsigned int);
7621 int vec_any_lt (vector unsigned int, vector bool int);
7622 int vec_any_lt (vector unsigned int, vector unsigned int);
7623 int vec_any_lt (vector bool int, vector signed int);
7624 int vec_any_lt (vector signed int, vector bool int);
7625 int vec_any_lt (vector signed int, vector signed int);
7626 int vec_any_lt (vector float, vector float);
7628 int vec_any_nan (vector float);
7630 int vec_any_ne (vector signed char, vector bool char);
7631 int vec_any_ne (vector signed char, vector signed char);
7632 int vec_any_ne (vector unsigned char, vector bool char);
7633 int vec_any_ne (vector unsigned char, vector unsigned char);
7634 int vec_any_ne (vector bool char, vector bool char);
7635 int vec_any_ne (vector bool char, vector unsigned char);
7636 int vec_any_ne (vector bool char, vector signed char);
7637 int vec_any_ne (vector signed short, vector bool short);
7638 int vec_any_ne (vector signed short, vector signed short);
7639 int vec_any_ne (vector unsigned short, vector bool short);
7640 int vec_any_ne (vector unsigned short, vector unsigned short);
7641 int vec_any_ne (vector bool short, vector bool short);
7642 int vec_any_ne (vector bool short, vector unsigned short);
7643 int vec_any_ne (vector bool short, vector signed short);
7644 int vec_any_ne (vector pixel, vector pixel);
7645 int vec_any_ne (vector signed int, vector bool int);
7646 int vec_any_ne (vector signed int, vector signed int);
7647 int vec_any_ne (vector unsigned int, vector bool int);
7648 int vec_any_ne (vector unsigned int, vector unsigned int);
7649 int vec_any_ne (vector bool int, vector bool int);
7650 int vec_any_ne (vector bool int, vector unsigned int);
7651 int vec_any_ne (vector bool int, vector signed int);
7652 int vec_any_ne (vector float, vector float);
7654 int vec_any_nge (vector float, vector float);
7656 int vec_any_ngt (vector float, vector float);
7658 int vec_any_nle (vector float, vector float);
7660 int vec_any_nlt (vector float, vector float);
7662 int vec_any_numeric (vector float);
7664 int vec_any_out (vector float, vector float);
7667 @node Target Format Checks
7668 @section Format Checks Specific to Particular Target Machines
7670 For some target machines, GCC supports additional options to the
7672 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
7675 * Solaris Format Checks::
7678 @node Solaris Format Checks
7679 @subsection Solaris Format Checks
7681 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
7682 check. @code{cmn_err} accepts a subset of the standard @code{printf}
7683 conversions, and the two-argument @code{%b} conversion for displaying
7684 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
7687 @section Pragmas Accepted by GCC
7691 GCC supports several types of pragmas, primarily in order to compile
7692 code originally written for other compilers. Note that in general
7693 we do not recommend the use of pragmas; @xref{Function Attributes},
7694 for further explanation.
7698 * RS/6000 and PowerPC Pragmas::
7701 * Symbol-Renaming Pragmas::
7702 * Structure-Packing Pragmas::
7706 @subsection ARM Pragmas
7708 The ARM target defines pragmas for controlling the default addition of
7709 @code{long_call} and @code{short_call} attributes to functions.
7710 @xref{Function Attributes}, for information about the effects of these
7715 @cindex pragma, long_calls
7716 Set all subsequent functions to have the @code{long_call} attribute.
7719 @cindex pragma, no_long_calls
7720 Set all subsequent functions to have the @code{short_call} attribute.
7722 @item long_calls_off
7723 @cindex pragma, long_calls_off
7724 Do not affect the @code{long_call} or @code{short_call} attributes of
7725 subsequent functions.
7728 @node RS/6000 and PowerPC Pragmas
7729 @subsection RS/6000 and PowerPC Pragmas
7731 The RS/6000 and PowerPC targets define one pragma for controlling
7732 whether or not the @code{longcall} attribute is added to function
7733 declarations by default. This pragma overrides the @option{-mlongcall}
7734 option, but not the @code{longcall} and @code{shortcall} attributes.
7735 @xref{RS/6000 and PowerPC Options}, for more information about when long
7736 calls are and are not necessary.
7740 @cindex pragma, longcall
7741 Apply the @code{longcall} attribute to all subsequent function
7745 Do not apply the @code{longcall} attribute to subsequent function
7749 @c Describe c4x pragmas here.
7750 @c Describe h8300 pragmas here.
7751 @c Describe sh pragmas here.
7752 @c Describe v850 pragmas here.
7754 @node Darwin Pragmas
7755 @subsection Darwin Pragmas
7757 The following pragmas are available for all architectures running the
7758 Darwin operating system. These are useful for compatibility with other
7762 @item mark @var{tokens}@dots{}
7763 @cindex pragma, mark
7764 This pragma is accepted, but has no effect.
7766 @item options align=@var{alignment}
7767 @cindex pragma, options align
7768 This pragma sets the alignment of fields in structures. The values of
7769 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
7770 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
7771 properly; to restore the previous setting, use @code{reset} for the
7774 @item segment @var{tokens}@dots{}
7775 @cindex pragma, segment
7776 This pragma is accepted, but has no effect.
7778 @item unused (@var{var} [, @var{var}]@dots{})
7779 @cindex pragma, unused
7780 This pragma declares variables to be possibly unused. GCC will not
7781 produce warnings for the listed variables. The effect is similar to
7782 that of the @code{unused} attribute, except that this pragma may appear
7783 anywhere within the variables' scopes.
7786 @node Solaris Pragmas
7787 @subsection Solaris Pragmas
7789 The Solaris target supports @code{#pragma redefine_extname}
7790 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
7791 @code{#pragma} directives for compatibility with the system compiler.
7794 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
7795 @cindex pragma, align
7797 Increase the minimum alignment of each @var{variable} to @var{alignment}.
7798 This is the same as GCC's @code{aligned} attribute @pxref{Variable
7801 @item fini (@var{function} [, @var{function}]...)
7802 @cindex pragma, fini
7804 This pragma causes each listed @var{function} to be called after
7805 main, or during shared module unloading, by adding a call to the
7806 @code{.fini} section.
7808 @item init (@var{function} [, @var{function}]...)
7809 @cindex pragma, init
7811 This pragma causes each listed @var{function} to be called during
7812 initialization (before @code{main}) or during shared module loading, by
7813 adding a call to the @code{.init} section.
7817 @node Symbol-Renaming Pragmas
7818 @subsection Symbol-Renaming Pragmas
7820 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
7821 supports two @code{#pragma} directives which change the name used in
7822 assembly for a given declaration. These pragmas are only available on
7823 platforms whose system headers need them. To get this effect on all
7824 platforms supported by GCC, use the asm labels extension (@pxref{Asm
7828 @item redefine_extname @var{oldname} @var{newname}
7829 @cindex pragma, redefine_extname
7831 This pragma gives the C function @var{oldname} the assembly symbol
7832 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
7833 will be defined if this pragma is available (currently only on
7836 @item extern_prefix @var{string}
7837 @cindex pragma, extern_prefix
7839 This pragma causes all subsequent external function and variable
7840 declarations to have @var{string} prepended to their assembly symbols.
7841 This effect may be terminated with another @code{extern_prefix} pragma
7842 whose argument is an empty string. The preprocessor macro
7843 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
7844 available (currently only on Tru64 UNIX).
7847 These pragmas and the asm labels extension interact in a complicated
7848 manner. Here are some corner cases you may want to be aware of.
7851 @item Both pragmas silently apply only to declarations with external
7852 linkage. Asm labels do not have this restriction.
7854 @item In C++, both pragmas silently apply only to declarations with
7855 ``C'' linkage. Again, asm labels do not have this restriction.
7857 @item If any of the three ways of changing the assembly name of a
7858 declaration is applied to a declaration whose assembly name has
7859 already been determined (either by a previous use of one of these
7860 features, or because the compiler needed the assembly name in order to
7861 generate code), and the new name is different, a warning issues and
7862 the name does not change.
7864 @item The @var{oldname} used by @code{#pragma redefine_extname} is
7865 always the C-language name.
7867 @item If @code{#pragma extern_prefix} is in effect, and a declaration
7868 occurs with an asm label attached, the prefix is silently ignored for
7871 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
7872 apply to the same declaration, whichever triggered first wins, and a
7873 warning issues if they contradict each other. (We would like to have
7874 @code{#pragma redefine_extname} always win, for consistency with asm
7875 labels, but if @code{#pragma extern_prefix} triggers first we have no
7876 way of knowing that that happened.)
7879 @node Structure-Packing Pragmas
7880 @subsection Structure-Packing Pragmas
7882 For compatibility with Win32, GCC supports as set of @code{#pragma}
7883 directives which change the maximum alignment of members of structures,
7884 unions, and classes subsequently defined. The @var{n} value below always
7885 is required to be a small power of two and specifies the new alignment
7889 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
7890 @item @code{#pragma pack()} sets the alignment to the one that was in
7891 effect when compilation started (see also command line option
7892 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
7893 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
7894 setting on an internal stack and then optionally sets the new alignment.
7895 @item @code{#pragma pack(pop)} restores the alignment setting to the one
7896 saved at the top of the internal stack (and removes that stack entry).
7897 Note that @code{#pragma pack([@var{n}])} does not influence this internal
7898 stack; thus it is possible to have @code{#pragma pack(push)} followed by
7899 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
7900 @code{#pragma pack(pop)}.
7903 @node Unnamed Fields
7904 @section Unnamed struct/union fields within structs/unions.
7908 For compatibility with other compilers, GCC allows you to define
7909 a structure or union that contains, as fields, structures and unions
7910 without names. For example:
7923 In this example, the user would be able to access members of the unnamed
7924 union with code like @samp{foo.b}. Note that only unnamed structs and
7925 unions are allowed, you may not have, for example, an unnamed
7928 You must never create such structures that cause ambiguous field definitions.
7929 For example, this structure:
7940 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
7941 Such constructs are not supported and must be avoided. In the future,
7942 such constructs may be detected and treated as compilation errors.
7945 @section Thread-Local Storage
7946 @cindex Thread-Local Storage
7947 @cindex @acronym{TLS}
7950 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
7951 are allocated such that there is one instance of the variable per extant
7952 thread. The run-time model GCC uses to implement this originates
7953 in the IA-64 processor-specific ABI, but has since been migrated
7954 to other processors as well. It requires significant support from
7955 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
7956 system libraries (@file{libc.so} and @file{libpthread.so}), so it
7957 is not available everywhere.
7959 At the user level, the extension is visible with a new storage
7960 class keyword: @code{__thread}. For example:
7964 extern __thread struct state s;
7965 static __thread char *p;
7968 The @code{__thread} specifier may be used alone, with the @code{extern}
7969 or @code{static} specifiers, but with no other storage class specifier.
7970 When used with @code{extern} or @code{static}, @code{__thread} must appear
7971 immediately after the other storage class specifier.
7973 The @code{__thread} specifier may be applied to any global, file-scoped
7974 static, function-scoped static, or static data member of a class. It may
7975 not be applied to block-scoped automatic or non-static data member.
7977 When the address-of operator is applied to a thread-local variable, it is
7978 evaluated at run-time and returns the address of the current thread's
7979 instance of that variable. An address so obtained may be used by any
7980 thread. When a thread terminates, any pointers to thread-local variables
7981 in that thread become invalid.
7983 No static initialization may refer to the address of a thread-local variable.
7985 In C++, if an initializer is present for a thread-local variable, it must
7986 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
7989 See @uref{http://people.redhat.com/drepper/tls.pdf,
7990 ELF Handling For Thread-Local Storage} for a detailed explanation of
7991 the four thread-local storage addressing models, and how the run-time
7992 is expected to function.
7995 * C99 Thread-Local Edits::
7996 * C++98 Thread-Local Edits::
7999 @node C99 Thread-Local Edits
8000 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
8002 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
8003 that document the exact semantics of the language extension.
8007 @cite{5.1.2 Execution environments}
8009 Add new text after paragraph 1
8012 Within either execution environment, a @dfn{thread} is a flow of
8013 control within a program. It is implementation defined whether
8014 or not there may be more than one thread associated with a program.
8015 It is implementation defined how threads beyond the first are
8016 created, the name and type of the function called at thread
8017 startup, and how threads may be terminated. However, objects
8018 with thread storage duration shall be initialized before thread
8023 @cite{6.2.4 Storage durations of objects}
8025 Add new text before paragraph 3
8028 An object whose identifier is declared with the storage-class
8029 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
8030 Its lifetime is the entire execution of the thread, and its
8031 stored value is initialized only once, prior to thread startup.
8035 @cite{6.4.1 Keywords}
8037 Add @code{__thread}.
8040 @cite{6.7.1 Storage-class specifiers}
8042 Add @code{__thread} to the list of storage class specifiers in
8045 Change paragraph 2 to
8048 With the exception of @code{__thread}, at most one storage-class
8049 specifier may be given [@dots{}]. The @code{__thread} specifier may
8050 be used alone, or immediately following @code{extern} or
8054 Add new text after paragraph 6
8057 The declaration of an identifier for a variable that has
8058 block scope that specifies @code{__thread} shall also
8059 specify either @code{extern} or @code{static}.
8061 The @code{__thread} specifier shall be used only with
8066 @node C++98 Thread-Local Edits
8067 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
8069 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
8070 that document the exact semantics of the language extension.
8074 @b{[intro.execution]}
8076 New text after paragraph 4
8079 A @dfn{thread} is a flow of control within the abstract machine.
8080 It is implementation defined whether or not there may be more than
8084 New text after paragraph 7
8087 It is unspecified whether additional action must be taken to
8088 ensure when and whether side effects are visible to other threads.
8094 Add @code{__thread}.
8097 @b{[basic.start.main]}
8099 Add after paragraph 5
8102 The thread that begins execution at the @code{main} function is called
8103 the @dfn{main thread}. It is implementation defined how functions
8104 beginning threads other than the main thread are designated or typed.
8105 A function so designated, as well as the @code{main} function, is called
8106 a @dfn{thread startup function}. It is implementation defined what
8107 happens if a thread startup function returns. It is implementation
8108 defined what happens to other threads when any thread calls @code{exit}.
8112 @b{[basic.start.init]}
8114 Add after paragraph 4
8117 The storage for an object of thread storage duration shall be
8118 statically initialized before the first statement of the thread startup
8119 function. An object of thread storage duration shall not require
8120 dynamic initialization.
8124 @b{[basic.start.term]}
8126 Add after paragraph 3
8129 The type of an object with thread storage duration shall not have a
8130 non-trivial destructor, nor shall it be an array type whose elements
8131 (directly or indirectly) have non-trivial destructors.
8137 Add ``thread storage duration'' to the list in paragraph 1.
8142 Thread, static, and automatic storage durations are associated with
8143 objects introduced by declarations [@dots{}].
8146 Add @code{__thread} to the list of specifiers in paragraph 3.
8149 @b{[basic.stc.thread]}
8151 New section before @b{[basic.stc.static]}
8154 The keyword @code{__thread} applied to a non-local object gives the
8155 object thread storage duration.
8157 A local variable or class data member declared both @code{static}
8158 and @code{__thread} gives the variable or member thread storage
8163 @b{[basic.stc.static]}
8168 All objects which have neither thread storage duration, dynamic
8169 storage duration nor are local [@dots{}].
8175 Add @code{__thread} to the list in paragraph 1.
8180 With the exception of @code{__thread}, at most one
8181 @var{storage-class-specifier} shall appear in a given
8182 @var{decl-specifier-seq}. The @code{__thread} specifier may
8183 be used alone, or immediately following the @code{extern} or
8184 @code{static} specifiers. [@dots{}]
8187 Add after paragraph 5
8190 The @code{__thread} specifier can be applied only to the names of objects
8191 and to anonymous unions.
8197 Add after paragraph 6
8200 Non-@code{static} members shall not be @code{__thread}.
8204 @node C++ Extensions
8205 @chapter Extensions to the C++ Language
8206 @cindex extensions, C++ language
8207 @cindex C++ language extensions
8209 The GNU compiler provides these extensions to the C++ language (and you
8210 can also use most of the C language extensions in your C++ programs). If you
8211 want to write code that checks whether these features are available, you can
8212 test for the GNU compiler the same way as for C programs: check for a
8213 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
8214 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
8215 Predefined Macros,cpp,The GNU C Preprocessor}).
8218 * Min and Max:: C++ Minimum and maximum operators.
8219 * Volatiles:: What constitutes an access to a volatile object.
8220 * Restricted Pointers:: C99 restricted pointers and references.
8221 * Vague Linkage:: Where G++ puts inlines, vtables and such.
8222 * C++ Interface:: You can use a single C++ header file for both
8223 declarations and definitions.
8224 * Template Instantiation:: Methods for ensuring that exactly one copy of
8225 each needed template instantiation is emitted.
8226 * Bound member functions:: You can extract a function pointer to the
8227 method denoted by a @samp{->*} or @samp{.*} expression.
8228 * C++ Attributes:: Variable, function, and type attributes for C++ only.
8229 * Strong Using:: Strong using-directives for namespace composition.
8230 * Java Exceptions:: Tweaking exception handling to work with Java.
8231 * Deprecated Features:: Things will disappear from g++.
8232 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
8236 @section Minimum and Maximum Operators in C++
8238 It is very convenient to have operators which return the ``minimum'' or the
8239 ``maximum'' of two arguments. In GNU C++ (but not in GNU C),
8242 @item @var{a} <? @var{b}
8244 @cindex minimum operator
8245 is the @dfn{minimum}, returning the smaller of the numeric values
8246 @var{a} and @var{b};
8248 @item @var{a} >? @var{b}
8250 @cindex maximum operator
8251 is the @dfn{maximum}, returning the larger of the numeric values @var{a}
8255 These operations are not primitive in ordinary C++, since you can
8256 use a macro to return the minimum of two things in C++, as in the
8260 #define MIN(X,Y) ((X) < (Y) ? : (X) : (Y))
8264 You might then use @w{@samp{int min = MIN (i, j);}} to set @var{min} to
8265 the minimum value of variables @var{i} and @var{j}.
8267 However, side effects in @code{X} or @code{Y} may cause unintended
8268 behavior. For example, @code{MIN (i++, j++)} will fail, incrementing
8269 the smaller counter twice. The GNU C @code{typeof} extension allows you
8270 to write safe macros that avoid this kind of problem (@pxref{Typeof}).
8271 However, writing @code{MIN} and @code{MAX} as macros also forces you to
8272 use function-call notation for a fundamental arithmetic operation.
8273 Using GNU C++ extensions, you can write @w{@samp{int min = i <? j;}}
8276 Since @code{<?} and @code{>?} are built into the compiler, they properly
8277 handle expressions with side-effects; @w{@samp{int min = i++ <? j++;}}
8281 @section When is a Volatile Object Accessed?
8282 @cindex accessing volatiles
8283 @cindex volatile read
8284 @cindex volatile write
8285 @cindex volatile access
8287 Both the C and C++ standard have the concept of volatile objects. These
8288 are normally accessed by pointers and used for accessing hardware. The
8289 standards encourage compilers to refrain from optimizations
8290 concerning accesses to volatile objects that it might perform on
8291 non-volatile objects. The C standard leaves it implementation defined
8292 as to what constitutes a volatile access. The C++ standard omits to
8293 specify this, except to say that C++ should behave in a similar manner
8294 to C with respect to volatiles, where possible. The minimum either
8295 standard specifies is that at a sequence point all previous accesses to
8296 volatile objects have stabilized and no subsequent accesses have
8297 occurred. Thus an implementation is free to reorder and combine
8298 volatile accesses which occur between sequence points, but cannot do so
8299 for accesses across a sequence point. The use of volatiles does not
8300 allow you to violate the restriction on updating objects multiple times
8301 within a sequence point.
8303 In most expressions, it is intuitively obvious what is a read and what is
8304 a write. For instance
8307 volatile int *dst = @var{somevalue};
8308 volatile int *src = @var{someothervalue};
8313 will cause a read of the volatile object pointed to by @var{src} and stores the
8314 value into the volatile object pointed to by @var{dst}. There is no
8315 guarantee that these reads and writes are atomic, especially for objects
8316 larger than @code{int}.
8318 Less obvious expressions are where something which looks like an access
8319 is used in a void context. An example would be,
8322 volatile int *src = @var{somevalue};
8326 With C, such expressions are rvalues, and as rvalues cause a read of
8327 the object, GCC interprets this as a read of the volatile being pointed
8328 to. The C++ standard specifies that such expressions do not undergo
8329 lvalue to rvalue conversion, and that the type of the dereferenced
8330 object may be incomplete. The C++ standard does not specify explicitly
8331 that it is this lvalue to rvalue conversion which is responsible for
8332 causing an access. However, there is reason to believe that it is,
8333 because otherwise certain simple expressions become undefined. However,
8334 because it would surprise most programmers, G++ treats dereferencing a
8335 pointer to volatile object of complete type in a void context as a read
8336 of the object. When the object has incomplete type, G++ issues a
8341 struct T @{int m;@};
8342 volatile S *ptr1 = @var{somevalue};
8343 volatile T *ptr2 = @var{somevalue};
8348 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
8349 causes a read of the object pointed to. If you wish to force an error on
8350 the first case, you must force a conversion to rvalue with, for instance
8351 a static cast, @code{static_cast<S>(*ptr1)}.
8353 When using a reference to volatile, G++ does not treat equivalent
8354 expressions as accesses to volatiles, but instead issues a warning that
8355 no volatile is accessed. The rationale for this is that otherwise it
8356 becomes difficult to determine where volatile access occur, and not
8357 possible to ignore the return value from functions returning volatile
8358 references. Again, if you wish to force a read, cast the reference to
8361 @node Restricted Pointers
8362 @section Restricting Pointer Aliasing
8363 @cindex restricted pointers
8364 @cindex restricted references
8365 @cindex restricted this pointer
8367 As with the C front end, G++ understands the C99 feature of restricted pointers,
8368 specified with the @code{__restrict__}, or @code{__restrict} type
8369 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
8370 language flag, @code{restrict} is not a keyword in C++.
8372 In addition to allowing restricted pointers, you can specify restricted
8373 references, which indicate that the reference is not aliased in the local
8377 void fn (int *__restrict__ rptr, int &__restrict__ rref)
8384 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
8385 @var{rref} refers to a (different) unaliased integer.
8387 You may also specify whether a member function's @var{this} pointer is
8388 unaliased by using @code{__restrict__} as a member function qualifier.
8391 void T::fn () __restrict__
8398 Within the body of @code{T::fn}, @var{this} will have the effective
8399 definition @code{T *__restrict__ const this}. Notice that the
8400 interpretation of a @code{__restrict__} member function qualifier is
8401 different to that of @code{const} or @code{volatile} qualifier, in that it
8402 is applied to the pointer rather than the object. This is consistent with
8403 other compilers which implement restricted pointers.
8405 As with all outermost parameter qualifiers, @code{__restrict__} is
8406 ignored in function definition matching. This means you only need to
8407 specify @code{__restrict__} in a function definition, rather than
8408 in a function prototype as well.
8411 @section Vague Linkage
8412 @cindex vague linkage
8414 There are several constructs in C++ which require space in the object
8415 file but are not clearly tied to a single translation unit. We say that
8416 these constructs have ``vague linkage''. Typically such constructs are
8417 emitted wherever they are needed, though sometimes we can be more
8421 @item Inline Functions
8422 Inline functions are typically defined in a header file which can be
8423 included in many different compilations. Hopefully they can usually be
8424 inlined, but sometimes an out-of-line copy is necessary, if the address
8425 of the function is taken or if inlining fails. In general, we emit an
8426 out-of-line copy in all translation units where one is needed. As an
8427 exception, we only emit inline virtual functions with the vtable, since
8428 it will always require a copy.
8430 Local static variables and string constants used in an inline function
8431 are also considered to have vague linkage, since they must be shared
8432 between all inlined and out-of-line instances of the function.
8436 C++ virtual functions are implemented in most compilers using a lookup
8437 table, known as a vtable. The vtable contains pointers to the virtual
8438 functions provided by a class, and each object of the class contains a
8439 pointer to its vtable (or vtables, in some multiple-inheritance
8440 situations). If the class declares any non-inline, non-pure virtual
8441 functions, the first one is chosen as the ``key method'' for the class,
8442 and the vtable is only emitted in the translation unit where the key
8445 @emph{Note:} If the chosen key method is later defined as inline, the
8446 vtable will still be emitted in every translation unit which defines it.
8447 Make sure that any inline virtuals are declared inline in the class
8448 body, even if they are not defined there.
8450 @item type_info objects
8453 C++ requires information about types to be written out in order to
8454 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
8455 For polymorphic classes (classes with virtual functions), the type_info
8456 object is written out along with the vtable so that @samp{dynamic_cast}
8457 can determine the dynamic type of a class object at runtime. For all
8458 other types, we write out the type_info object when it is used: when
8459 applying @samp{typeid} to an expression, throwing an object, or
8460 referring to a type in a catch clause or exception specification.
8462 @item Template Instantiations
8463 Most everything in this section also applies to template instantiations,
8464 but there are other options as well.
8465 @xref{Template Instantiation,,Where's the Template?}.
8469 When used with GNU ld version 2.8 or later on an ELF system such as
8470 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
8471 these constructs will be discarded at link time. This is known as
8474 On targets that don't support COMDAT, but do support weak symbols, GCC
8475 will use them. This way one copy will override all the others, but
8476 the unused copies will still take up space in the executable.
8478 For targets which do not support either COMDAT or weak symbols,
8479 most entities with vague linkage will be emitted as local symbols to
8480 avoid duplicate definition errors from the linker. This will not happen
8481 for local statics in inlines, however, as having multiple copies will
8482 almost certainly break things.
8484 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
8485 another way to control placement of these constructs.
8488 @section #pragma interface and implementation
8490 @cindex interface and implementation headers, C++
8491 @cindex C++ interface and implementation headers
8492 @cindex pragmas, interface and implementation
8494 @code{#pragma interface} and @code{#pragma implementation} provide the
8495 user with a way of explicitly directing the compiler to emit entities
8496 with vague linkage (and debugging information) in a particular
8499 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
8500 most cases, because of COMDAT support and the ``key method'' heuristic
8501 mentioned in @ref{Vague Linkage}. Using them can actually cause your
8502 program to grow due to unnecesary out-of-line copies of inline
8503 functions. Currently (3.4) the only benefit of these
8504 @code{#pragma}s is reduced duplication of debugging information, and
8505 that should be addressed soon on DWARF 2 targets with the use of
8509 @item #pragma interface
8510 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
8511 @kindex #pragma interface
8512 Use this directive in @emph{header files} that define object classes, to save
8513 space in most of the object files that use those classes. Normally,
8514 local copies of certain information (backup copies of inline member
8515 functions, debugging information, and the internal tables that implement
8516 virtual functions) must be kept in each object file that includes class
8517 definitions. You can use this pragma to avoid such duplication. When a
8518 header file containing @samp{#pragma interface} is included in a
8519 compilation, this auxiliary information will not be generated (unless
8520 the main input source file itself uses @samp{#pragma implementation}).
8521 Instead, the object files will contain references to be resolved at link
8524 The second form of this directive is useful for the case where you have
8525 multiple headers with the same name in different directories. If you
8526 use this form, you must specify the same string to @samp{#pragma
8529 @item #pragma implementation
8530 @itemx #pragma implementation "@var{objects}.h"
8531 @kindex #pragma implementation
8532 Use this pragma in a @emph{main input file}, when you want full output from
8533 included header files to be generated (and made globally visible). The
8534 included header file, in turn, should use @samp{#pragma interface}.
8535 Backup copies of inline member functions, debugging information, and the
8536 internal tables used to implement virtual functions are all generated in
8537 implementation files.
8539 @cindex implied @code{#pragma implementation}
8540 @cindex @code{#pragma implementation}, implied
8541 @cindex naming convention, implementation headers
8542 If you use @samp{#pragma implementation} with no argument, it applies to
8543 an include file with the same basename@footnote{A file's @dfn{basename}
8544 was the name stripped of all leading path information and of trailing
8545 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
8546 file. For example, in @file{allclass.cc}, giving just
8547 @samp{#pragma implementation}
8548 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
8550 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
8551 an implementation file whenever you would include it from
8552 @file{allclass.cc} even if you never specified @samp{#pragma
8553 implementation}. This was deemed to be more trouble than it was worth,
8554 however, and disabled.
8556 Use the string argument if you want a single implementation file to
8557 include code from multiple header files. (You must also use
8558 @samp{#include} to include the header file; @samp{#pragma
8559 implementation} only specifies how to use the file---it doesn't actually
8562 There is no way to split up the contents of a single header file into
8563 multiple implementation files.
8566 @cindex inlining and C++ pragmas
8567 @cindex C++ pragmas, effect on inlining
8568 @cindex pragmas in C++, effect on inlining
8569 @samp{#pragma implementation} and @samp{#pragma interface} also have an
8570 effect on function inlining.
8572 If you define a class in a header file marked with @samp{#pragma
8573 interface}, the effect on an inline function defined in that class is
8574 similar to an explicit @code{extern} declaration---the compiler emits
8575 no code at all to define an independent version of the function. Its
8576 definition is used only for inlining with its callers.
8578 @opindex fno-implement-inlines
8579 Conversely, when you include the same header file in a main source file
8580 that declares it as @samp{#pragma implementation}, the compiler emits
8581 code for the function itself; this defines a version of the function
8582 that can be found via pointers (or by callers compiled without
8583 inlining). If all calls to the function can be inlined, you can avoid
8584 emitting the function by compiling with @option{-fno-implement-inlines}.
8585 If any calls were not inlined, you will get linker errors.
8587 @node Template Instantiation
8588 @section Where's the Template?
8589 @cindex template instantiation
8591 C++ templates are the first language feature to require more
8592 intelligence from the environment than one usually finds on a UNIX
8593 system. Somehow the compiler and linker have to make sure that each
8594 template instance occurs exactly once in the executable if it is needed,
8595 and not at all otherwise. There are two basic approaches to this
8596 problem, which are referred to as the Borland model and the Cfront model.
8600 Borland C++ solved the template instantiation problem by adding the code
8601 equivalent of common blocks to their linker; the compiler emits template
8602 instances in each translation unit that uses them, and the linker
8603 collapses them together. The advantage of this model is that the linker
8604 only has to consider the object files themselves; there is no external
8605 complexity to worry about. This disadvantage is that compilation time
8606 is increased because the template code is being compiled repeatedly.
8607 Code written for this model tends to include definitions of all
8608 templates in the header file, since they must be seen to be
8612 The AT&T C++ translator, Cfront, solved the template instantiation
8613 problem by creating the notion of a template repository, an
8614 automatically maintained place where template instances are stored. A
8615 more modern version of the repository works as follows: As individual
8616 object files are built, the compiler places any template definitions and
8617 instantiations encountered in the repository. At link time, the link
8618 wrapper adds in the objects in the repository and compiles any needed
8619 instances that were not previously emitted. The advantages of this
8620 model are more optimal compilation speed and the ability to use the
8621 system linker; to implement the Borland model a compiler vendor also
8622 needs to replace the linker. The disadvantages are vastly increased
8623 complexity, and thus potential for error; for some code this can be
8624 just as transparent, but in practice it can been very difficult to build
8625 multiple programs in one directory and one program in multiple
8626 directories. Code written for this model tends to separate definitions
8627 of non-inline member templates into a separate file, which should be
8628 compiled separately.
8631 When used with GNU ld version 2.8 or later on an ELF system such as
8632 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
8633 Borland model. On other systems, G++ implements neither automatic
8636 A future version of G++ will support a hybrid model whereby the compiler
8637 will emit any instantiations for which the template definition is
8638 included in the compile, and store template definitions and
8639 instantiation context information into the object file for the rest.
8640 The link wrapper will extract that information as necessary and invoke
8641 the compiler to produce the remaining instantiations. The linker will
8642 then combine duplicate instantiations.
8644 In the mean time, you have the following options for dealing with
8645 template instantiations:
8650 Compile your template-using code with @option{-frepo}. The compiler will
8651 generate files with the extension @samp{.rpo} listing all of the
8652 template instantiations used in the corresponding object files which
8653 could be instantiated there; the link wrapper, @samp{collect2}, will
8654 then update the @samp{.rpo} files to tell the compiler where to place
8655 those instantiations and rebuild any affected object files. The
8656 link-time overhead is negligible after the first pass, as the compiler
8657 will continue to place the instantiations in the same files.
8659 This is your best option for application code written for the Borland
8660 model, as it will just work. Code written for the Cfront model will
8661 need to be modified so that the template definitions are available at
8662 one or more points of instantiation; usually this is as simple as adding
8663 @code{#include <tmethods.cc>} to the end of each template header.
8665 For library code, if you want the library to provide all of the template
8666 instantiations it needs, just try to link all of its object files
8667 together; the link will fail, but cause the instantiations to be
8668 generated as a side effect. Be warned, however, that this may cause
8669 conflicts if multiple libraries try to provide the same instantiations.
8670 For greater control, use explicit instantiation as described in the next
8674 @opindex fno-implicit-templates
8675 Compile your code with @option{-fno-implicit-templates} to disable the
8676 implicit generation of template instances, and explicitly instantiate
8677 all the ones you use. This approach requires more knowledge of exactly
8678 which instances you need than do the others, but it's less
8679 mysterious and allows greater control. You can scatter the explicit
8680 instantiations throughout your program, perhaps putting them in the
8681 translation units where the instances are used or the translation units
8682 that define the templates themselves; you can put all of the explicit
8683 instantiations you need into one big file; or you can create small files
8690 template class Foo<int>;
8691 template ostream& operator <<
8692 (ostream&, const Foo<int>&);
8695 for each of the instances you need, and create a template instantiation
8698 If you are using Cfront-model code, you can probably get away with not
8699 using @option{-fno-implicit-templates} when compiling files that don't
8700 @samp{#include} the member template definitions.
8702 If you use one big file to do the instantiations, you may want to
8703 compile it without @option{-fno-implicit-templates} so you get all of the
8704 instances required by your explicit instantiations (but not by any
8705 other files) without having to specify them as well.
8707 G++ has extended the template instantiation syntax given in the ISO
8708 standard to allow forward declaration of explicit instantiations
8709 (with @code{extern}), instantiation of the compiler support data for a
8710 template class (i.e.@: the vtable) without instantiating any of its
8711 members (with @code{inline}), and instantiation of only the static data
8712 members of a template class, without the support data or member
8713 functions (with (@code{static}):
8716 extern template int max (int, int);
8717 inline template class Foo<int>;
8718 static template class Foo<int>;
8722 Do nothing. Pretend G++ does implement automatic instantiation
8723 management. Code written for the Borland model will work fine, but
8724 each translation unit will contain instances of each of the templates it
8725 uses. In a large program, this can lead to an unacceptable amount of code
8729 @node Bound member functions
8730 @section Extracting the function pointer from a bound pointer to member function
8732 @cindex pointer to member function
8733 @cindex bound pointer to member function
8735 In C++, pointer to member functions (PMFs) are implemented using a wide
8736 pointer of sorts to handle all the possible call mechanisms; the PMF
8737 needs to store information about how to adjust the @samp{this} pointer,
8738 and if the function pointed to is virtual, where to find the vtable, and
8739 where in the vtable to look for the member function. If you are using
8740 PMFs in an inner loop, you should really reconsider that decision. If
8741 that is not an option, you can extract the pointer to the function that
8742 would be called for a given object/PMF pair and call it directly inside
8743 the inner loop, to save a bit of time.
8745 Note that you will still be paying the penalty for the call through a
8746 function pointer; on most modern architectures, such a call defeats the
8747 branch prediction features of the CPU@. This is also true of normal
8748 virtual function calls.
8750 The syntax for this extension is
8754 extern int (A::*fp)();
8755 typedef int (*fptr)(A *);
8757 fptr p = (fptr)(a.*fp);
8760 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
8761 no object is needed to obtain the address of the function. They can be
8762 converted to function pointers directly:
8765 fptr p1 = (fptr)(&A::foo);
8768 @opindex Wno-pmf-conversions
8769 You must specify @option{-Wno-pmf-conversions} to use this extension.
8771 @node C++ Attributes
8772 @section C++-Specific Variable, Function, and Type Attributes
8774 Some attributes only make sense for C++ programs.
8777 @item init_priority (@var{priority})
8778 @cindex init_priority attribute
8781 In Standard C++, objects defined at namespace scope are guaranteed to be
8782 initialized in an order in strict accordance with that of their definitions
8783 @emph{in a given translation unit}. No guarantee is made for initializations
8784 across translation units. However, GNU C++ allows users to control the
8785 order of initialization of objects defined at namespace scope with the
8786 @code{init_priority} attribute by specifying a relative @var{priority},
8787 a constant integral expression currently bounded between 101 and 65535
8788 inclusive. Lower numbers indicate a higher priority.
8790 In the following example, @code{A} would normally be created before
8791 @code{B}, but the @code{init_priority} attribute has reversed that order:
8794 Some_Class A __attribute__ ((init_priority (2000)));
8795 Some_Class B __attribute__ ((init_priority (543)));
8799 Note that the particular values of @var{priority} do not matter; only their
8802 @item java_interface
8803 @cindex java_interface attribute
8805 This type attribute informs C++ that the class is a Java interface. It may
8806 only be applied to classes declared within an @code{extern "Java"} block.
8807 Calls to methods declared in this interface will be dispatched using GCJ's
8808 interface table mechanism, instead of regular virtual table dispatch.
8812 See also @xref{Strong Using}.
8815 @section Strong Using
8817 @strong{Caution:} The semantics of this extension are not fully
8818 defined. Users should refrain from using this extension as its
8819 semantics may change subtly over time. It is possible that this
8820 extension wil be removed in future versions of G++.
8822 A using-directive with @code{__attribute ((strong))} is stronger
8823 than a normal using-directive in two ways:
8827 Templates from the used namespace can be specialized as though they were members of the using namespace.
8830 The using namespace is considered an associated namespace of all
8831 templates in the used namespace for purposes of argument-dependent
8835 This is useful for composing a namespace transparently from
8836 implementation namespaces. For example:
8841 template <class T> struct A @{ @};
8843 using namespace debug __attribute ((__strong__));
8844 template <> struct A<int> @{ @}; // ok to specialize
8846 template <class T> void f (A<T>);
8851 f (std::A<float>()); // lookup finds std::f
8856 @node Java Exceptions
8857 @section Java Exceptions
8859 The Java language uses a slightly different exception handling model
8860 from C++. Normally, GNU C++ will automatically detect when you are
8861 writing C++ code that uses Java exceptions, and handle them
8862 appropriately. However, if C++ code only needs to execute destructors
8863 when Java exceptions are thrown through it, GCC will guess incorrectly.
8864 Sample problematic code is:
8867 struct S @{ ~S(); @};
8868 extern void bar(); // is written in Java, and may throw exceptions
8877 The usual effect of an incorrect guess is a link failure, complaining of
8878 a missing routine called @samp{__gxx_personality_v0}.
8880 You can inform the compiler that Java exceptions are to be used in a
8881 translation unit, irrespective of what it might think, by writing
8882 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
8883 @samp{#pragma} must appear before any functions that throw or catch
8884 exceptions, or run destructors when exceptions are thrown through them.
8886 You cannot mix Java and C++ exceptions in the same translation unit. It
8887 is believed to be safe to throw a C++ exception from one file through
8888 another file compiled for the Java exception model, or vice versa, but
8889 there may be bugs in this area.
8891 @node Deprecated Features
8892 @section Deprecated Features
8894 In the past, the GNU C++ compiler was extended to experiment with new
8895 features, at a time when the C++ language was still evolving. Now that
8896 the C++ standard is complete, some of those features are superseded by
8897 superior alternatives. Using the old features might cause a warning in
8898 some cases that the feature will be dropped in the future. In other
8899 cases, the feature might be gone already.
8901 While the list below is not exhaustive, it documents some of the options
8902 that are now deprecated:
8905 @item -fexternal-templates
8906 @itemx -falt-external-templates
8907 These are two of the many ways for G++ to implement template
8908 instantiation. @xref{Template Instantiation}. The C++ standard clearly
8909 defines how template definitions have to be organized across
8910 implementation units. G++ has an implicit instantiation mechanism that
8911 should work just fine for standard-conforming code.
8913 @item -fstrict-prototype
8914 @itemx -fno-strict-prototype
8915 Previously it was possible to use an empty prototype parameter list to
8916 indicate an unspecified number of parameters (like C), rather than no
8917 parameters, as C++ demands. This feature has been removed, except where
8918 it is required for backwards compatibility @xref{Backwards Compatibility}.
8921 The named return value extension has been deprecated, and is now
8924 The use of initializer lists with new expressions has been deprecated,
8925 and is now removed from G++.
8927 Floating and complex non-type template parameters have been deprecated,
8928 and are now removed from G++.
8930 The implicit typename extension has been deprecated and is now
8933 The use of default arguments in function pointers, function typedefs and
8934 and other places where they are not permitted by the standard is
8935 deprecated and will be removed from a future version of G++.
8937 @node Backwards Compatibility
8938 @section Backwards Compatibility
8939 @cindex Backwards Compatibility
8940 @cindex ARM [Annotated C++ Reference Manual]
8942 Now that there is a definitive ISO standard C++, G++ has a specification
8943 to adhere to. The C++ language evolved over time, and features that
8944 used to be acceptable in previous drafts of the standard, such as the ARM
8945 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
8946 compilation of C++ written to such drafts, G++ contains some backwards
8947 compatibilities. @emph{All such backwards compatibility features are
8948 liable to disappear in future versions of G++.} They should be considered
8949 deprecated @xref{Deprecated Features}.
8953 If a variable is declared at for scope, it used to remain in scope until
8954 the end of the scope which contained the for statement (rather than just
8955 within the for scope). G++ retains this, but issues a warning, if such a
8956 variable is accessed outside the for scope.
8958 @item Implicit C language
8959 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
8960 scope to set the language. On such systems, all header files are
8961 implicitly scoped inside a C language scope. Also, an empty prototype
8962 @code{()} will be treated as an unspecified number of arguments, rather
8963 than no arguments, as C++ demands.