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 C Implementation-defined behavior
8 @cindex implementation-defined behavior, C language
10 A conforming implementation of ISO C is required to document its
11 choice of behavior in each of the areas that are designated
12 ``implementation defined.'' The following lists all such areas,
13 along with the section number from the ISO/IEC 9899:1999 standard.
16 * Translation implementation::
17 * Environment implementation::
18 * Identifiers implementation::
19 * Characters implementation::
20 * Integers implementation::
21 * Floating point implementation::
22 * Arrays and pointers implementation::
23 * Hints implementation::
24 * Structures unions enumerations and bit-fields implementation::
25 * Qualifiers implementation::
26 * Preprocessing directives implementation::
27 * Library functions implementation::
28 * Architecture implementation::
29 * Locale-specific behavior implementation::
32 @node Translation implementation
37 @cite{How a diagnostic is identified (3.10, 5.1.1.3).}
39 Diagnostics consist of all the output sent to stderr by GCC.
42 @cite{Whether each nonempty sequence of white-space characters other than
43 new-line is retained or replaced by one space character in translation
47 @node Environment implementation
50 The behavior of these points are dependent on the implementation
51 of the C library, and are not defined by GCC itself.
53 @node Identifiers implementation
58 @cite{Which additional multibyte characters may appear in identifiers
59 and their correspondence to universal character names (6.4.2).}
62 @cite{The number of significant initial characters in an identifier
65 For internal names, all characters are significant. For external names,
66 the number of significant characters are defined by the linker; for
67 almost all targets, all characters are significant.
71 @node Characters implementation
76 @cite{The number of bits in a byte (3.6).}
79 @cite{The values of the members of the execution character set (5.2.1).}
82 @cite{The unique value of the member of the execution character set produced
83 for each of the standard alphabetic escape sequences (5.2.2).}
86 @cite{The value of a @code{char} object into which has been stored any
87 character other than a member of the basic execution character set (6.2.5).}
90 @cite{Which of @code{signed char} or @code{unsigned char} has the same range,
91 representation, and behavior as ``plain'' @code{char} (6.2.5, 6.3.1.1).}
94 @cite{The mapping of members of the source character set (in character
95 constants and string literals) to members of the execution character
96 set (6.4.4.4, 5.1.1.2).}
99 @cite{The value of an integer character constant containing more than one
100 character or containing a character or escape sequence that does not map
101 to a single-byte execution character (6.4.4.4).}
104 @cite{The value of a wide character constant containing more than one
105 multibyte character, or containing a multibyte character or escape
106 sequence not represented in the extended execution character set (6.4.4.4).}
109 @cite{The current locale used to convert a wide character constant consisting
110 of a single multibyte character that maps to a member of the extended
111 execution character set into a corresponding wide character code (6.4.4.4).}
114 @cite{The current locale used to convert a wide string literal into
115 corresponding wide character codes (6.4.5).}
118 @cite{The value of a string literal containing a multibyte character or escape
119 sequence not represented in the execution character set (6.4.5).}
122 @node Integers implementation
127 @cite{Any extended integer types that exist in the implementation (6.2.5).}
130 @cite{Whether signed integer types are represented using sign and magnitude,
131 two's complement, or one's complement, and whether the extraordinary value
132 is a trap representation or an ordinary value (6.2.6.2).}
134 GCC supports only two's complement integer types, and all bit patterns
138 @cite{The rank of any extended integer type relative to another extended
139 integer type with the same precision (6.3.1.1).}
142 @cite{The result of, or the signal raised by, converting an integer to a
143 signed integer type when the value cannot be represented in an object of
144 that type (6.3.1.3).}
147 @cite{The results of some bitwise operations on signed integers (6.5).}
150 @node Floating point implementation
151 @section Floating point
155 @cite{The accuracy of the floating-point operations and of the library
156 functions in @code{<math.h>} and @code{<complex.h>} that return floating-point
157 results (5.2.4.2.2).}
160 @cite{The rounding behaviors characterized by non-standard values
161 of @code{FLT_ROUNDS} @gol
165 @cite{The evaluation methods characterized by non-standard negative
166 values of @code{FLT_EVAL_METHOD} (5.2.4.2.2).}
169 @cite{The direction of rounding when an integer is converted to a
170 floating-point number that cannot exactly represent the original
174 @cite{The direction of rounding when a floating-point number is
175 converted to a narrower floating-point number (6.3.1.5).}
178 @cite{How the nearest representable value or the larger or smaller
179 representable value immediately adjacent to the nearest representable
180 value is chosen for certain floating constants (6.4.4.2).}
183 @cite{Whether and how floating expressions are contracted when not
184 disallowed by the @code{FP_CONTRACT} pragma (6.5).}
187 @cite{The default state for the @code{FENV_ACCESS} pragma (7.6.1).}
190 @cite{Additional floating-point exceptions, rounding modes, environments,
191 and classifications, and their macro names (7.6, 7.12).}
194 @cite{The default state for the @code{FP_CONTRACT} pragma (7.12.2).}
197 @cite{Whether the ``inexact'' floating-point exception can be raised
198 when the rounded result actually does equal the mathematical result
199 in an IEC 60559 conformant implementation (F.9).}
202 @cite{Whether the ``underflow'' (and ``inexact'') floating-point
203 exception can be raised when a result is tiny but not inexact in an
204 IEC 60559 conformant implementation (F.9).}
208 @node Arrays and pointers implementation
209 @section Arrays and pointers
213 @cite{The result of converting a pointer to an integer or
214 vice versa (6.3.2.3).}
216 A cast from pointer to integer discards most-significant bits if the
217 pointer representation is larger than the integer type,
218 sign-extends@footnote{Future versions of GCC may zero-extend, or use
219 a target-defined @code{ptr_extend} pattern. Do not rely on sign extension.}
220 if the pointer representation is smaller than the integer type, otherwise
221 the bits are unchanged.
222 @c ??? We've always claimed that pointers were unsigned entities.
223 @c Shouldn't we therefore be doing zero-extension? If so, the bug
224 @c is in convert_to_integer, where we call type_for_size and request
225 @c a signed integral type. On the other hand, it might be most useful
226 @c for the target if we extend according to POINTERS_EXTEND_UNSIGNED.
228 A cast from integer to pointer discards most-significant bits if the
229 pointer representation is smaller than the integer type, extends according
230 to the signedness of the integer type if the pointer representation
231 is larger than the integer type, otherwise the bits are unchanged.
233 When casting from pointer to integer and back again, the resulting
234 pointer must reference the same object as the original pointer, otherwise
235 the behavior is undefined. That is, one may not use integer arithmetic to
236 avoid the undefined behavior of pointer arithmetic as proscribed in 6.5.6/8.
239 @cite{The size of the result of subtracting two pointers to elements
240 of the same array (6.5.6).}
244 @node Hints implementation
249 @cite{The extent to which suggestions made by using the @code{register}
250 storage-class specifier are effective (6.7.1).}
252 The @code{register} specifier affects code generation only in these ways:
256 When used as part of the register variable extension, see
257 @ref{Explicit Reg Vars}.
260 When @option{-O0} is in use, the compiler allocates distinct stack
261 memory for all variables that do not have the @code{register}
262 storage-class specifier; if @code{register} is specified, the variable
263 may have a shorter lifespan than the code would indicate and may never
267 On some rare x86 targets, @code{setjmp} doesn't save the registers in
268 all circumstances. In those cases, GCC doesn't allocate any variables
269 in registers unless they are marked @code{register}.
274 @cite{The extent to which suggestions made by using the inline function
275 specifier are effective (6.7.4).}
277 GCC will not inline any functions if the @option{-fno-inline} option is
278 used or if @option{-O0} is used. Otherwise, GCC may still be unable to
279 inline a function for many reasons; the @option{-Winline} option may be
280 used to determine if a function has not been inlined and why not.
284 @node Structures unions enumerations and bit-fields implementation
285 @section Structures, unions, enumerations, and bit-fields
289 @cite{Whether a ``plain'' int bit-field is treated as a @code{signed int}
290 bit-field or as an @code{unsigned int} bit-field (6.7.2, 6.7.2.1).}
293 @cite{Allowable bit-field types other than @code{_Bool}, @code{signed int},
294 and @code{unsigned int} (6.7.2.1).}
297 @cite{Whether a bit-field can straddle a storage-unit boundary (6.7.2.1).}
300 @cite{The order of allocation of bit-fields within a unit (6.7.2.1).}
303 @cite{The alignment of non-bit-field members of structures (6.7.2.1).}
306 @cite{The integer type compatible with each enumerated type (6.7.2.2).}
310 @node Qualifiers implementation
315 @cite{What constitutes an access to an object that has volatile-qualified
320 @node Preprocessing directives implementation
321 @section Preprocessing directives
325 @cite{How sequences in both forms of header names are mapped to headers
326 or external source file names (6.4.7).}
329 @cite{Whether the value of a character constant in a constant expression
330 that controls conditional inclusion matches the value of the same character
331 constant in the execution character set (6.10.1).}
334 @cite{Whether the value of a single-character character constant in a
335 constant expression that controls conditional inclusion may have a
336 negative value (6.10.1).}
339 @cite{The places that are searched for an included @samp{<>} delimited
340 header, and how the places are specified or the header is
341 identified (6.10.2).}
344 @cite{How the named source file is searched for in an included @samp{""}
345 delimited header (6.10.2).}
348 @cite{The method by which preprocessing tokens (possibly resulting from
349 macro expansion) in a @code{#include} directive are combined into a header
353 @cite{The nesting limit for @code{#include} processing (6.10.2).}
355 GCC imposes a limit of 200 nested @code{#include}s.
358 @cite{Whether the @samp{#} operator inserts a @samp{\} character before
359 the @samp{\} character that begins a universal character name in a
360 character constant or string literal (6.10.3.2).}
363 @cite{The behavior on each recognized non-@code{STDC #pragma}
367 @cite{The definitions for @code{__DATE__} and @code{__TIME__} when
368 respectively, the date and time of translation are not available (6.10.8).}
370 If the date and time are not available, @code{__DATE__} expands to
371 @code{@w{"??? ?? ????"}} and @code{__TIME__} expands to
376 @node Library functions implementation
377 @section Library functions
379 The behavior of these points are dependent on the implementation
380 of the C library, and are not defined by GCC itself.
382 @node Architecture implementation
383 @section Architecture
387 @cite{The values or expressions assigned to the macros specified in the
388 headers @code{<float.h>}, @code{<limits.h>}, and @code{<stdint.h>}
389 (5.2.4.2, 7.18.2, 7.18.3).}
392 @cite{The number, order, and encoding of bytes in any object
393 (when not explicitly specified in this International Standard) (6.2.6.1).}
396 @cite{The value of the result of the sizeof operator (6.5.3.4).}
400 @node Locale-specific behavior implementation
401 @section Locale-specific behavior
403 The behavior of these points are dependent on the implementation
404 of the C library, and are not defined by GCC itself.
407 @chapter Extensions to the C Language Family
408 @cindex extensions, C language
409 @cindex C language extensions
412 GNU C provides several language features not found in ISO standard C@.
413 (The @option{-pedantic} option directs GCC to print a warning message if
414 any of these features is used.) To test for the availability of these
415 features in conditional compilation, check for a predefined macro
416 @code{__GNUC__}, which is always defined under GCC@.
418 These extensions are available in C and Objective-C@. Most of them are
419 also available in C++. @xref{C++ Extensions,,Extensions to the
420 C++ Language}, for extensions that apply @emph{only} to C++.
422 Some features that are in ISO C99 but not C89 or C++ are also, as
423 extensions, accepted by GCC in C89 mode and in C++.
426 * Statement Exprs:: Putting statements and declarations inside expressions.
427 * Local Labels:: Labels local to a block.
428 * Labels as Values:: Getting pointers to labels, and computed gotos.
429 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
430 * Constructing Calls:: Dispatching a call to another function.
431 * Typeof:: @code{typeof}: referring to the type of an expression.
432 * Lvalues:: Using @samp{?:}, @samp{,} and casts in lvalues.
433 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
434 * Long Long:: Double-word integers---@code{long long int}.
435 * Complex:: Data types for complex numbers.
436 * Hex Floats:: Hexadecimal floating-point constants.
437 * Zero Length:: Zero-length arrays.
438 * Variable Length:: Arrays whose length is computed at run time.
439 * Empty Structures:: Structures with no members.
440 * Variadic Macros:: Macros with a variable number of arguments.
441 * Escaped Newlines:: Slightly looser rules for escaped newlines.
442 * Subscripting:: Any array can be subscripted, even if not an lvalue.
443 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
444 * Initializers:: Non-constant initializers.
445 * Compound Literals:: Compound literals give structures, unions
447 * Designated Inits:: Labeling elements of initializers.
448 * Cast to Union:: Casting to union type from any member of the union.
449 * Case Ranges:: `case 1 ... 9' and such.
450 * Mixed Declarations:: Mixing declarations and code.
451 * Function Attributes:: Declaring that functions have no side effects,
452 or that they can never return.
453 * Attribute Syntax:: Formal syntax for attributes.
454 * Function Prototypes:: Prototype declarations and old-style definitions.
455 * C++ Comments:: C++ comments are recognized.
456 * Dollar Signs:: Dollar sign is allowed in identifiers.
457 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
458 * Variable Attributes:: Specifying attributes of variables.
459 * Type Attributes:: Specifying attributes of types.
460 * Alignment:: Inquiring about the alignment of a type or variable.
461 * Inline:: Defining inline functions (as fast as macros).
462 * Extended Asm:: Assembler instructions with C expressions as operands.
463 (With them you can define ``built-in'' functions.)
464 * Constraints:: Constraints for asm operands
465 * Asm Labels:: Specifying the assembler name to use for a C symbol.
466 * Explicit Reg Vars:: Defining variables residing in specified registers.
467 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
468 * Incomplete Enums:: @code{enum foo;}, with details to follow.
469 * Function Names:: Printable strings which are the name of the current
471 * Return Address:: Getting the return or frame address of a function.
472 * Vector Extensions:: Using vector instructions through built-in functions.
473 * Other Builtins:: Other built-in functions.
474 * Target Builtins:: Built-in functions specific to particular targets.
475 * Pragmas:: Pragmas accepted by GCC.
476 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
477 * Thread-Local:: Per-thread variables.
480 @node Statement Exprs
481 @section Statements and Declarations in Expressions
482 @cindex statements inside expressions
483 @cindex declarations inside expressions
484 @cindex expressions containing statements
485 @cindex macros, statements in expressions
487 @c the above section title wrapped and causes an underfull hbox.. i
488 @c changed it from "within" to "in". --mew 4feb93
489 A compound statement enclosed in parentheses may appear as an expression
490 in GNU C@. This allows you to use loops, switches, and local variables
491 within an expression.
493 Recall that a compound statement is a sequence of statements surrounded
494 by braces; in this construct, parentheses go around the braces. For
498 (@{ int y = foo (); int z;
505 is a valid (though slightly more complex than necessary) expression
506 for the absolute value of @code{foo ()}.
508 The last thing in the compound statement should be an expression
509 followed by a semicolon; the value of this subexpression serves as the
510 value of the entire construct. (If you use some other kind of statement
511 last within the braces, the construct has type @code{void}, and thus
512 effectively no value.)
514 This feature is especially useful in making macro definitions ``safe'' (so
515 that they evaluate each operand exactly once). For example, the
516 ``maximum'' function is commonly defined as a macro in standard C as
520 #define max(a,b) ((a) > (b) ? (a) : (b))
524 @cindex side effects, macro argument
525 But this definition computes either @var{a} or @var{b} twice, with bad
526 results if the operand has side effects. In GNU C, if you know the
527 type of the operands (here taken as @code{int}), you can define
528 the macro safely as follows:
531 #define maxint(a,b) \
532 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
535 Embedded statements are not allowed in constant expressions, such as
536 the value of an enumeration constant, the width of a bit-field, or
537 the initial value of a static variable.
539 If you don't know the type of the operand, you can still do this, but you
540 must use @code{typeof} (@pxref{Typeof}).
542 In G++, the result value of a statement expression undergoes array and
543 function pointer decay, and is returned by value to the enclosing
544 expression. For instance, if @code{A} is a class, then
553 will construct a temporary @code{A} object to hold the result of the
554 statement expression, and that will be used to invoke @code{Foo}.
555 Therefore the @code{this} pointer observed by @code{Foo} will not be the
558 Any temporaries created within a statement within a statement expression
559 will be destroyed at the statement's end. This makes statement
560 expressions inside macros slightly different from function calls. In
561 the latter case temporaries introduced during argument evaluation will
562 be destroyed at the end of the statement that includes the function
563 call. In the statement expression case they will be destroyed during
564 the statement expression. For instance,
567 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
568 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
578 will have different places where temporaries are destroyed. For the
579 @code{macro} case, the temporary @code{X} will be destroyed just after
580 the initialization of @code{b}. In the @code{function} case that
581 temporary will be destroyed when the function returns.
583 These considerations mean that it is probably a bad idea to use
584 statement-expressions of this form in header files that are designed to
585 work with C++. (Note that some versions of the GNU C Library contained
586 header files using statement-expression that lead to precisely this
590 @section Locally Declared Labels
592 @cindex macros, local labels
594 GCC allows you to declare @dfn{local labels} in any nested block
595 scope. A local label is just like an ordinary label, but you can
596 only reference it (with a @code{goto} statement, or by taking its
597 address) within the block in which it was declared.
599 A local label declaration looks like this:
602 __label__ @var{label};
609 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
612 Local label declarations must come at the beginning of the block,
613 before any ordinary declarations or statements.
615 The label declaration defines the label @emph{name}, but does not define
616 the label itself. You must do this in the usual way, with
617 @code{@var{label}:}, within the statements of the statement expression.
619 The local label feature is useful for complex macros. If a macro
620 contains nested loops, a @code{goto} can be useful for breaking out of
621 them. However, an ordinary label whose scope is the whole function
622 cannot be used: if the macro can be expanded several times in one
623 function, the label will be multiply defined in that function. A
624 local label avoids this problem. For example:
627 #define SEARCH(value, array, target) \
630 typeof (target) _SEARCH_target = (target); \
631 typeof (*(array)) *_SEARCH_array = (array); \
634 for (i = 0; i < max; i++) \
635 for (j = 0; j < max; j++) \
636 if (_SEARCH_array[i][j] == _SEARCH_target) \
637 @{ (value) = i; goto found; @} \
643 This could also be written using a statement-expression:
646 #define SEARCH(array, target) \
649 typeof (target) _SEARCH_target = (target); \
650 typeof (*(array)) *_SEARCH_array = (array); \
653 for (i = 0; i < max; i++) \
654 for (j = 0; j < max; j++) \
655 if (_SEARCH_array[i][j] == _SEARCH_target) \
656 @{ value = i; goto found; @} \
663 Local label declarations also make the labels they declare visible to
664 nested functions, if there are any. @xref{Nested Functions}, for details.
666 @node Labels as Values
667 @section Labels as Values
668 @cindex labels as values
669 @cindex computed gotos
670 @cindex goto with computed label
671 @cindex address of a label
673 You can get the address of a label defined in the current function
674 (or a containing function) with the unary operator @samp{&&}. The
675 value has type @code{void *}. This value is a constant and can be used
676 wherever a constant of that type is valid. For example:
684 To use these values, you need to be able to jump to one. This is done
685 with the computed goto statement@footnote{The analogous feature in
686 Fortran is called an assigned goto, but that name seems inappropriate in
687 C, where one can do more than simply store label addresses in label
688 variables.}, @code{goto *@var{exp};}. For example,
695 Any expression of type @code{void *} is allowed.
697 One way of using these constants is in initializing a static array that
698 will serve as a jump table:
701 static void *array[] = @{ &&foo, &&bar, &&hack @};
704 Then you can select a label with indexing, like this:
711 Note that this does not check whether the subscript is in bounds---array
712 indexing in C never does that.
714 Such an array of label values serves a purpose much like that of the
715 @code{switch} statement. The @code{switch} statement is cleaner, so
716 use that rather than an array unless the problem does not fit a
717 @code{switch} statement very well.
719 Another use of label values is in an interpreter for threaded code.
720 The labels within the interpreter function can be stored in the
721 threaded code for super-fast dispatching.
723 You may not use this mechanism to jump to code in a different function.
724 If you do that, totally unpredictable things will happen. The best way to
725 avoid this is to store the label address only in automatic variables and
726 never pass it as an argument.
728 An alternate way to write the above example is
731 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
733 goto *(&&foo + array[i]);
737 This is more friendly to code living in shared libraries, as it reduces
738 the number of dynamic relocations that are needed, and by consequence,
739 allows the data to be read-only.
741 @node Nested Functions
742 @section Nested Functions
743 @cindex nested functions
744 @cindex downward funargs
747 A @dfn{nested function} is a function defined inside another function.
748 (Nested functions are not supported for GNU C++.) The nested function's
749 name is local to the block where it is defined. For example, here we
750 define a nested function named @code{square}, and call it twice:
754 foo (double a, double b)
756 double square (double z) @{ return z * z; @}
758 return square (a) + square (b);
763 The nested function can access all the variables of the containing
764 function that are visible at the point of its definition. This is
765 called @dfn{lexical scoping}. For example, here we show a nested
766 function which uses an inherited variable named @code{offset}:
770 bar (int *array, int offset, int size)
772 int access (int *array, int index)
773 @{ return array[index + offset]; @}
776 for (i = 0; i < size; i++)
777 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
782 Nested function definitions are permitted within functions in the places
783 where variable definitions are allowed; that is, in any block, before
784 the first statement in the block.
786 It is possible to call the nested function from outside the scope of its
787 name by storing its address or passing the address to another function:
790 hack (int *array, int size)
792 void store (int index, int value)
793 @{ array[index] = value; @}
795 intermediate (store, size);
799 Here, the function @code{intermediate} receives the address of
800 @code{store} as an argument. If @code{intermediate} calls @code{store},
801 the arguments given to @code{store} are used to store into @code{array}.
802 But this technique works only so long as the containing function
803 (@code{hack}, in this example) does not exit.
805 If you try to call the nested function through its address after the
806 containing function has exited, all hell will break loose. If you try
807 to call it after a containing scope level has exited, and if it refers
808 to some of the variables that are no longer in scope, you may be lucky,
809 but it's not wise to take the risk. If, however, the nested function
810 does not refer to anything that has gone out of scope, you should be
813 GCC implements taking the address of a nested function using a technique
814 called @dfn{trampolines}. A paper describing them is available as
817 @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}.
819 A nested function can jump to a label inherited from a containing
820 function, provided the label was explicitly declared in the containing
821 function (@pxref{Local Labels}). Such a jump returns instantly to the
822 containing function, exiting the nested function which did the
823 @code{goto} and any intermediate functions as well. Here is an example:
827 bar (int *array, int offset, int size)
830 int access (int *array, int index)
834 return array[index + offset];
838 for (i = 0; i < size; i++)
839 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
843 /* @r{Control comes here from @code{access}
844 if it detects an error.} */
851 A nested function always has internal linkage. Declaring one with
852 @code{extern} is erroneous. If you need to declare the nested function
853 before its definition, use @code{auto} (which is otherwise meaningless
854 for function declarations).
857 bar (int *array, int offset, int size)
860 auto int access (int *, int);
862 int access (int *array, int index)
866 return array[index + offset];
872 @node Constructing Calls
873 @section Constructing Function Calls
874 @cindex constructing calls
875 @cindex forwarding calls
877 Using the built-in functions described below, you can record
878 the arguments a function received, and call another function
879 with the same arguments, without knowing the number or types
882 You can also record the return value of that function call,
883 and later return that value, without knowing what data type
884 the function tried to return (as long as your caller expects
887 However, these built-in functions may interact badly with some
888 sophisticated features or other extensions of the language. It
889 is, therefore, not recommended to use them outside very simple
890 functions acting as mere forwarders for their arguments.
892 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
893 This built-in function returns a pointer to data
894 describing how to perform a call with the same arguments as were passed
895 to the current function.
897 The function saves the arg pointer register, structure value address,
898 and all registers that might be used to pass arguments to a function
899 into a block of memory allocated on the stack. Then it returns the
900 address of that block.
903 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
904 This built-in function invokes @var{function}
905 with a copy of the parameters described by @var{arguments}
908 The value of @var{arguments} should be the value returned by
909 @code{__builtin_apply_args}. The argument @var{size} specifies the size
910 of the stack argument data, in bytes.
912 This function returns a pointer to data describing
913 how to return whatever value was returned by @var{function}. The data
914 is saved in a block of memory allocated on the stack.
916 It is not always simple to compute the proper value for @var{size}. The
917 value is used by @code{__builtin_apply} to compute the amount of data
918 that should be pushed on the stack and copied from the incoming argument
922 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
923 This built-in function returns the value described by @var{result} from
924 the containing function. You should specify, for @var{result}, a value
925 returned by @code{__builtin_apply}.
929 @section Referring to a Type with @code{typeof}
932 @cindex macros, types of arguments
934 Another way to refer to the type of an expression is with @code{typeof}.
935 The syntax of using of this keyword looks like @code{sizeof}, but the
936 construct acts semantically like a type name defined with @code{typedef}.
938 There are two ways of writing the argument to @code{typeof}: with an
939 expression or with a type. Here is an example with an expression:
946 This assumes that @code{x} is an array of pointers to functions;
947 the type described is that of the values of the functions.
949 Here is an example with a typename as the argument:
956 Here the type described is that of pointers to @code{int}.
958 If you are writing a header file that must work when included in ISO C
959 programs, write @code{__typeof__} instead of @code{typeof}.
960 @xref{Alternate Keywords}.
962 A @code{typeof}-construct can be used anywhere a typedef name could be
963 used. For example, you can use it in a declaration, in a cast, or inside
964 of @code{sizeof} or @code{typeof}.
966 @code{typeof} is often useful in conjunction with the
967 statements-within-expressions feature. Here is how the two together can
968 be used to define a safe ``maximum'' macro that operates on any
969 arithmetic type and evaluates each of its arguments exactly once:
973 (@{ typeof (a) _a = (a); \
974 typeof (b) _b = (b); \
975 _a > _b ? _a : _b; @})
978 @cindex underscores in variables in macros
979 @cindex @samp{_} in variables in macros
980 @cindex local variables in macros
981 @cindex variables, local, in macros
982 @cindex macros, local variables in
984 The reason for using names that start with underscores for the local
985 variables is to avoid conflicts with variable names that occur within the
986 expressions that are substituted for @code{a} and @code{b}. Eventually we
987 hope to design a new form of declaration syntax that allows you to declare
988 variables whose scopes start only after their initializers; this will be a
989 more reliable way to prevent such conflicts.
992 Some more examples of the use of @code{typeof}:
996 This declares @code{y} with the type of what @code{x} points to.
1003 This declares @code{y} as an array of such values.
1010 This declares @code{y} as an array of pointers to characters:
1013 typeof (typeof (char *)[4]) y;
1017 It is equivalent to the following traditional C declaration:
1023 To see the meaning of the declaration using @code{typeof}, and why it
1024 might be a useful way to write, rewrite it with these macros:
1027 #define pointer(T) typeof(T *)
1028 #define array(T, N) typeof(T [N])
1032 Now the declaration can be rewritten this way:
1035 array (pointer (char), 4) y;
1039 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
1040 pointers to @code{char}.
1043 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
1044 a more limited extension which permitted one to write
1047 typedef @var{T} = @var{expr};
1051 with the effect of declaring @var{T} to have the type of the expression
1052 @var{expr}. This extension does not work with GCC 3 (versions between
1053 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
1054 relies on it should be rewritten to use @code{typeof}:
1057 typedef typeof(@var{expr}) @var{T};
1061 This will work with all versions of GCC@.
1064 @section Generalized Lvalues
1065 @cindex compound expressions as lvalues
1066 @cindex expressions, compound, as lvalues
1067 @cindex conditional expressions as lvalues
1068 @cindex expressions, conditional, as lvalues
1069 @cindex casts as lvalues
1070 @cindex generalized lvalues
1071 @cindex lvalues, generalized
1072 @cindex extensions, @code{?:}
1073 @cindex @code{?:} extensions
1075 Compound expressions, conditional expressions and casts are allowed as
1076 lvalues provided their operands are lvalues. This means that you can take
1077 their addresses or store values into them. All these extensions are
1080 Standard C++ allows compound expressions and conditional expressions
1081 as lvalues, and permits casts to reference type, so use of this
1082 extension is not supported for C++ code.
1084 For example, a compound expression can be assigned, provided the last
1085 expression in the sequence is an lvalue. These two expressions are
1093 Similarly, the address of the compound expression can be taken. These two
1094 expressions are equivalent:
1101 A conditional expression is a valid lvalue if its type is not void and the
1102 true and false branches are both valid lvalues. For example, these two
1103 expressions are equivalent:
1107 (a ? b = 5 : (c = 5))
1110 A cast is a valid lvalue if its operand is an lvalue. This extension
1111 is deprecated. A simple
1112 assignment whose left-hand side is a cast works by converting the
1113 right-hand side first to the specified type, then to the type of the
1114 inner left-hand side expression. After this is stored, the value is
1115 converted back to the specified type to become the value of the
1116 assignment. Thus, if @code{a} has type @code{char *}, the following two
1117 expressions are equivalent:
1121 (int)(a = (char *)(int)5)
1124 An assignment-with-arithmetic operation such as @samp{+=} applied to a cast
1125 performs the arithmetic using the type resulting from the cast, and then
1126 continues as in the previous case. Therefore, these two expressions are
1131 (int)(a = (char *)(int) ((int)a + 5))
1134 You cannot take the address of an lvalue cast, because the use of its
1135 address would not work out coherently. Suppose that @code{&(int)f} were
1136 permitted, where @code{f} has type @code{float}. Then the following
1137 statement would try to store an integer bit-pattern where a floating
1138 point number belongs:
1144 This is quite different from what @code{(int)f = 1} would do---that
1145 would convert 1 to floating point and store it. Rather than cause this
1146 inconsistency, we think it is better to prohibit use of @samp{&} on a cast.
1148 If you really do want an @code{int *} pointer with the address of
1149 @code{f}, you can simply write @code{(int *)&f}.
1152 @section Conditionals with Omitted Operands
1153 @cindex conditional expressions, extensions
1154 @cindex omitted middle-operands
1155 @cindex middle-operands, omitted
1156 @cindex extensions, @code{?:}
1157 @cindex @code{?:} extensions
1159 The middle operand in a conditional expression may be omitted. Then
1160 if the first operand is nonzero, its value is the value of the conditional
1163 Therefore, the expression
1170 has the value of @code{x} if that is nonzero; otherwise, the value of
1173 This example is perfectly equivalent to
1179 @cindex side effect in ?:
1180 @cindex ?: side effect
1182 In this simple case, the ability to omit the middle operand is not
1183 especially useful. When it becomes useful is when the first operand does,
1184 or may (if it is a macro argument), contain a side effect. Then repeating
1185 the operand in the middle would perform the side effect twice. Omitting
1186 the middle operand uses the value already computed without the undesirable
1187 effects of recomputing it.
1190 @section Double-Word Integers
1191 @cindex @code{long long} data types
1192 @cindex double-word arithmetic
1193 @cindex multiprecision arithmetic
1194 @cindex @code{LL} integer suffix
1195 @cindex @code{ULL} integer suffix
1197 ISO C99 supports data types for integers that are at least 64 bits wide,
1198 and as an extension GCC supports them in C89 mode and in C++.
1199 Simply write @code{long long int} for a signed integer, or
1200 @code{unsigned long long int} for an unsigned integer. To make an
1201 integer constant of type @code{long long int}, add the suffix @samp{LL}
1202 to the integer. To make an integer constant of type @code{unsigned long
1203 long int}, add the suffix @samp{ULL} to the integer.
1205 You can use these types in arithmetic like any other integer types.
1206 Addition, subtraction, and bitwise boolean operations on these types
1207 are open-coded on all types of machines. Multiplication is open-coded
1208 if the machine supports fullword-to-doubleword a widening multiply
1209 instruction. Division and shifts are open-coded only on machines that
1210 provide special support. The operations that are not open-coded use
1211 special library routines that come with GCC@.
1213 There may be pitfalls when you use @code{long long} types for function
1214 arguments, unless you declare function prototypes. If a function
1215 expects type @code{int} for its argument, and you pass a value of type
1216 @code{long long int}, confusion will result because the caller and the
1217 subroutine will disagree about the number of bytes for the argument.
1218 Likewise, if the function expects @code{long long int} and you pass
1219 @code{int}. The best way to avoid such problems is to use prototypes.
1222 @section Complex Numbers
1223 @cindex complex numbers
1224 @cindex @code{_Complex} keyword
1225 @cindex @code{__complex__} keyword
1227 ISO C99 supports complex floating data types, and as an extension GCC
1228 supports them in C89 mode and in C++, and supports complex integer data
1229 types which are not part of ISO C99. You can declare complex types
1230 using the keyword @code{_Complex}. As an extension, the older GNU
1231 keyword @code{__complex__} is also supported.
1233 For example, @samp{_Complex double x;} declares @code{x} as a
1234 variable whose real part and imaginary part are both of type
1235 @code{double}. @samp{_Complex short int y;} declares @code{y} to
1236 have real and imaginary parts of type @code{short int}; this is not
1237 likely to be useful, but it shows that the set of complex types is
1240 To write a constant with a complex data type, use the suffix @samp{i} or
1241 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
1242 has type @code{_Complex float} and @code{3i} has type
1243 @code{_Complex int}. Such a constant always has a pure imaginary
1244 value, but you can form any complex value you like by adding one to a
1245 real constant. This is a GNU extension; if you have an ISO C99
1246 conforming C library (such as GNU libc), and want to construct complex
1247 constants of floating type, you should include @code{<complex.h>} and
1248 use the macros @code{I} or @code{_Complex_I} instead.
1250 @cindex @code{__real__} keyword
1251 @cindex @code{__imag__} keyword
1252 To extract the real part of a complex-valued expression @var{exp}, write
1253 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
1254 extract the imaginary part. This is a GNU extension; for values of
1255 floating type, you should use the ISO C99 functions @code{crealf},
1256 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
1257 @code{cimagl}, declared in @code{<complex.h>} and also provided as
1258 built-in functions by GCC@.
1260 @cindex complex conjugation
1261 The operator @samp{~} performs complex conjugation when used on a value
1262 with a complex type. This is a GNU extension; for values of
1263 floating type, you should use the ISO C99 functions @code{conjf},
1264 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
1265 provided as built-in functions by GCC@.
1267 GCC can allocate complex automatic variables in a noncontiguous
1268 fashion; it's even possible for the real part to be in a register while
1269 the imaginary part is on the stack (or vice-versa). Only the DWARF2
1270 debug info format can represent this, so use of DWARF2 is recommended.
1271 If you are using the stabs debug info format, GCC describes a noncontiguous
1272 complex variable as if it were two separate variables of noncomplex type.
1273 If the variable's actual name is @code{foo}, the two fictitious
1274 variables are named @code{foo$real} and @code{foo$imag}. You can
1275 examine and set these two fictitious variables with your debugger.
1281 ISO C99 supports floating-point numbers written not only in the usual
1282 decimal notation, such as @code{1.55e1}, but also numbers such as
1283 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
1284 supports this in C89 mode (except in some cases when strictly
1285 conforming) and in C++. In that format the
1286 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
1287 mandatory. The exponent is a decimal number that indicates the power of
1288 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
1295 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
1296 is the same as @code{1.55e1}.
1298 Unlike for floating-point numbers in the decimal notation the exponent
1299 is always required in the hexadecimal notation. Otherwise the compiler
1300 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
1301 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
1302 extension for floating-point constants of type @code{float}.
1305 @section Arrays of Length Zero
1306 @cindex arrays of length zero
1307 @cindex zero-length arrays
1308 @cindex length-zero arrays
1309 @cindex flexible array members
1311 Zero-length arrays are allowed in GNU C@. They are very useful as the
1312 last element of a structure which is really a header for a variable-length
1321 struct line *thisline = (struct line *)
1322 malloc (sizeof (struct line) + this_length);
1323 thisline->length = this_length;
1326 In ISO C90, you would have to give @code{contents} a length of 1, which
1327 means either you waste space or complicate the argument to @code{malloc}.
1329 In ISO C99, you would use a @dfn{flexible array member}, which is
1330 slightly different in syntax and semantics:
1334 Flexible array members are written as @code{contents[]} without
1338 Flexible array members have incomplete type, and so the @code{sizeof}
1339 operator may not be applied. As a quirk of the original implementation
1340 of zero-length arrays, @code{sizeof} evaluates to zero.
1343 Flexible array members may only appear as the last member of a
1344 @code{struct} that is otherwise non-empty.
1347 A structure containing a flexible array member, or a union containing
1348 such a structure (possibly recursively), may not be a member of a
1349 structure or an element of an array. (However, these uses are
1350 permitted by GCC as extensions.)
1353 GCC versions before 3.0 allowed zero-length arrays to be statically
1354 initialized, as if they were flexible arrays. In addition to those
1355 cases that were useful, it also allowed initializations in situations
1356 that would corrupt later data. Non-empty initialization of zero-length
1357 arrays is now treated like any case where there are more initializer
1358 elements than the array holds, in that a suitable warning about "excess
1359 elements in array" is given, and the excess elements (all of them, in
1360 this case) are ignored.
1362 Instead GCC allows static initialization of flexible array members.
1363 This is equivalent to defining a new structure containing the original
1364 structure followed by an array of sufficient size to contain the data.
1365 I.e.@: in the following, @code{f1} is constructed as if it were declared
1371 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
1374 struct f1 f1; int data[3];
1375 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1379 The convenience of this extension is that @code{f1} has the desired
1380 type, eliminating the need to consistently refer to @code{f2.f1}.
1382 This has symmetry with normal static arrays, in that an array of
1383 unknown size is also written with @code{[]}.
1385 Of course, this extension only makes sense if the extra data comes at
1386 the end of a top-level object, as otherwise we would be overwriting
1387 data at subsequent offsets. To avoid undue complication and confusion
1388 with initialization of deeply nested arrays, we simply disallow any
1389 non-empty initialization except when the structure is the top-level
1390 object. For example:
1393 struct foo @{ int x; int y[]; @};
1394 struct bar @{ struct foo z; @};
1396 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1397 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1398 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1399 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1402 @node Empty Structures
1403 @section Structures With No Members
1404 @cindex empty structures
1405 @cindex zero-size structures
1407 GCC permits a C structure to have no members:
1414 The structure will have size zero. In C++, empty structures are part
1415 of the language. G++ treats empty structures as if they had a single
1416 member of type @code{char}.
1418 @node Variable Length
1419 @section Arrays of Variable Length
1420 @cindex variable-length arrays
1421 @cindex arrays of variable length
1424 Variable-length automatic arrays are allowed in ISO C99, and as an
1425 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1426 implementation of variable-length arrays does not yet conform in detail
1427 to the ISO C99 standard.) These arrays are
1428 declared like any other automatic arrays, but with a length that is not
1429 a constant expression. The storage is allocated at the point of
1430 declaration and deallocated when the brace-level is exited. For
1435 concat_fopen (char *s1, char *s2, char *mode)
1437 char str[strlen (s1) + strlen (s2) + 1];
1440 return fopen (str, mode);
1444 @cindex scope of a variable length array
1445 @cindex variable-length array scope
1446 @cindex deallocating variable length arrays
1447 Jumping or breaking out of the scope of the array name deallocates the
1448 storage. Jumping into the scope is not allowed; you get an error
1451 @cindex @code{alloca} vs variable-length arrays
1452 You can use the function @code{alloca} to get an effect much like
1453 variable-length arrays. The function @code{alloca} is available in
1454 many other C implementations (but not in all). On the other hand,
1455 variable-length arrays are more elegant.
1457 There are other differences between these two methods. Space allocated
1458 with @code{alloca} exists until the containing @emph{function} returns.
1459 The space for a variable-length array is deallocated as soon as the array
1460 name's scope ends. (If you use both variable-length arrays and
1461 @code{alloca} in the same function, deallocation of a variable-length array
1462 will also deallocate anything more recently allocated with @code{alloca}.)
1464 You can also use variable-length arrays as arguments to functions:
1468 tester (int len, char data[len][len])
1474 The length of an array is computed once when the storage is allocated
1475 and is remembered for the scope of the array in case you access it with
1478 If you want to pass the array first and the length afterward, you can
1479 use a forward declaration in the parameter list---another GNU extension.
1483 tester (int len; char data[len][len], int len)
1489 @cindex parameter forward declaration
1490 The @samp{int len} before the semicolon is a @dfn{parameter forward
1491 declaration}, and it serves the purpose of making the name @code{len}
1492 known when the declaration of @code{data} is parsed.
1494 You can write any number of such parameter forward declarations in the
1495 parameter list. They can be separated by commas or semicolons, but the
1496 last one must end with a semicolon, which is followed by the ``real''
1497 parameter declarations. Each forward declaration must match a ``real''
1498 declaration in parameter name and data type. ISO C99 does not support
1499 parameter forward declarations.
1501 @node Variadic Macros
1502 @section Macros with a Variable Number of Arguments.
1503 @cindex variable number of arguments
1504 @cindex macro with variable arguments
1505 @cindex rest argument (in macro)
1506 @cindex variadic macros
1508 In the ISO C standard of 1999, a macro can be declared to accept a
1509 variable number of arguments much as a function can. The syntax for
1510 defining the macro is similar to that of a function. Here is an
1514 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1517 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1518 such a macro, it represents the zero or more tokens until the closing
1519 parenthesis that ends the invocation, including any commas. This set of
1520 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1521 wherever it appears. See the CPP manual for more information.
1523 GCC has long supported variadic macros, and used a different syntax that
1524 allowed you to give a name to the variable arguments just like any other
1525 argument. Here is an example:
1528 #define debug(format, args...) fprintf (stderr, format, args)
1531 This is in all ways equivalent to the ISO C example above, but arguably
1532 more readable and descriptive.
1534 GNU CPP has two further variadic macro extensions, and permits them to
1535 be used with either of the above forms of macro definition.
1537 In standard C, you are not allowed to leave the variable argument out
1538 entirely; but you are allowed to pass an empty argument. For example,
1539 this invocation is invalid in ISO C, because there is no comma after
1546 GNU CPP permits you to completely omit the variable arguments in this
1547 way. In the above examples, the compiler would complain, though since
1548 the expansion of the macro still has the extra comma after the format
1551 To help solve this problem, CPP behaves specially for variable arguments
1552 used with the token paste operator, @samp{##}. If instead you write
1555 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1558 and if the variable arguments are omitted or empty, the @samp{##}
1559 operator causes the preprocessor to remove the comma before it. If you
1560 do provide some variable arguments in your macro invocation, GNU CPP
1561 does not complain about the paste operation and instead places the
1562 variable arguments after the comma. Just like any other pasted macro
1563 argument, these arguments are not macro expanded.
1565 @node Escaped Newlines
1566 @section Slightly Looser Rules for Escaped Newlines
1567 @cindex escaped newlines
1568 @cindex newlines (escaped)
1570 Recently, the preprocessor has relaxed its treatment of escaped
1571 newlines. Previously, the newline had to immediately follow a
1572 backslash. The current implementation allows whitespace in the form
1573 of spaces, horizontal and vertical tabs, and form feeds between the
1574 backslash and the subsequent newline. The preprocessor issues a
1575 warning, but treats it as a valid escaped newline and combines the two
1576 lines to form a single logical line. This works within comments and
1577 tokens, as well as between tokens. Comments are @emph{not} treated as
1578 whitespace for the purposes of this relaxation, since they have not
1579 yet been replaced with spaces.
1582 @section Non-Lvalue Arrays May Have Subscripts
1583 @cindex subscripting
1584 @cindex arrays, non-lvalue
1586 @cindex subscripting and function values
1587 In ISO C99, arrays that are not lvalues still decay to pointers, and
1588 may be subscripted, although they may not be modified or used after
1589 the next sequence point and the unary @samp{&} operator may not be
1590 applied to them. As an extension, GCC allows such arrays to be
1591 subscripted in C89 mode, though otherwise they do not decay to
1592 pointers outside C99 mode. For example,
1593 this is valid in GNU C though not valid in C89:
1597 struct foo @{int a[4];@};
1603 return f().a[index];
1609 @section Arithmetic on @code{void}- and Function-Pointers
1610 @cindex void pointers, arithmetic
1611 @cindex void, size of pointer to
1612 @cindex function pointers, arithmetic
1613 @cindex function, size of pointer to
1615 In GNU C, addition and subtraction operations are supported on pointers to
1616 @code{void} and on pointers to functions. This is done by treating the
1617 size of a @code{void} or of a function as 1.
1619 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1620 and on function types, and returns 1.
1622 @opindex Wpointer-arith
1623 The option @option{-Wpointer-arith} requests a warning if these extensions
1627 @section Non-Constant Initializers
1628 @cindex initializers, non-constant
1629 @cindex non-constant initializers
1631 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1632 automatic variable are not required to be constant expressions in GNU C@.
1633 Here is an example of an initializer with run-time varying elements:
1636 foo (float f, float g)
1638 float beat_freqs[2] = @{ f-g, f+g @};
1643 @node Compound Literals
1644 @section Compound Literals
1645 @cindex constructor expressions
1646 @cindex initializations in expressions
1647 @cindex structures, constructor expression
1648 @cindex expressions, constructor
1649 @cindex compound literals
1650 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1652 ISO C99 supports compound literals. A compound literal looks like
1653 a cast containing an initializer. Its value is an object of the
1654 type specified in the cast, containing the elements specified in
1655 the initializer; it is an lvalue. As an extension, GCC supports
1656 compound literals in C89 mode and in C++.
1658 Usually, the specified type is a structure. Assume that
1659 @code{struct foo} and @code{structure} are declared as shown:
1662 struct foo @{int a; char b[2];@} structure;
1666 Here is an example of constructing a @code{struct foo} with a compound literal:
1669 structure = ((struct foo) @{x + y, 'a', 0@});
1673 This is equivalent to writing the following:
1677 struct foo temp = @{x + y, 'a', 0@};
1682 You can also construct an array. If all the elements of the compound literal
1683 are (made up of) simple constant expressions, suitable for use in
1684 initializers of objects of static storage duration, then the compound
1685 literal can be coerced to a pointer to its first element and used in
1686 such an initializer, as shown here:
1689 char **foo = (char *[]) @{ "x", "y", "z" @};
1692 Compound literals for scalar types and union types are is
1693 also allowed, but then the compound literal is equivalent
1696 As a GNU extension, GCC allows initialization of objects with static storage
1697 duration by compound literals (which is not possible in ISO C99, because
1698 the initializer is not a constant).
1699 It is handled as if the object was initialized only with the bracket
1700 enclosed list if compound literal's and object types match.
1701 The initializer list of the compound literal must be constant.
1702 If the object being initialized has array type of unknown size, the size is
1703 determined by compound literal size.
1706 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1707 static int y[] = (int []) @{1, 2, 3@};
1708 static int z[] = (int [3]) @{1@};
1712 The above lines are equivalent to the following:
1714 static struct foo x = @{1, 'a', 'b'@};
1715 static int y[] = @{1, 2, 3@};
1716 static int z[] = @{1, 0, 0@};
1719 @node Designated Inits
1720 @section Designated Initializers
1721 @cindex initializers with labeled elements
1722 @cindex labeled elements in initializers
1723 @cindex case labels in initializers
1724 @cindex designated initializers
1726 Standard C89 requires the elements of an initializer to appear in a fixed
1727 order, the same as the order of the elements in the array or structure
1730 In ISO C99 you can give the elements in any order, specifying the array
1731 indices or structure field names they apply to, and GNU C allows this as
1732 an extension in C89 mode as well. This extension is not
1733 implemented in GNU C++.
1735 To specify an array index, write
1736 @samp{[@var{index}] =} before the element value. For example,
1739 int a[6] = @{ [4] = 29, [2] = 15 @};
1746 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1750 The index values must be constant expressions, even if the array being
1751 initialized is automatic.
1753 An alternative syntax for this which has been obsolete since GCC 2.5 but
1754 GCC still accepts is to write @samp{[@var{index}]} before the element
1755 value, with no @samp{=}.
1757 To initialize a range of elements to the same value, write
1758 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1759 extension. For example,
1762 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1766 If the value in it has side-effects, the side-effects will happen only once,
1767 not for each initialized field by the range initializer.
1770 Note that the length of the array is the highest value specified
1773 In a structure initializer, specify the name of a field to initialize
1774 with @samp{.@var{fieldname} =} before the element value. For example,
1775 given the following structure,
1778 struct point @{ int x, y; @};
1782 the following initialization
1785 struct point p = @{ .y = yvalue, .x = xvalue @};
1792 struct point p = @{ xvalue, yvalue @};
1795 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1796 @samp{@var{fieldname}:}, as shown here:
1799 struct point p = @{ y: yvalue, x: xvalue @};
1803 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1804 @dfn{designator}. You can also use a designator (or the obsolete colon
1805 syntax) when initializing a union, to specify which element of the union
1806 should be used. For example,
1809 union foo @{ int i; double d; @};
1811 union foo f = @{ .d = 4 @};
1815 will convert 4 to a @code{double} to store it in the union using
1816 the second element. By contrast, casting 4 to type @code{union foo}
1817 would store it into the union as the integer @code{i}, since it is
1818 an integer. (@xref{Cast to Union}.)
1820 You can combine this technique of naming elements with ordinary C
1821 initialization of successive elements. Each initializer element that
1822 does not have a designator applies to the next consecutive element of the
1823 array or structure. For example,
1826 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1833 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1836 Labeling the elements of an array initializer is especially useful
1837 when the indices are characters or belong to an @code{enum} type.
1842 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1843 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1846 @cindex designator lists
1847 You can also write a series of @samp{.@var{fieldname}} and
1848 @samp{[@var{index}]} designators before an @samp{=} to specify a
1849 nested subobject to initialize; the list is taken relative to the
1850 subobject corresponding to the closest surrounding brace pair. For
1851 example, with the @samp{struct point} declaration above:
1854 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1858 If the same field is initialized multiple times, it will have value from
1859 the last initialization. If any such overridden initialization has
1860 side-effect, it is unspecified whether the side-effect happens or not.
1861 Currently, GCC will discard them and issue a warning.
1864 @section Case Ranges
1866 @cindex ranges in case statements
1868 You can specify a range of consecutive values in a single @code{case} label,
1872 case @var{low} ... @var{high}:
1876 This has the same effect as the proper number of individual @code{case}
1877 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1879 This feature is especially useful for ranges of ASCII character codes:
1885 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1886 it may be parsed wrong when you use it with integer values. For example,
1901 @section Cast to a Union Type
1902 @cindex cast to a union
1903 @cindex union, casting to a
1905 A cast to union type is similar to other casts, except that the type
1906 specified is a union type. You can specify the type either with
1907 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1908 a constructor though, not a cast, and hence does not yield an lvalue like
1909 normal casts. (@xref{Compound Literals}.)
1911 The types that may be cast to the union type are those of the members
1912 of the union. Thus, given the following union and variables:
1915 union foo @{ int i; double d; @};
1921 both @code{x} and @code{y} can be cast to type @code{union foo}.
1923 Using the cast as the right-hand side of an assignment to a variable of
1924 union type is equivalent to storing in a member of the union:
1929 u = (union foo) x @equiv{} u.i = x
1930 u = (union foo) y @equiv{} u.d = y
1933 You can also use the union cast as a function argument:
1936 void hack (union foo);
1938 hack ((union foo) x);
1941 @node Mixed Declarations
1942 @section Mixed Declarations and Code
1943 @cindex mixed declarations and code
1944 @cindex declarations, mixed with code
1945 @cindex code, mixed with declarations
1947 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1948 within compound statements. As an extension, GCC also allows this in
1949 C89 mode. For example, you could do:
1958 Each identifier is visible from where it is declared until the end of
1959 the enclosing block.
1961 @node Function Attributes
1962 @section Declaring Attributes of Functions
1963 @cindex function attributes
1964 @cindex declaring attributes of functions
1965 @cindex functions that never return
1966 @cindex functions that have no side effects
1967 @cindex functions in arbitrary sections
1968 @cindex functions that behave like malloc
1969 @cindex @code{volatile} applied to function
1970 @cindex @code{const} applied to function
1971 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1972 @cindex functions with non-null pointer arguments
1973 @cindex functions that are passed arguments in registers on the 386
1974 @cindex functions that pop the argument stack on the 386
1975 @cindex functions that do not pop the argument stack on the 386
1977 In GNU C, you declare certain things about functions called in your program
1978 which help the compiler optimize function calls and check your code more
1981 The keyword @code{__attribute__} allows you to specify special
1982 attributes when making a declaration. This keyword is followed by an
1983 attribute specification inside double parentheses. The following
1984 attributes are currently defined for functions on all targets:
1985 @code{noreturn}, @code{noinline}, @code{always_inline},
1986 @code{pure}, @code{const}, @code{nothrow},
1987 @code{format}, @code{format_arg}, @code{no_instrument_function},
1988 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1989 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
1990 @code{alias}, @code{warn_unused_result} and @code{nonnull}. Several other
1991 attributes are defined for functions on particular target systems. Other
1992 attributes, including @code{section} are supported for variables declarations
1993 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1995 You may also specify attributes with @samp{__} preceding and following
1996 each keyword. This allows you to use them in header files without
1997 being concerned about a possible macro of the same name. For example,
1998 you may use @code{__noreturn__} instead of @code{noreturn}.
2000 @xref{Attribute Syntax}, for details of the exact syntax for using
2004 @cindex @code{noreturn} function attribute
2006 A few standard library functions, such as @code{abort} and @code{exit},
2007 cannot return. GCC knows this automatically. Some programs define
2008 their own functions that never return. You can declare them
2009 @code{noreturn} to tell the compiler this fact. For example,
2013 void fatal () __attribute__ ((noreturn));
2016 fatal (/* @r{@dots{}} */)
2018 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2024 The @code{noreturn} keyword tells the compiler to assume that
2025 @code{fatal} cannot return. It can then optimize without regard to what
2026 would happen if @code{fatal} ever did return. This makes slightly
2027 better code. More importantly, it helps avoid spurious warnings of
2028 uninitialized variables.
2030 The @code{noreturn} keyword does not affect the exceptional path when that
2031 applies: a @code{noreturn}-marked function may still return to the caller
2032 by throwing an exception.
2034 Do not assume that registers saved by the calling function are
2035 restored before calling the @code{noreturn} function.
2037 It does not make sense for a @code{noreturn} function to have a return
2038 type other than @code{void}.
2040 The attribute @code{noreturn} is not implemented in GCC versions
2041 earlier than 2.5. An alternative way to declare that a function does
2042 not return, which works in the current version and in some older
2043 versions, is as follows:
2046 typedef void voidfn ();
2048 volatile voidfn fatal;
2051 This approach does not work in GNU C++.
2053 @cindex @code{noinline} function attribute
2055 This function attribute prevents a function from being considered for
2058 @cindex @code{always_inline} function attribute
2060 Generally, functions are not inlined unless optimization is specified.
2061 For functions declared inline, this attribute inlines the function even
2062 if no optimization level was specified.
2064 @cindex @code{pure} function attribute
2066 Many functions have no effects except the return value and their
2067 return value depends only on the parameters and/or global variables.
2068 Such a function can be subject
2069 to common subexpression elimination and loop optimization just as an
2070 arithmetic operator would be. These functions should be declared
2071 with the attribute @code{pure}. For example,
2074 int square (int) __attribute__ ((pure));
2078 says that the hypothetical function @code{square} is safe to call
2079 fewer times than the program says.
2081 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2082 Interesting non-pure functions are functions with infinite loops or those
2083 depending on volatile memory or other system resource, that may change between
2084 two consecutive calls (such as @code{feof} in a multithreading environment).
2086 The attribute @code{pure} is not implemented in GCC versions earlier
2088 @cindex @code{const} function attribute
2090 Many functions do not examine any values except their arguments, and
2091 have no effects except the return value. Basically this is just slightly
2092 more strict class than the @code{pure} attribute above, since function is not
2093 allowed to read global memory.
2095 @cindex pointer arguments
2096 Note that a function that has pointer arguments and examines the data
2097 pointed to must @emph{not} be declared @code{const}. Likewise, a
2098 function that calls a non-@code{const} function usually must not be
2099 @code{const}. It does not make sense for a @code{const} function to
2102 The attribute @code{const} is not implemented in GCC versions earlier
2103 than 2.5. An alternative way to declare that a function has no side
2104 effects, which works in the current version and in some older versions,
2108 typedef int intfn ();
2110 extern const intfn square;
2113 This approach does not work in GNU C++ from 2.6.0 on, since the language
2114 specifies that the @samp{const} must be attached to the return value.
2116 @cindex @code{nothrow} function attribute
2118 The @code{nothrow} attribute is used to inform the compiler that a
2119 function cannot throw an exception. For example, most functions in
2120 the standard C library can be guaranteed not to throw an exception
2121 with the notable exceptions of @code{qsort} and @code{bsearch} that
2122 take function pointer arguments. The @code{nothrow} attribute is not
2123 implemented in GCC versions earlier than 3.2.
2125 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2126 @cindex @code{format} function attribute
2128 The @code{format} attribute specifies that a function takes @code{printf},
2129 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
2130 should be type-checked against a format string. For example, the
2135 my_printf (void *my_object, const char *my_format, ...)
2136 __attribute__ ((format (printf, 2, 3)));
2140 causes the compiler to check the arguments in calls to @code{my_printf}
2141 for consistency with the @code{printf} style format string argument
2144 The parameter @var{archetype} determines how the format string is
2145 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
2146 or @code{strfmon}. (You can also use @code{__printf__},
2147 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
2148 parameter @var{string-index} specifies which argument is the format
2149 string argument (starting from 1), while @var{first-to-check} is the
2150 number of the first argument to check against the format string. For
2151 functions where the arguments are not available to be checked (such as
2152 @code{vprintf}), specify the third parameter as zero. In this case the
2153 compiler only checks the format string for consistency. For
2154 @code{strftime} formats, the third parameter is required to be zero.
2155 Since non-static C++ methods have an implicit @code{this} argument, the
2156 arguments of such methods should be counted from two, not one, when
2157 giving values for @var{string-index} and @var{first-to-check}.
2159 In the example above, the format string (@code{my_format}) is the second
2160 argument of the function @code{my_print}, and the arguments to check
2161 start with the third argument, so the correct parameters for the format
2162 attribute are 2 and 3.
2164 @opindex ffreestanding
2165 The @code{format} attribute allows you to identify your own functions
2166 which take format strings as arguments, so that GCC can check the
2167 calls to these functions for errors. The compiler always (unless
2168 @option{-ffreestanding} is used) checks formats
2169 for the standard library functions @code{printf}, @code{fprintf},
2170 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2171 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2172 warnings are requested (using @option{-Wformat}), so there is no need to
2173 modify the header file @file{stdio.h}. In C99 mode, the functions
2174 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2175 @code{vsscanf} are also checked. Except in strictly conforming C
2176 standard modes, the X/Open function @code{strfmon} is also checked as
2177 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2178 @xref{C Dialect Options,,Options Controlling C Dialect}.
2180 @item format_arg (@var{string-index})
2181 @cindex @code{format_arg} function attribute
2182 @opindex Wformat-nonliteral
2183 The @code{format_arg} attribute specifies that a function takes a format
2184 string for a @code{printf}, @code{scanf}, @code{strftime} or
2185 @code{strfmon} style function and modifies it (for example, to translate
2186 it into another language), so the result can be passed to a
2187 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2188 function (with the remaining arguments to the format function the same
2189 as they would have been for the unmodified string). For example, the
2194 my_dgettext (char *my_domain, const char *my_format)
2195 __attribute__ ((format_arg (2)));
2199 causes the compiler to check the arguments in calls to a @code{printf},
2200 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2201 format string argument is a call to the @code{my_dgettext} function, for
2202 consistency with the format string argument @code{my_format}. If the
2203 @code{format_arg} attribute had not been specified, all the compiler
2204 could tell in such calls to format functions would be that the format
2205 string argument is not constant; this would generate a warning when
2206 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2207 without the attribute.
2209 The parameter @var{string-index} specifies which argument is the format
2210 string argument (starting from one). Since non-static C++ methods have
2211 an implicit @code{this} argument, the arguments of such methods should
2212 be counted from two.
2214 The @code{format-arg} attribute allows you to identify your own
2215 functions which modify format strings, so that GCC can check the
2216 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2217 type function whose operands are a call to one of your own function.
2218 The compiler always treats @code{gettext}, @code{dgettext}, and
2219 @code{dcgettext} in this manner except when strict ISO C support is
2220 requested by @option{-ansi} or an appropriate @option{-std} option, or
2221 @option{-ffreestanding} is used. @xref{C Dialect Options,,Options
2222 Controlling C Dialect}.
2224 @item nonnull (@var{arg-index}, @dots{})
2225 @cindex @code{nonnull} function attribute
2226 The @code{nonnull} attribute specifies that some function parameters should
2227 be non-null pointers. For instance, the declaration:
2231 my_memcpy (void *dest, const void *src, size_t len)
2232 __attribute__((nonnull (1, 2)));
2236 causes the compiler to check that, in calls to @code{my_memcpy},
2237 arguments @var{dest} and @var{src} are non-null. If the compiler
2238 determines that a null pointer is passed in an argument slot marked
2239 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2240 is issued. The compiler may also choose to make optimizations based
2241 on the knowledge that certain function arguments will not be null.
2243 If no argument index list is given to the @code{nonnull} attribute,
2244 all pointer arguments are marked as non-null. To illustrate, the
2245 following declaration is equivalent to the previous example:
2249 my_memcpy (void *dest, const void *src, size_t len)
2250 __attribute__((nonnull));
2253 @item no_instrument_function
2254 @cindex @code{no_instrument_function} function attribute
2255 @opindex finstrument-functions
2256 If @option{-finstrument-functions} is given, profiling function calls will
2257 be generated at entry and exit of most user-compiled functions.
2258 Functions with this attribute will not be so instrumented.
2260 @item section ("@var{section-name}")
2261 @cindex @code{section} function attribute
2262 Normally, the compiler places the code it generates in the @code{text} section.
2263 Sometimes, however, you need additional sections, or you need certain
2264 particular functions to appear in special sections. The @code{section}
2265 attribute specifies that a function lives in a particular section.
2266 For example, the declaration:
2269 extern void foobar (void) __attribute__ ((section ("bar")));
2273 puts the function @code{foobar} in the @code{bar} section.
2275 Some file formats do not support arbitrary sections so the @code{section}
2276 attribute is not available on all platforms.
2277 If you need to map the entire contents of a module to a particular
2278 section, consider using the facilities of the linker instead.
2282 @cindex @code{constructor} function attribute
2283 @cindex @code{destructor} function attribute
2284 The @code{constructor} attribute causes the function to be called
2285 automatically before execution enters @code{main ()}. Similarly, the
2286 @code{destructor} attribute causes the function to be called
2287 automatically after @code{main ()} has completed or @code{exit ()} has
2288 been called. Functions with these attributes are useful for
2289 initializing data that will be used implicitly during the execution of
2292 These attributes are not currently implemented for Objective-C@.
2294 @cindex @code{unused} attribute.
2296 This attribute, attached to a function, means that the function is meant
2297 to be possibly unused. GCC will not produce a warning for this
2300 @cindex @code{used} attribute.
2302 This attribute, attached to a function, means that code must be emitted
2303 for the function even if it appears that the function is not referenced.
2304 This is useful, for example, when the function is referenced only in
2307 @cindex @code{deprecated} attribute.
2309 The @code{deprecated} attribute results in a warning if the function
2310 is used anywhere in the source file. This is useful when identifying
2311 functions that are expected to be removed in a future version of a
2312 program. The warning also includes the location of the declaration
2313 of the deprecated function, to enable users to easily find further
2314 information about why the function is deprecated, or what they should
2315 do instead. Note that the warnings only occurs for uses:
2318 int old_fn () __attribute__ ((deprecated));
2320 int (*fn_ptr)() = old_fn;
2323 results in a warning on line 3 but not line 2.
2325 The @code{deprecated} attribute can also be used for variables and
2326 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2328 @item warn_unused_result
2329 @cindex @code{warn_unused_result} attribute
2330 The @code{warn_unused_result} attribute causes a warning to be emitted
2331 if a caller of the function with this attribute does not use its
2332 return value. This is useful for functions where not checking
2333 the result is either a security problem or always a bug, such as
2337 int fn () __attribute__ ((warn_unused_result));
2340 if (fn () < 0) return -1;
2346 results in warning on line 5.
2349 @cindex @code{weak} attribute
2350 The @code{weak} attribute causes the declaration to be emitted as a weak
2351 symbol rather than a global. This is primarily useful in defining
2352 library functions which can be overridden in user code, though it can
2353 also be used with non-function declarations. Weak symbols are supported
2354 for ELF targets, and also for a.out targets when using the GNU assembler
2358 @cindex @code{malloc} attribute
2359 The @code{malloc} attribute is used to tell the compiler that a function
2360 may be treated as if any non-@code{NULL} pointer it returns cannot
2361 alias any other pointer valid when the function returns.
2362 This will often improve optimization.
2363 Standard functions with this property include @code{malloc} and
2364 @code{calloc}. @code{realloc}-like functions have this property as
2365 long as the old pointer is never referred to (including comparing it
2366 to the new pointer) after the function returns a non-@code{NULL}
2369 @item alias ("@var{target}")
2370 @cindex @code{alias} attribute
2371 The @code{alias} attribute causes the declaration to be emitted as an
2372 alias for another symbol, which must be specified. For instance,
2375 void __f () @{ /* @r{Do something.} */; @}
2376 void f () __attribute__ ((weak, alias ("__f")));
2379 declares @samp{f} to be a weak alias for @samp{__f}. In C++, the
2380 mangled name for the target must be used.
2382 Not all target machines support this attribute.
2384 @item visibility ("@var{visibility_type}")
2385 @cindex @code{visibility} attribute
2386 The @code{visibility} attribute on ELF targets causes the declaration
2387 to be emitted with default, hidden, protected or internal visibility.
2390 void __attribute__ ((visibility ("protected")))
2391 f () @{ /* @r{Do something.} */; @}
2392 int i __attribute__ ((visibility ("hidden")));
2395 See the ELF gABI for complete details, but the short story is:
2399 Default visibility is the normal case for ELF. This value is
2400 available for the visibility attribute to override other options
2401 that may change the assumed visibility of symbols.
2404 Hidden visibility indicates that the symbol will not be placed into
2405 the dynamic symbol table, so no other @dfn{module} (executable or
2406 shared library) can reference it directly.
2409 Protected visibility indicates that the symbol will be placed in the
2410 dynamic symbol table, but that references within the defining module
2411 will bind to the local symbol. That is, the symbol cannot be overridden
2415 Internal visibility is like hidden visibility, but with additional
2416 processor specific semantics. Unless otherwise specified by the psABI,
2417 GCC defines internal visibility to mean that the function is @emph{never}
2418 called from another module. Note that hidden symbols, while they cannot
2419 be referenced directly by other modules, can be referenced indirectly via
2420 function pointers. By indicating that a symbol cannot be called from
2421 outside the module, GCC may for instance omit the load of a PIC register
2422 since it is known that the calling function loaded the correct value.
2425 Not all ELF targets support this attribute.
2427 @item regparm (@var{number})
2428 @cindex @code{regparm} attribute
2429 @cindex functions that are passed arguments in registers on the 386
2430 On the Intel 386, the @code{regparm} attribute causes the compiler to
2431 pass up to @var{number} integer arguments in registers EAX,
2432 EDX, and ECX instead of on the stack. Functions that take a
2433 variable number of arguments will continue to be passed all of their
2434 arguments on the stack.
2436 Beware that on some ELF systems this attribute is unsuitable for
2437 global functions in shared libraries with lazy binding (which is the
2438 default). Lazy binding will send the first call via resolving code in
2439 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2440 per the standard calling conventions. Solaris 8 is affected by this.
2441 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2442 safe since the loaders there save all registers. (Lazy binding can be
2443 disabled with the linker or the loader if desired, to avoid the
2447 @cindex functions that pop the argument stack on the 386
2448 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2449 assume that the called function will pop off the stack space used to
2450 pass arguments, unless it takes a variable number of arguments.
2453 @cindex functions that pop the argument stack on the 386
2454 On the Intel 386, the @code{fastcall} attribute causes the compiler to
2455 pass the first two arguments in the registers ECX and EDX. Subsequent
2456 arguments are passed on the stack. The called function will pop the
2457 arguments off the stack. If the number of arguments is variable all
2458 arguments are pushed on the stack.
2461 @cindex functions that do pop the argument stack on the 386
2463 On the Intel 386, the @code{cdecl} attribute causes the compiler to
2464 assume that the calling function will pop off the stack space used to
2465 pass arguments. This is
2466 useful to override the effects of the @option{-mrtd} switch.
2468 @item longcall/shortcall
2469 @cindex functions called via pointer on the RS/6000 and PowerPC
2470 On the RS/6000 and PowerPC, the @code{longcall} attribute causes the
2471 compiler to always call this function via a pointer, just as it would if
2472 the @option{-mlongcall} option had been specified. The @code{shortcall}
2473 attribute causes the compiler not to do this. These attributes override
2474 both the @option{-mlongcall} switch and the @code{#pragma longcall}
2477 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2478 calls are necessary.
2480 @item long_call/short_call
2481 @cindex indirect calls on ARM
2482 This attribute specifies how a particular function is called on
2483 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2484 command line switch and @code{#pragma long_calls} settings. The
2485 @code{long_call} attribute causes the compiler to always call the
2486 function by first loading its address into a register and then using the
2487 contents of that register. The @code{short_call} attribute always places
2488 the offset to the function from the call site into the @samp{BL}
2489 instruction directly.
2491 @item function_vector
2492 @cindex calling functions through the function vector on the H8/300 processors
2493 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2494 function should be called through the function vector. Calling a
2495 function through the function vector will reduce code size, however;
2496 the function vector has a limited size (maximum 128 entries on the H8/300
2497 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2499 You must use GAS and GLD from GNU binutils version 2.7 or later for
2500 this attribute to work correctly.
2503 @cindex interrupt handler functions
2504 Use this attribute on the ARM, AVR, C4x, M32R/D and Xstormy16 ports to indicate
2505 that the specified function is an interrupt handler. The compiler will
2506 generate function entry and exit sequences suitable for use in an
2507 interrupt handler when this attribute is present.
2509 Note, interrupt handlers for the m68k, H8/300, H8/300H, H8S, and SH processors
2510 can be specified via the @code{interrupt_handler} attribute.
2512 Note, on the AVR, interrupts will be enabled inside the function.
2514 Note, for the ARM, you can specify the kind of interrupt to be handled by
2515 adding an optional parameter to the interrupt attribute like this:
2518 void f () __attribute__ ((interrupt ("IRQ")));
2521 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2523 @item interrupt_handler
2524 @cindex interrupt handler functions on the m68k, H8/300 and SH processors
2525 Use this attribute on the m68k, H8/300, H8/300H, H8S, and SH to indicate that
2526 the specified function is an interrupt handler. The compiler will generate
2527 function entry and exit sequences suitable for use in an interrupt
2528 handler when this attribute is present.
2531 Use this attribute on the SH to indicate an @code{interrupt_handler}
2532 function should switch to an alternate stack. It expects a string
2533 argument that names a global variable holding the address of the
2538 void f () __attribute__ ((interrupt_handler,
2539 sp_switch ("alt_stack")));
2543 Use this attribute on the SH for an @code{interrupt_handler} to return using
2544 @code{trapa} instead of @code{rte}. This attribute expects an integer
2545 argument specifying the trap number to be used.
2548 @cindex eight bit data on the H8/300, H8/300H, and H8S
2549 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2550 variable should be placed into the eight bit data section.
2551 The compiler will generate more efficient code for certain operations
2552 on data in the eight bit data area. Note the eight bit data area is limited to
2555 You must use GAS and GLD from GNU binutils version 2.7 or later for
2556 this attribute to work correctly.
2559 @cindex tiny data section on the H8/300H and H8S
2560 Use this attribute on the H8/300H and H8S to indicate that the specified
2561 variable should be placed into the tiny data section.
2562 The compiler will generate more efficient code for loads and stores
2563 on data in the tiny data section. Note the tiny data area is limited to
2564 slightly under 32kbytes of data.
2567 @cindex save all registers on the H8/300, H8/300H, and H8S
2568 Use this attribute on the H8/300, H8/300H, and H8S to indicate that
2569 all registers except the stack pointer should be saved in the prologue
2570 regardless of whether they are used or not.
2573 @cindex signal handler functions on the AVR processors
2574 Use this attribute on the AVR to indicate that the specified
2575 function is a signal handler. The compiler will generate function
2576 entry and exit sequences suitable for use in a signal handler when this
2577 attribute is present. Interrupts will be disabled inside the function.
2580 @cindex function without a prologue/epilogue code
2581 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
2582 specified function does not need prologue/epilogue sequences generated by
2583 the compiler. It is up to the programmer to provide these sequences.
2585 @item model (@var{model-name})
2586 @cindex function addressability on the M32R/D
2587 @cindex variable addressability on the IA-64
2589 On the M32R/D, use this attribute to set the addressability of an
2590 object, and of the code generated for a function. The identifier
2591 @var{model-name} is one of @code{small}, @code{medium}, or
2592 @code{large}, representing each of the code models.
2594 Small model objects live in the lower 16MB of memory (so that their
2595 addresses can be loaded with the @code{ld24} instruction), and are
2596 callable with the @code{bl} instruction.
2598 Medium model objects may live anywhere in the 32-bit address space (the
2599 compiler will generate @code{seth/add3} instructions to load their addresses),
2600 and are callable with the @code{bl} instruction.
2602 Large model objects may live anywhere in the 32-bit address space (the
2603 compiler will generate @code{seth/add3} instructions to load their addresses),
2604 and may not be reachable with the @code{bl} instruction (the compiler will
2605 generate the much slower @code{seth/add3/jl} instruction sequence).
2607 On IA-64, use this attribute to set the addressability of an object.
2608 At present, the only supported identifier for @var{model-name} is
2609 @code{small}, indicating addressability via ``small'' (22-bit)
2610 addresses (so that their addresses can be loaded with the @code{addl}
2611 instruction). Caveat: such addressing is by definition not position
2612 independent and hence this attribute must not be used for objects
2613 defined by shared libraries.
2616 @cindex functions which handle memory bank switching
2617 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
2618 use a calling convention that takes care of switching memory banks when
2619 entering and leaving a function. This calling convention is also the
2620 default when using the @option{-mlong-calls} option.
2622 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
2623 to call and return from a function.
2625 On 68HC11 the compiler will generate a sequence of instructions
2626 to invoke a board-specific routine to switch the memory bank and call the
2627 real function. The board-specific routine simulates a @code{call}.
2628 At the end of a function, it will jump to a board-specific routine
2629 instead of using @code{rts}. The board-specific return routine simulates
2633 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2634 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2635 use the normal calling convention based on @code{jsr} and @code{rts}.
2636 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2640 @cindex @code{__declspec(dllimport)}
2641 On Microsoft Windows targets, the @code{dllimport} attribute causes the compiler
2642 to reference a function or variable via a global pointer to a pointer
2643 that is set up by the Microsoft Windows dll library. The pointer name is formed by
2644 combining @code{_imp__} and the function or variable name. The attribute
2645 implies @code{extern} storage.
2647 Currently, the attribute is ignored for inlined functions. If the
2648 attribute is applied to a symbol @emph{definition}, an error is reported.
2649 If a symbol previously declared @code{dllimport} is later defined, the
2650 attribute is ignored in subsequent references, and a warning is emitted.
2651 The attribute is also overridden by a subsequent declaration as
2654 When applied to C++ classes, the attribute marks non-inlined
2655 member functions and static data members as imports. However, the
2656 attribute is ignored for virtual methods to allow creation of vtables
2659 On cygwin, mingw and arm-pe targets, @code{__declspec(dllimport)} is
2660 recognized as a synonym for @code{__attribute__ ((dllimport))} for
2661 compatibility with other Microsoft Windows compilers.
2663 The use of the @code{dllimport} attribute on functions is not necessary,
2664 but provides a small performance benefit by eliminating a thunk in the
2665 dll. The use of the @code{dllimport} attribute on imported variables was
2666 required on older versions of GNU ld, but can now be avoided by passing
2667 the @option{--enable-auto-import} switch to ld. As with functions, using
2668 the attribute for a variable eliminates a thunk in the dll.
2670 One drawback to using this attribute is that a pointer to a function or
2671 variable marked as dllimport cannot be used as a constant address. The
2672 attribute can be disabled for functions by setting the
2673 @option{-mnop-fun-dllimport} flag.
2676 @cindex @code{__declspec(dllexport)}
2677 On Microsoft Windows targets the @code{dllexport} attribute causes the compiler to
2678 provide a global pointer to a pointer in a dll, so that it can be
2679 referenced with the @code{dllimport} attribute. The pointer name is
2680 formed by combining @code{_imp__} and the function or variable name.
2682 Currently, the @code{dllexport}attribute is ignored for inlined
2683 functions, but export can be forced by using the
2684 @option{-fkeep-inline-functions} flag. The attribute is also ignored for
2687 When applied to C++ classes. the attribute marks defined non-inlined
2688 member functions and static data members as exports. Static consts
2689 initialized in-class are not marked unless they are also defined
2692 On cygwin, mingw and arm-pe targets, @code{__declspec(dllexport)} is
2693 recognized as a synonym for @code{__attribute__ ((dllexport))} for
2694 compatibility with other Microsoft Windows compilers.
2696 Alternative methods for including the symbol in the dll's export table
2697 are to use a .def file with an @code{EXPORTS} section or, with GNU ld,
2698 using the @option{--export-all} linker flag.
2702 You can specify multiple attributes in a declaration by separating them
2703 by commas within the double parentheses or by immediately following an
2704 attribute declaration with another attribute declaration.
2706 @cindex @code{#pragma}, reason for not using
2707 @cindex pragma, reason for not using
2708 Some people object to the @code{__attribute__} feature, suggesting that
2709 ISO C's @code{#pragma} should be used instead. At the time
2710 @code{__attribute__} was designed, there were two reasons for not doing
2715 It is impossible to generate @code{#pragma} commands from a macro.
2718 There is no telling what the same @code{#pragma} might mean in another
2722 These two reasons applied to almost any application that might have been
2723 proposed for @code{#pragma}. It was basically a mistake to use
2724 @code{#pragma} for @emph{anything}.
2726 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2727 to be generated from macros. In addition, a @code{#pragma GCC}
2728 namespace is now in use for GCC-specific pragmas. However, it has been
2729 found convenient to use @code{__attribute__} to achieve a natural
2730 attachment of attributes to their corresponding declarations, whereas
2731 @code{#pragma GCC} is of use for constructs that do not naturally form
2732 part of the grammar. @xref{Other Directives,,Miscellaneous
2733 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2735 @node Attribute Syntax
2736 @section Attribute Syntax
2737 @cindex attribute syntax
2739 This section describes the syntax with which @code{__attribute__} may be
2740 used, and the constructs to which attribute specifiers bind, for the C
2741 language. Some details may vary for C++ and Objective-C@. Because of
2742 infelicities in the grammar for attributes, some forms described here
2743 may not be successfully parsed in all cases.
2745 There are some problems with the semantics of attributes in C++. For
2746 example, there are no manglings for attributes, although they may affect
2747 code generation, so problems may arise when attributed types are used in
2748 conjunction with templates or overloading. Similarly, @code{typeid}
2749 does not distinguish between types with different attributes. Support
2750 for attributes in C++ may be restricted in future to attributes on
2751 declarations only, but not on nested declarators.
2753 @xref{Function Attributes}, for details of the semantics of attributes
2754 applying to functions. @xref{Variable Attributes}, for details of the
2755 semantics of attributes applying to variables. @xref{Type Attributes},
2756 for details of the semantics of attributes applying to structure, union
2757 and enumerated types.
2759 An @dfn{attribute specifier} is of the form
2760 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2761 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2762 each attribute is one of the following:
2766 Empty. Empty attributes are ignored.
2769 A word (which may be an identifier such as @code{unused}, or a reserved
2770 word such as @code{const}).
2773 A word, followed by, in parentheses, parameters for the attribute.
2774 These parameters take one of the following forms:
2778 An identifier. For example, @code{mode} attributes use this form.
2781 An identifier followed by a comma and a non-empty comma-separated list
2782 of expressions. For example, @code{format} attributes use this form.
2785 A possibly empty comma-separated list of expressions. For example,
2786 @code{format_arg} attributes use this form with the list being a single
2787 integer constant expression, and @code{alias} attributes use this form
2788 with the list being a single string constant.
2792 An @dfn{attribute specifier list} is a sequence of one or more attribute
2793 specifiers, not separated by any other tokens.
2795 In GNU C, an attribute specifier list may appear after the colon following a
2796 label, other than a @code{case} or @code{default} label. The only
2797 attribute it makes sense to use after a label is @code{unused}. This
2798 feature is intended for code generated by programs which contains labels
2799 that may be unused but which is compiled with @option{-Wall}. It would
2800 not normally be appropriate to use in it human-written code, though it
2801 could be useful in cases where the code that jumps to the label is
2802 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2803 such placement of attribute lists, as it is permissible for a
2804 declaration, which could begin with an attribute list, to be labelled in
2805 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2806 does not arise there.
2808 An attribute specifier list may appear as part of a @code{struct},
2809 @code{union} or @code{enum} specifier. It may go either immediately
2810 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2811 the closing brace. It is ignored if the content of the structure, union
2812 or enumerated type is not defined in the specifier in which the
2813 attribute specifier list is used---that is, in usages such as
2814 @code{struct __attribute__((foo)) bar} with no following opening brace.
2815 Where attribute specifiers follow the closing brace, they are considered
2816 to relate to the structure, union or enumerated type defined, not to any
2817 enclosing declaration the type specifier appears in, and the type
2818 defined is not complete until after the attribute specifiers.
2819 @c Otherwise, there would be the following problems: a shift/reduce
2820 @c conflict between attributes binding the struct/union/enum and
2821 @c binding to the list of specifiers/qualifiers; and "aligned"
2822 @c attributes could use sizeof for the structure, but the size could be
2823 @c changed later by "packed" attributes.
2825 Otherwise, an attribute specifier appears as part of a declaration,
2826 counting declarations of unnamed parameters and type names, and relates
2827 to that declaration (which may be nested in another declaration, for
2828 example in the case of a parameter declaration), or to a particular declarator
2829 within a declaration. Where an
2830 attribute specifier is applied to a parameter declared as a function or
2831 an array, it should apply to the function or array rather than the
2832 pointer to which the parameter is implicitly converted, but this is not
2833 yet correctly implemented.
2835 Any list of specifiers and qualifiers at the start of a declaration may
2836 contain attribute specifiers, whether or not such a list may in that
2837 context contain storage class specifiers. (Some attributes, however,
2838 are essentially in the nature of storage class specifiers, and only make
2839 sense where storage class specifiers may be used; for example,
2840 @code{section}.) There is one necessary limitation to this syntax: the
2841 first old-style parameter declaration in a function definition cannot
2842 begin with an attribute specifier, because such an attribute applies to
2843 the function instead by syntax described below (which, however, is not
2844 yet implemented in this case). In some other cases, attribute
2845 specifiers are permitted by this grammar but not yet supported by the
2846 compiler. All attribute specifiers in this place relate to the
2847 declaration as a whole. In the obsolescent usage where a type of
2848 @code{int} is implied by the absence of type specifiers, such a list of
2849 specifiers and qualifiers may be an attribute specifier list with no
2850 other specifiers or qualifiers.
2852 An attribute specifier list may appear immediately before a declarator
2853 (other than the first) in a comma-separated list of declarators in a
2854 declaration of more than one identifier using a single list of
2855 specifiers and qualifiers. Such attribute specifiers apply
2856 only to the identifier before whose declarator they appear. For
2860 __attribute__((noreturn)) void d0 (void),
2861 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2866 the @code{noreturn} attribute applies to all the functions
2867 declared; the @code{format} attribute only applies to @code{d1}.
2869 An attribute specifier list may appear immediately before the comma,
2870 @code{=} or semicolon terminating the declaration of an identifier other
2871 than a function definition. At present, such attribute specifiers apply
2872 to the declared object or function, but in future they may attach to the
2873 outermost adjacent declarator. In simple cases there is no difference,
2874 but, for example, in
2877 void (****f)(void) __attribute__((noreturn));
2881 at present the @code{noreturn} attribute applies to @code{f}, which
2882 causes a warning since @code{f} is not a function, but in future it may
2883 apply to the function @code{****f}. The precise semantics of what
2884 attributes in such cases will apply to are not yet specified. Where an
2885 assembler name for an object or function is specified (@pxref{Asm
2886 Labels}), at present the attribute must follow the @code{asm}
2887 specification; in future, attributes before the @code{asm} specification
2888 may apply to the adjacent declarator, and those after it to the declared
2891 An attribute specifier list may, in future, be permitted to appear after
2892 the declarator in a function definition (before any old-style parameter
2893 declarations or the function body).
2895 Attribute specifiers may be mixed with type qualifiers appearing inside
2896 the @code{[]} of a parameter array declarator, in the C99 construct by
2897 which such qualifiers are applied to the pointer to which the array is
2898 implicitly converted. Such attribute specifiers apply to the pointer,
2899 not to the array, but at present this is not implemented and they are
2902 An attribute specifier list may appear at the start of a nested
2903 declarator. At present, there are some limitations in this usage: the
2904 attributes correctly apply to the declarator, but for most individual
2905 attributes the semantics this implies are not implemented.
2906 When attribute specifiers follow the @code{*} of a pointer
2907 declarator, they may be mixed with any type qualifiers present.
2908 The following describes the formal semantics of this syntax. It will make the
2909 most sense if you are familiar with the formal specification of
2910 declarators in the ISO C standard.
2912 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2913 D1}, where @code{T} contains declaration specifiers that specify a type
2914 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2915 contains an identifier @var{ident}. The type specified for @var{ident}
2916 for derived declarators whose type does not include an attribute
2917 specifier is as in the ISO C standard.
2919 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2920 and the declaration @code{T D} specifies the type
2921 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2922 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2923 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2925 If @code{D1} has the form @code{*
2926 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2927 declaration @code{T D} specifies the type
2928 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2929 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2930 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2936 void (__attribute__((noreturn)) ****f) (void);
2940 specifies the type ``pointer to pointer to pointer to pointer to
2941 non-returning function returning @code{void}''. As another example,
2944 char *__attribute__((aligned(8))) *f;
2948 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2949 Note again that this does not work with most attributes; for example,
2950 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2951 is not yet supported.
2953 For compatibility with existing code written for compiler versions that
2954 did not implement attributes on nested declarators, some laxity is
2955 allowed in the placing of attributes. If an attribute that only applies
2956 to types is applied to a declaration, it will be treated as applying to
2957 the type of that declaration. If an attribute that only applies to
2958 declarations is applied to the type of a declaration, it will be treated
2959 as applying to that declaration; and, for compatibility with code
2960 placing the attributes immediately before the identifier declared, such
2961 an attribute applied to a function return type will be treated as
2962 applying to the function type, and such an attribute applied to an array
2963 element type will be treated as applying to the array type. If an
2964 attribute that only applies to function types is applied to a
2965 pointer-to-function type, it will be treated as applying to the pointer
2966 target type; if such an attribute is applied to a function return type
2967 that is not a pointer-to-function type, it will be treated as applying
2968 to the function type.
2970 @node Function Prototypes
2971 @section Prototypes and Old-Style Function Definitions
2972 @cindex function prototype declarations
2973 @cindex old-style function definitions
2974 @cindex promotion of formal parameters
2976 GNU C extends ISO C to allow a function prototype to override a later
2977 old-style non-prototype definition. Consider the following example:
2980 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2987 /* @r{Prototype function declaration.} */
2988 int isroot P((uid_t));
2990 /* @r{Old-style function definition.} */
2992 isroot (x) /* ??? lossage here ??? */
2999 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
3000 not allow this example, because subword arguments in old-style
3001 non-prototype definitions are promoted. Therefore in this example the
3002 function definition's argument is really an @code{int}, which does not
3003 match the prototype argument type of @code{short}.
3005 This restriction of ISO C makes it hard to write code that is portable
3006 to traditional C compilers, because the programmer does not know
3007 whether the @code{uid_t} type is @code{short}, @code{int}, or
3008 @code{long}. Therefore, in cases like these GNU C allows a prototype
3009 to override a later old-style definition. More precisely, in GNU C, a
3010 function prototype argument type overrides the argument type specified
3011 by a later old-style definition if the former type is the same as the
3012 latter type before promotion. Thus in GNU C the above example is
3013 equivalent to the following:
3026 GNU C++ does not support old-style function definitions, so this
3027 extension is irrelevant.
3030 @section C++ Style Comments
3032 @cindex C++ comments
3033 @cindex comments, C++ style
3035 In GNU C, you may use C++ style comments, which start with @samp{//} and
3036 continue until the end of the line. Many other C implementations allow
3037 such comments, and they are included in the 1999 C standard. However,
3038 C++ style comments are not recognized if you specify an @option{-std}
3039 option specifying a version of ISO C before C99, or @option{-ansi}
3040 (equivalent to @option{-std=c89}).
3043 @section Dollar Signs in Identifier Names
3045 @cindex dollar signs in identifier names
3046 @cindex identifier names, dollar signs in
3048 In GNU C, you may normally use dollar signs in identifier names.
3049 This is because many traditional C implementations allow such identifiers.
3050 However, dollar signs in identifiers are not supported on a few target
3051 machines, typically because the target assembler does not allow them.
3053 @node Character Escapes
3054 @section The Character @key{ESC} in Constants
3056 You can use the sequence @samp{\e} in a string or character constant to
3057 stand for the ASCII character @key{ESC}.
3060 @section Inquiring on Alignment of Types or Variables
3062 @cindex type alignment
3063 @cindex variable alignment
3065 The keyword @code{__alignof__} allows you to inquire about how an object
3066 is aligned, or the minimum alignment usually required by a type. Its
3067 syntax is just like @code{sizeof}.
3069 For example, if the target machine requires a @code{double} value to be
3070 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
3071 This is true on many RISC machines. On more traditional machine
3072 designs, @code{__alignof__ (double)} is 4 or even 2.
3074 Some machines never actually require alignment; they allow reference to any
3075 data type even at an odd address. For these machines, @code{__alignof__}
3076 reports the @emph{recommended} alignment of a type.
3078 If the operand of @code{__alignof__} is an lvalue rather than a type,
3079 its value is the required alignment for its type, taking into account
3080 any minimum alignment specified with GCC's @code{__attribute__}
3081 extension (@pxref{Variable Attributes}). For example, after this
3085 struct foo @{ int x; char y; @} foo1;
3089 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
3090 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
3092 It is an error to ask for the alignment of an incomplete type.
3094 @node Variable Attributes
3095 @section Specifying Attributes of Variables
3096 @cindex attribute of variables
3097 @cindex variable attributes
3099 The keyword @code{__attribute__} allows you to specify special
3100 attributes of variables or structure fields. This keyword is followed
3101 by an attribute specification inside double parentheses. Some
3102 attributes are currently defined generically for variables.
3103 Other attributes are defined for variables on particular target
3104 systems. Other attributes are available for functions
3105 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
3106 Other front ends might define more attributes
3107 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
3109 You may also specify attributes with @samp{__} preceding and following
3110 each keyword. This allows you to use them in header files without
3111 being concerned about a possible macro of the same name. For example,
3112 you may use @code{__aligned__} instead of @code{aligned}.
3114 @xref{Attribute Syntax}, for details of the exact syntax for using
3118 @cindex @code{aligned} attribute
3119 @item aligned (@var{alignment})
3120 This attribute specifies a minimum alignment for the variable or
3121 structure field, measured in bytes. For example, the declaration:
3124 int x __attribute__ ((aligned (16))) = 0;
3128 causes the compiler to allocate the global variable @code{x} on a
3129 16-byte boundary. On a 68040, this could be used in conjunction with
3130 an @code{asm} expression to access the @code{move16} instruction which
3131 requires 16-byte aligned operands.
3133 You can also specify the alignment of structure fields. For example, to
3134 create a double-word aligned @code{int} pair, you could write:
3137 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
3141 This is an alternative to creating a union with a @code{double} member
3142 that forces the union to be double-word aligned.
3144 As in the preceding examples, you can explicitly specify the alignment
3145 (in bytes) that you wish the compiler to use for a given variable or
3146 structure field. Alternatively, you can leave out the alignment factor
3147 and just ask the compiler to align a variable or field to the maximum
3148 useful alignment for the target machine you are compiling for. For
3149 example, you could write:
3152 short array[3] __attribute__ ((aligned));
3155 Whenever you leave out the alignment factor in an @code{aligned} attribute
3156 specification, the compiler automatically sets the alignment for the declared
3157 variable or field to the largest alignment which is ever used for any data
3158 type on the target machine you are compiling for. Doing this can often make
3159 copy operations more efficient, because the compiler can use whatever
3160 instructions copy the biggest chunks of memory when performing copies to
3161 or from the variables or fields that you have aligned this way.
3163 The @code{aligned} attribute can only increase the alignment; but you
3164 can decrease it by specifying @code{packed} as well. See below.
3166 Note that the effectiveness of @code{aligned} attributes may be limited
3167 by inherent limitations in your linker. On many systems, the linker is
3168 only able to arrange for variables to be aligned up to a certain maximum
3169 alignment. (For some linkers, the maximum supported alignment may
3170 be very very small.) If your linker is only able to align variables
3171 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3172 in an @code{__attribute__} will still only provide you with 8 byte
3173 alignment. See your linker documentation for further information.
3175 @item cleanup (@var{cleanup_function})
3176 @cindex @code{cleanup} attribute
3177 The @code{cleanup} attribute runs a function when the variable goes
3178 out of scope. This attribute can only be applied to auto function
3179 scope variables; it may not be applied to parameters or variables
3180 with static storage duration. The function must take one parameter,
3181 a pointer to a type compatible with the variable. The return value
3182 of the function (if any) is ignored.
3184 If @option{-fexceptions} is enabled, then @var{cleanup_function}
3185 will be run during the stack unwinding that happens during the
3186 processing of the exception. Note that the @code{cleanup} attribute
3187 does not allow the exception to be caught, only to perform an action.
3188 It is undefined what happens if @var{cleanup_function} does not
3193 @cindex @code{common} attribute
3194 @cindex @code{nocommon} attribute
3197 The @code{common} attribute requests GCC to place a variable in
3198 ``common'' storage. The @code{nocommon} attribute requests the
3199 opposite -- to allocate space for it directly.
3201 These attributes override the default chosen by the
3202 @option{-fno-common} and @option{-fcommon} flags respectively.
3205 @cindex @code{deprecated} attribute
3206 The @code{deprecated} attribute results in a warning if the variable
3207 is used anywhere in the source file. This is useful when identifying
3208 variables that are expected to be removed in a future version of a
3209 program. The warning also includes the location of the declaration
3210 of the deprecated variable, to enable users to easily find further
3211 information about why the variable is deprecated, or what they should
3212 do instead. Note that the warning only occurs for uses:
3215 extern int old_var __attribute__ ((deprecated));
3217 int new_fn () @{ return old_var; @}
3220 results in a warning on line 3 but not line 2.
3222 The @code{deprecated} attribute can also be used for functions and
3223 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3225 @item mode (@var{mode})
3226 @cindex @code{mode} attribute
3227 This attribute specifies the data type for the declaration---whichever
3228 type corresponds to the mode @var{mode}. This in effect lets you
3229 request an integer or floating point type according to its width.
3231 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3232 indicate the mode corresponding to a one-byte integer, @samp{word} or
3233 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3234 or @samp{__pointer__} for the mode used to represent pointers.
3237 @cindex @code{packed} attribute
3238 The @code{packed} attribute specifies that a variable or structure field
3239 should have the smallest possible alignment---one byte for a variable,
3240 and one bit for a field, unless you specify a larger value with the
3241 @code{aligned} attribute.
3243 Here is a structure in which the field @code{x} is packed, so that it
3244 immediately follows @code{a}:
3250 int x[2] __attribute__ ((packed));
3254 @item section ("@var{section-name}")
3255 @cindex @code{section} variable attribute
3256 Normally, the compiler places the objects it generates in sections like
3257 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3258 or you need certain particular variables to appear in special sections,
3259 for example to map to special hardware. The @code{section}
3260 attribute specifies that a variable (or function) lives in a particular
3261 section. For example, this small program uses several specific section names:
3264 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3265 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3266 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3267 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3271 /* Initialize stack pointer */
3272 init_sp (stack + sizeof (stack));
3274 /* Initialize initialized data */
3275 memcpy (&init_data, &data, &edata - &data);
3277 /* Turn on the serial ports */
3284 Use the @code{section} attribute with an @emph{initialized} definition
3285 of a @emph{global} variable, as shown in the example. GCC issues
3286 a warning and otherwise ignores the @code{section} attribute in
3287 uninitialized variable declarations.
3289 You may only use the @code{section} attribute with a fully initialized
3290 global definition because of the way linkers work. The linker requires
3291 each object be defined once, with the exception that uninitialized
3292 variables tentatively go in the @code{common} (or @code{bss}) section
3293 and can be multiply ``defined''. You can force a variable to be
3294 initialized with the @option{-fno-common} flag or the @code{nocommon}
3297 Some file formats do not support arbitrary sections so the @code{section}
3298 attribute is not available on all platforms.
3299 If you need to map the entire contents of a module to a particular
3300 section, consider using the facilities of the linker instead.
3303 @cindex @code{shared} variable attribute
3304 On Microsoft Windows, in addition to putting variable definitions in a named
3305 section, the section can also be shared among all running copies of an
3306 executable or DLL@. For example, this small program defines shared data
3307 by putting it in a named section @code{shared} and marking the section
3311 int foo __attribute__((section ("shared"), shared)) = 0;
3316 /* Read and write foo. All running
3317 copies see the same value. */
3323 You may only use the @code{shared} attribute along with @code{section}
3324 attribute with a fully initialized global definition because of the way
3325 linkers work. See @code{section} attribute for more information.
3327 The @code{shared} attribute is only available on Microsoft Windows@.
3329 @item tls_model ("@var{tls_model}")
3330 @cindex @code{tls_model} attribute
3331 The @code{tls_model} attribute sets thread-local storage model
3332 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3333 overriding @code{-ftls-model=} command line switch on a per-variable
3335 The @var{tls_model} argument should be one of @code{global-dynamic},
3336 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3338 Not all targets support this attribute.
3340 @item transparent_union
3341 This attribute, attached to a function parameter which is a union, means
3342 that the corresponding argument may have the type of any union member,
3343 but the argument is passed as if its type were that of the first union
3344 member. For more details see @xref{Type Attributes}. You can also use
3345 this attribute on a @code{typedef} for a union data type; then it
3346 applies to all function parameters with that type.
3349 This attribute, attached to a variable, means that the variable is meant
3350 to be possibly unused. GCC will not produce a warning for this
3353 @item vector_size (@var{bytes})
3354 This attribute specifies the vector size for the variable, measured in
3355 bytes. For example, the declaration:
3358 int foo __attribute__ ((vector_size (16)));
3362 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3363 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3364 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3366 This attribute is only applicable to integral and float scalars,
3367 although arrays, pointers, and function return values are allowed in
3368 conjunction with this construct.
3370 Aggregates with this attribute are invalid, even if they are of the same
3371 size as a corresponding scalar. For example, the declaration:
3374 struct S @{ int a; @};
3375 struct S __attribute__ ((vector_size (16))) foo;
3379 is invalid even if the size of the structure is the same as the size of
3383 The @code{weak} attribute is described in @xref{Function Attributes}.
3386 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3389 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3393 @subsection M32R/D Variable Attributes
3395 One attribute is currently defined for the M32R/D.
3398 @item model (@var{model-name})
3399 @cindex variable addressability on the M32R/D
3400 Use this attribute on the M32R/D to set the addressability of an object.
3401 The identifier @var{model-name} is one of @code{small}, @code{medium},
3402 or @code{large}, representing each of the code models.
3404 Small model objects live in the lower 16MB of memory (so that their
3405 addresses can be loaded with the @code{ld24} instruction).
3407 Medium and large model objects may live anywhere in the 32-bit address space
3408 (the compiler will generate @code{seth/add3} instructions to load their
3412 @subsection i386 Variable Attributes
3414 Two attributes are currently defined for i386 configurations:
3415 @code{ms_struct} and @code{gcc_struct}
3420 @cindex @code{ms_struct} attribute
3421 @cindex @code{gcc_struct} attribute
3423 If @code{packed} is used on a structure, or if bit-fields are used
3424 it may be that the Microsoft ABI packs them differently
3425 than GCC would normally pack them. Particularly when moving packed
3426 data between functions compiled with GCC and the native Microsoft compiler
3427 (either via function call or as data in a file), it may be necessary to access
3430 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3431 compilers to match the native Microsoft compiler.
3434 @node Type Attributes
3435 @section Specifying Attributes of Types
3436 @cindex attribute of types
3437 @cindex type attributes
3439 The keyword @code{__attribute__} allows you to specify special
3440 attributes of @code{struct} and @code{union} types when you define such
3441 types. This keyword is followed by an attribute specification inside
3442 double parentheses. Six attributes are currently defined for types:
3443 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3444 @code{deprecated} and @code{may_alias}. Other attributes are defined for
3445 functions (@pxref{Function Attributes}) and for variables
3446 (@pxref{Variable Attributes}).
3448 You may also specify any one of these attributes with @samp{__}
3449 preceding and following its keyword. This allows you to use these
3450 attributes in header files without being concerned about a possible
3451 macro of the same name. For example, you may use @code{__aligned__}
3452 instead of @code{aligned}.
3454 You may specify the @code{aligned} and @code{transparent_union}
3455 attributes either in a @code{typedef} declaration or just past the
3456 closing curly brace of a complete enum, struct or union type
3457 @emph{definition} and the @code{packed} attribute only past the closing
3458 brace of a definition.
3460 You may also specify attributes between the enum, struct or union
3461 tag and the name of the type rather than after the closing brace.
3463 @xref{Attribute Syntax}, for details of the exact syntax for using
3467 @cindex @code{aligned} attribute
3468 @item aligned (@var{alignment})
3469 This attribute specifies a minimum alignment (in bytes) for variables
3470 of the specified type. For example, the declarations:
3473 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3474 typedef int more_aligned_int __attribute__ ((aligned (8)));
3478 force the compiler to insure (as far as it can) that each variable whose
3479 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3480 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3481 variables of type @code{struct S} aligned to 8-byte boundaries allows
3482 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3483 store) instructions when copying one variable of type @code{struct S} to
3484 another, thus improving run-time efficiency.
3486 Note that the alignment of any given @code{struct} or @code{union} type
3487 is required by the ISO C standard to be at least a perfect multiple of
3488 the lowest common multiple of the alignments of all of the members of
3489 the @code{struct} or @code{union} in question. This means that you @emph{can}
3490 effectively adjust the alignment of a @code{struct} or @code{union}
3491 type by attaching an @code{aligned} attribute to any one of the members
3492 of such a type, but the notation illustrated in the example above is a
3493 more obvious, intuitive, and readable way to request the compiler to
3494 adjust the alignment of an entire @code{struct} or @code{union} type.
3496 As in the preceding example, you can explicitly specify the alignment
3497 (in bytes) that you wish the compiler to use for a given @code{struct}
3498 or @code{union} type. Alternatively, you can leave out the alignment factor
3499 and just ask the compiler to align a type to the maximum
3500 useful alignment for the target machine you are compiling for. For
3501 example, you could write:
3504 struct S @{ short f[3]; @} __attribute__ ((aligned));
3507 Whenever you leave out the alignment factor in an @code{aligned}
3508 attribute specification, the compiler automatically sets the alignment
3509 for the type to the largest alignment which is ever used for any data
3510 type on the target machine you are compiling for. Doing this can often
3511 make copy operations more efficient, because the compiler can use
3512 whatever instructions copy the biggest chunks of memory when performing
3513 copies to or from the variables which have types that you have aligned
3516 In the example above, if the size of each @code{short} is 2 bytes, then
3517 the size of the entire @code{struct S} type is 6 bytes. The smallest
3518 power of two which is greater than or equal to that is 8, so the
3519 compiler sets the alignment for the entire @code{struct S} type to 8
3522 Note that although you can ask the compiler to select a time-efficient
3523 alignment for a given type and then declare only individual stand-alone
3524 objects of that type, the compiler's ability to select a time-efficient
3525 alignment is primarily useful only when you plan to create arrays of
3526 variables having the relevant (efficiently aligned) type. If you
3527 declare or use arrays of variables of an efficiently-aligned type, then
3528 it is likely that your program will also be doing pointer arithmetic (or
3529 subscripting, which amounts to the same thing) on pointers to the
3530 relevant type, and the code that the compiler generates for these
3531 pointer arithmetic operations will often be more efficient for
3532 efficiently-aligned types than for other types.
3534 The @code{aligned} attribute can only increase the alignment; but you
3535 can decrease it by specifying @code{packed} as well. See below.
3537 Note that the effectiveness of @code{aligned} attributes may be limited
3538 by inherent limitations in your linker. On many systems, the linker is
3539 only able to arrange for variables to be aligned up to a certain maximum
3540 alignment. (For some linkers, the maximum supported alignment may
3541 be very very small.) If your linker is only able to align variables
3542 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3543 in an @code{__attribute__} will still only provide you with 8 byte
3544 alignment. See your linker documentation for further information.
3547 This attribute, attached to @code{struct} or @code{union} type
3548 definition, specifies that each member of the structure or union is
3549 placed to minimize the memory required. When attached to an @code{enum}
3550 definition, it indicates that the smallest integral type should be used.
3552 @opindex fshort-enums
3553 Specifying this attribute for @code{struct} and @code{union} types is
3554 equivalent to specifying the @code{packed} attribute on each of the
3555 structure or union members. Specifying the @option{-fshort-enums}
3556 flag on the line is equivalent to specifying the @code{packed}
3557 attribute on all @code{enum} definitions.
3559 In the following example @code{struct my_packed_struct}'s members are
3560 packed closely together, but the internal layout of its @code{s} member
3561 is not packed -- to do that, @code{struct my_unpacked_struct} would need to
3565 struct my_unpacked_struct
3571 struct my_packed_struct __attribute__ ((__packed__))
3575 struct my_unpacked_struct s;
3579 You may only specify this attribute on the definition of a @code{enum},
3580 @code{struct} or @code{union}, not on a @code{typedef} which does not
3581 also define the enumerated type, structure or union.
3583 @item transparent_union
3584 This attribute, attached to a @code{union} type definition, indicates
3585 that any function parameter having that union type causes calls to that
3586 function to be treated in a special way.
3588 First, the argument corresponding to a transparent union type can be of
3589 any type in the union; no cast is required. Also, if the union contains
3590 a pointer type, the corresponding argument can be a null pointer
3591 constant or a void pointer expression; and if the union contains a void
3592 pointer type, the corresponding argument can be any pointer expression.
3593 If the union member type is a pointer, qualifiers like @code{const} on
3594 the referenced type must be respected, just as with normal pointer
3597 Second, the argument is passed to the function using the calling
3598 conventions of the first member of the transparent union, not the calling
3599 conventions of the union itself. All members of the union must have the
3600 same machine representation; this is necessary for this argument passing
3603 Transparent unions are designed for library functions that have multiple
3604 interfaces for compatibility reasons. For example, suppose the
3605 @code{wait} function must accept either a value of type @code{int *} to
3606 comply with Posix, or a value of type @code{union wait *} to comply with
3607 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3608 @code{wait} would accept both kinds of arguments, but it would also
3609 accept any other pointer type and this would make argument type checking
3610 less useful. Instead, @code{<sys/wait.h>} might define the interface
3618 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3620 pid_t wait (wait_status_ptr_t);
3623 This interface allows either @code{int *} or @code{union wait *}
3624 arguments to be passed, using the @code{int *} calling convention.
3625 The program can call @code{wait} with arguments of either type:
3628 int w1 () @{ int w; return wait (&w); @}
3629 int w2 () @{ union wait w; return wait (&w); @}
3632 With this interface, @code{wait}'s implementation might look like this:
3635 pid_t wait (wait_status_ptr_t p)
3637 return waitpid (-1, p.__ip, 0);
3642 When attached to a type (including a @code{union} or a @code{struct}),
3643 this attribute means that variables of that type are meant to appear
3644 possibly unused. GCC will not produce a warning for any variables of
3645 that type, even if the variable appears to do nothing. This is often
3646 the case with lock or thread classes, which are usually defined and then
3647 not referenced, but contain constructors and destructors that have
3648 nontrivial bookkeeping functions.
3651 The @code{deprecated} attribute results in a warning if the type
3652 is used anywhere in the source file. This is useful when identifying
3653 types that are expected to be removed in a future version of a program.
3654 If possible, the warning also includes the location of the declaration
3655 of the deprecated type, to enable users to easily find further
3656 information about why the type is deprecated, or what they should do
3657 instead. Note that the warnings only occur for uses and then only
3658 if the type is being applied to an identifier that itself is not being
3659 declared as deprecated.
3662 typedef int T1 __attribute__ ((deprecated));
3666 typedef T1 T3 __attribute__ ((deprecated));
3667 T3 z __attribute__ ((deprecated));
3670 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3671 warning is issued for line 4 because T2 is not explicitly
3672 deprecated. Line 5 has no warning because T3 is explicitly
3673 deprecated. Similarly for line 6.
3675 The @code{deprecated} attribute can also be used for functions and
3676 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3679 Accesses to objects with types with this attribute are not subjected to
3680 type-based alias analysis, but are instead assumed to be able to alias
3681 any other type of objects, just like the @code{char} type. See
3682 @option{-fstrict-aliasing} for more information on aliasing issues.
3687 typedef short __attribute__((__may_alias__)) short_a;
3693 short_a *b = (short_a *) &a;
3697 if (a == 0x12345678)
3704 If you replaced @code{short_a} with @code{short} in the variable
3705 declaration, the above program would abort when compiled with
3706 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3707 above in recent GCC versions.
3709 @subsection i386 Type Attributes
3711 Two attributes are currently defined for i386 configurations:
3712 @code{ms_struct} and @code{gcc_struct}
3716 @cindex @code{ms_struct}
3717 @cindex @code{gcc_struct}
3719 If @code{packed} is used on a structure, or if bit-fields are used
3720 it may be that the Microsoft ABI packs them differently
3721 than GCC would normally pack them. Particularly when moving packed
3722 data between functions compiled with GCC and the native Microsoft compiler
3723 (either via function call or as data in a file), it may be necessary to access
3726 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3727 compilers to match the native Microsoft compiler.
3730 To specify multiple attributes, separate them by commas within the
3731 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3735 @section An Inline Function is As Fast As a Macro
3736 @cindex inline functions
3737 @cindex integrating function code
3739 @cindex macros, inline alternative
3741 By declaring a function @code{inline}, you can direct GCC to
3742 integrate that function's code into the code for its callers. This
3743 makes execution faster by eliminating the function-call overhead; in
3744 addition, if any of the actual argument values are constant, their known
3745 values may permit simplifications at compile time so that not all of the
3746 inline function's code needs to be included. The effect on code size is
3747 less predictable; object code may be larger or smaller with function
3748 inlining, depending on the particular case. Inlining of functions is an
3749 optimization and it really ``works'' only in optimizing compilation. If
3750 you don't use @option{-O}, no function is really inline.
3752 Inline functions are included in the ISO C99 standard, but there are
3753 currently substantial differences between what GCC implements and what
3754 the ISO C99 standard requires.
3756 To declare a function inline, use the @code{inline} keyword in its
3757 declaration, like this:
3767 (If you are writing a header file to be included in ISO C programs, write
3768 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3769 You can also make all ``simple enough'' functions inline with the option
3770 @option{-finline-functions}.
3773 Note that certain usages in a function definition can make it unsuitable
3774 for inline substitution. Among these usages are: use of varargs, use of
3775 alloca, use of variable sized data types (@pxref{Variable Length}),
3776 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3777 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3778 will warn when a function marked @code{inline} could not be substituted,
3779 and will give the reason for the failure.
3781 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3782 does not affect the linkage of the function.
3784 @cindex automatic @code{inline} for C++ member fns
3785 @cindex @code{inline} automatic for C++ member fns
3786 @cindex member fns, automatically @code{inline}
3787 @cindex C++ member fns, automatically @code{inline}
3788 @opindex fno-default-inline
3789 GCC automatically inlines member functions defined within the class
3790 body of C++ programs even if they are not explicitly declared
3791 @code{inline}. (You can override this with @option{-fno-default-inline};
3792 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3794 @cindex inline functions, omission of
3795 @opindex fkeep-inline-functions
3796 When a function is both inline and @code{static}, if all calls to the
3797 function are integrated into the caller, and the function's address is
3798 never used, then the function's own assembler code is never referenced.
3799 In this case, GCC does not actually output assembler code for the
3800 function, unless you specify the option @option{-fkeep-inline-functions}.
3801 Some calls cannot be integrated for various reasons (in particular,
3802 calls that precede the function's definition cannot be integrated, and
3803 neither can recursive calls within the definition). If there is a
3804 nonintegrated call, then the function is compiled to assembler code as
3805 usual. The function must also be compiled as usual if the program
3806 refers to its address, because that can't be inlined.
3808 @cindex non-static inline function
3809 When an inline function is not @code{static}, then the compiler must assume
3810 that there may be calls from other source files; since a global symbol can
3811 be defined only once in any program, the function must not be defined in
3812 the other source files, so the calls therein cannot be integrated.
3813 Therefore, a non-@code{static} inline function is always compiled on its
3814 own in the usual fashion.
3816 If you specify both @code{inline} and @code{extern} in the function
3817 definition, then the definition is used only for inlining. In no case
3818 is the function compiled on its own, not even if you refer to its
3819 address explicitly. Such an address becomes an external reference, as
3820 if you had only declared the function, and had not defined it.
3822 This combination of @code{inline} and @code{extern} has almost the
3823 effect of a macro. The way to use it is to put a function definition in
3824 a header file with these keywords, and put another copy of the
3825 definition (lacking @code{inline} and @code{extern}) in a library file.
3826 The definition in the header file will cause most calls to the function
3827 to be inlined. If any uses of the function remain, they will refer to
3828 the single copy in the library.
3830 Since GCC eventually will implement ISO C99 semantics for
3831 inline functions, it is best to use @code{static inline} only
3832 to guarantee compatibility. (The
3833 existing semantics will remain available when @option{-std=gnu89} is
3834 specified, but eventually the default will be @option{-std=gnu99} and
3835 that will implement the C99 semantics, though it does not do so yet.)
3837 GCC does not inline any functions when not optimizing unless you specify
3838 the @samp{always_inline} attribute for the function, like this:
3842 inline void foo (const char) __attribute__((always_inline));
3846 @section Assembler Instructions with C Expression Operands
3847 @cindex extended @code{asm}
3848 @cindex @code{asm} expressions
3849 @cindex assembler instructions
3852 In an assembler instruction using @code{asm}, you can specify the
3853 operands of the instruction using C expressions. This means you need not
3854 guess which registers or memory locations will contain the data you want
3857 You must specify an assembler instruction template much like what
3858 appears in a machine description, plus an operand constraint string for
3861 For example, here is how to use the 68881's @code{fsinx} instruction:
3864 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3868 Here @code{angle} is the C expression for the input operand while
3869 @code{result} is that of the output operand. Each has @samp{"f"} as its
3870 operand constraint, saying that a floating point register is required.
3871 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3872 output operands' constraints must use @samp{=}. The constraints use the
3873 same language used in the machine description (@pxref{Constraints}).
3875 Each operand is described by an operand-constraint string followed by
3876 the C expression in parentheses. A colon separates the assembler
3877 template from the first output operand and another separates the last
3878 output operand from the first input, if any. Commas separate the
3879 operands within each group. The total number of operands is currently
3880 limited to 30; this limitation may be lifted in some future version of
3883 If there are no output operands but there are input operands, you must
3884 place two consecutive colons surrounding the place where the output
3887 As of GCC version 3.1, it is also possible to specify input and output
3888 operands using symbolic names which can be referenced within the
3889 assembler code. These names are specified inside square brackets
3890 preceding the constraint string, and can be referenced inside the
3891 assembler code using @code{%[@var{name}]} instead of a percentage sign
3892 followed by the operand number. Using named operands the above example
3896 asm ("fsinx %[angle],%[output]"
3897 : [output] "=f" (result)
3898 : [angle] "f" (angle));
3902 Note that the symbolic operand names have no relation whatsoever to
3903 other C identifiers. You may use any name you like, even those of
3904 existing C symbols, but you must ensure that no two operands within the same
3905 assembler construct use the same symbolic name.
3907 Output operand expressions must be lvalues; the compiler can check this.
3908 The input operands need not be lvalues. The compiler cannot check
3909 whether the operands have data types that are reasonable for the
3910 instruction being executed. It does not parse the assembler instruction
3911 template and does not know what it means or even whether it is valid
3912 assembler input. The extended @code{asm} feature is most often used for
3913 machine instructions the compiler itself does not know exist. If
3914 the output expression cannot be directly addressed (for example, it is a
3915 bit-field), your constraint must allow a register. In that case, GCC
3916 will use the register as the output of the @code{asm}, and then store
3917 that register into the output.
3919 The ordinary output operands must be write-only; GCC will assume that
3920 the values in these operands before the instruction are dead and need
3921 not be generated. Extended asm supports input-output or read-write
3922 operands. Use the constraint character @samp{+} to indicate such an
3923 operand and list it with the output operands. You should only use
3924 read-write operands when the constraints for the operand (or the
3925 operand in which only some of the bits are to be changed) allow a
3928 You may, as an alternative, logically split its function into two
3929 separate operands, one input operand and one write-only output
3930 operand. The connection between them is expressed by constraints
3931 which say they need to be in the same location when the instruction
3932 executes. You can use the same C expression for both operands, or
3933 different expressions. For example, here we write the (fictitious)
3934 @samp{combine} instruction with @code{bar} as its read-only source
3935 operand and @code{foo} as its read-write destination:
3938 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3942 The constraint @samp{"0"} for operand 1 says that it must occupy the
3943 same location as operand 0. A number in constraint is allowed only in
3944 an input operand and it must refer to an output operand.
3946 Only a number in the constraint can guarantee that one operand will be in
3947 the same place as another. The mere fact that @code{foo} is the value
3948 of both operands is not enough to guarantee that they will be in the
3949 same place in the generated assembler code. The following would not
3953 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3956 Various optimizations or reloading could cause operands 0 and 1 to be in
3957 different registers; GCC knows no reason not to do so. For example, the
3958 compiler might find a copy of the value of @code{foo} in one register and
3959 use it for operand 1, but generate the output operand 0 in a different
3960 register (copying it afterward to @code{foo}'s own address). Of course,
3961 since the register for operand 1 is not even mentioned in the assembler
3962 code, the result will not work, but GCC can't tell that.
3964 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3965 the operand number for a matching constraint. For example:
3968 asm ("cmoveq %1,%2,%[result]"
3969 : [result] "=r"(result)
3970 : "r" (test), "r"(new), "[result]"(old));
3973 Some instructions clobber specific hard registers. To describe this,
3974 write a third colon after the input operands, followed by the names of
3975 the clobbered hard registers (given as strings). Here is a realistic
3976 example for the VAX:
3979 asm volatile ("movc3 %0,%1,%2"
3981 : "g" (from), "g" (to), "g" (count)
3982 : "r0", "r1", "r2", "r3", "r4", "r5");
3985 You may not write a clobber description in a way that overlaps with an
3986 input or output operand. For example, you may not have an operand
3987 describing a register class with one member if you mention that register
3988 in the clobber list. Variables declared to live in specific registers
3989 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
3990 have no part mentioned in the clobber description.
3991 There is no way for you to specify that an input
3992 operand is modified without also specifying it as an output
3993 operand. Note that if all the output operands you specify are for this
3994 purpose (and hence unused), you will then also need to specify
3995 @code{volatile} for the @code{asm} construct, as described below, to
3996 prevent GCC from deleting the @code{asm} statement as unused.
3998 If you refer to a particular hardware register from the assembler code,
3999 you will probably have to list the register after the third colon to
4000 tell the compiler the register's value is modified. In some assemblers,
4001 the register names begin with @samp{%}; to produce one @samp{%} in the
4002 assembler code, you must write @samp{%%} in the input.
4004 If your assembler instruction can alter the condition code register, add
4005 @samp{cc} to the list of clobbered registers. GCC on some machines
4006 represents the condition codes as a specific hardware register;
4007 @samp{cc} serves to name this register. On other machines, the
4008 condition code is handled differently, and specifying @samp{cc} has no
4009 effect. But it is valid no matter what the machine.
4011 If your assembler instructions access memory in an unpredictable
4012 fashion, add @samp{memory} to the list of clobbered registers. This
4013 will cause GCC to not keep memory values cached in registers across the
4014 assembler instruction and not optimize stores or loads to that memory.
4015 You will also want to add the @code{volatile} keyword if the memory
4016 affected is not listed in the inputs or outputs of the @code{asm}, as
4017 the @samp{memory} clobber does not count as a side-effect of the
4018 @code{asm}. If you know how large the accessed memory is, you can add
4019 it as input or output but if this is not known, you should add
4020 @samp{memory}. As an example, if you access ten bytes of a string, you
4021 can use a memory input like:
4024 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
4027 Note that in the following example the memory input is necessary,
4028 otherwise GCC might optimize the store to @code{x} away:
4035 asm ("magic stuff accessing an 'int' pointed to by '%1'"
4036 "=&d" (r) : "a" (y), "m" (*y));
4041 You can put multiple assembler instructions together in a single
4042 @code{asm} template, separated by the characters normally used in assembly
4043 code for the system. A combination that works in most places is a newline
4044 to break the line, plus a tab character to move to the instruction field
4045 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
4046 assembler allows semicolons as a line-breaking character. Note that some
4047 assembler dialects use semicolons to start a comment.
4048 The input operands are guaranteed not to use any of the clobbered
4049 registers, and neither will the output operands' addresses, so you can
4050 read and write the clobbered registers as many times as you like. Here
4051 is an example of multiple instructions in a template; it assumes the
4052 subroutine @code{_foo} accepts arguments in registers 9 and 10:
4055 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
4057 : "g" (from), "g" (to)
4061 Unless an output operand has the @samp{&} constraint modifier, GCC
4062 may allocate it in the same register as an unrelated input operand, on
4063 the assumption the inputs are consumed before the outputs are produced.
4064 This assumption may be false if the assembler code actually consists of
4065 more than one instruction. In such a case, use @samp{&} for each output
4066 operand that may not overlap an input. @xref{Modifiers}.
4068 If you want to test the condition code produced by an assembler
4069 instruction, you must include a branch and a label in the @code{asm}
4070 construct, as follows:
4073 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
4079 This assumes your assembler supports local labels, as the GNU assembler
4080 and most Unix assemblers do.
4082 Speaking of labels, jumps from one @code{asm} to another are not
4083 supported. The compiler's optimizers do not know about these jumps, and
4084 therefore they cannot take account of them when deciding how to
4087 @cindex macros containing @code{asm}
4088 Usually the most convenient way to use these @code{asm} instructions is to
4089 encapsulate them in macros that look like functions. For example,
4093 (@{ double __value, __arg = (x); \
4094 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
4099 Here the variable @code{__arg} is used to make sure that the instruction
4100 operates on a proper @code{double} value, and to accept only those
4101 arguments @code{x} which can convert automatically to a @code{double}.
4103 Another way to make sure the instruction operates on the correct data
4104 type is to use a cast in the @code{asm}. This is different from using a
4105 variable @code{__arg} in that it converts more different types. For
4106 example, if the desired type were @code{int}, casting the argument to
4107 @code{int} would accept a pointer with no complaint, while assigning the
4108 argument to an @code{int} variable named @code{__arg} would warn about
4109 using a pointer unless the caller explicitly casts it.
4111 If an @code{asm} has output operands, GCC assumes for optimization
4112 purposes the instruction has no side effects except to change the output
4113 operands. This does not mean instructions with a side effect cannot be
4114 used, but you must be careful, because the compiler may eliminate them
4115 if the output operands aren't used, or move them out of loops, or
4116 replace two with one if they constitute a common subexpression. Also,
4117 if your instruction does have a side effect on a variable that otherwise
4118 appears not to change, the old value of the variable may be reused later
4119 if it happens to be found in a register.
4121 You can prevent an @code{asm} instruction from being deleted, moved
4122 significantly, or combined, by writing the keyword @code{volatile} after
4123 the @code{asm}. For example:
4126 #define get_and_set_priority(new) \
4128 asm volatile ("get_and_set_priority %0, %1" \
4129 : "=g" (__old) : "g" (new)); \
4134 If you write an @code{asm} instruction with no outputs, GCC will know
4135 the instruction has side-effects and will not delete the instruction or
4136 move it outside of loops.
4138 The @code{volatile} keyword indicates that the instruction has
4139 important side-effects. GCC will not delete a volatile @code{asm} if
4140 it is reachable. (The instruction can still be deleted if GCC can
4141 prove that control-flow will never reach the location of the
4142 instruction.) In addition, GCC will not reschedule instructions
4143 across a volatile @code{asm} instruction. For example:
4146 *(volatile int *)addr = foo;
4147 asm volatile ("eieio" : : );
4151 Assume @code{addr} contains the address of a memory mapped device
4152 register. The PowerPC @code{eieio} instruction (Enforce In-order
4153 Execution of I/O) tells the CPU to make sure that the store to that
4154 device register happens before it issues any other I/O@.
4156 Note that even a volatile @code{asm} instruction can be moved in ways
4157 that appear insignificant to the compiler, such as across jump
4158 instructions. You can't expect a sequence of volatile @code{asm}
4159 instructions to remain perfectly consecutive. If you want consecutive
4160 output, use a single @code{asm}. Also, GCC will perform some
4161 optimizations across a volatile @code{asm} instruction; GCC does not
4162 ``forget everything'' when it encounters a volatile @code{asm}
4163 instruction the way some other compilers do.
4165 An @code{asm} instruction without any operands or clobbers (an ``old
4166 style'' @code{asm}) will be treated identically to a volatile
4167 @code{asm} instruction.
4169 It is a natural idea to look for a way to give access to the condition
4170 code left by the assembler instruction. However, when we attempted to
4171 implement this, we found no way to make it work reliably. The problem
4172 is that output operands might need reloading, which would result in
4173 additional following ``store'' instructions. On most machines, these
4174 instructions would alter the condition code before there was time to
4175 test it. This problem doesn't arise for ordinary ``test'' and
4176 ``compare'' instructions because they don't have any output operands.
4178 For reasons similar to those described above, it is not possible to give
4179 an assembler instruction access to the condition code left by previous
4182 If you are writing a header file that should be includable in ISO C
4183 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
4186 @subsection Size of an @code{asm}
4188 Some targets require that GCC track the size of each instruction used in
4189 order to generate correct code. Because the final length of an
4190 @code{asm} is only known by the assembler, GCC must make an estimate as
4191 to how big it will be. The estimate is formed by counting the number of
4192 statements in the pattern of the @code{asm} and multiplying that by the
4193 length of the longest instruction on that processor. Statements in the
4194 @code{asm} are identified by newline characters and whatever statement
4195 separator characters are supported by the assembler; on most processors
4196 this is the `@code{;}' character.
4198 Normally, GCC's estimate is perfectly adequate to ensure that correct
4199 code is generated, but it is possible to confuse the compiler if you use
4200 pseudo instructions or assembler macros that expand into multiple real
4201 instructions or if you use assembler directives that expand to more
4202 space in the object file than would be needed for a single instruction.
4203 If this happens then the assembler will produce a diagnostic saying that
4204 a label is unreachable.
4206 @subsection i386 floating point asm operands
4208 There are several rules on the usage of stack-like regs in
4209 asm_operands insns. These rules apply only to the operands that are
4214 Given a set of input regs that die in an asm_operands, it is
4215 necessary to know which are implicitly popped by the asm, and
4216 which must be explicitly popped by gcc.
4218 An input reg that is implicitly popped by the asm must be
4219 explicitly clobbered, unless it is constrained to match an
4223 For any input reg that is implicitly popped by an asm, it is
4224 necessary to know how to adjust the stack to compensate for the pop.
4225 If any non-popped input is closer to the top of the reg-stack than
4226 the implicitly popped reg, it would not be possible to know what the
4227 stack looked like---it's not clear how the rest of the stack ``slides
4230 All implicitly popped input regs must be closer to the top of
4231 the reg-stack than any input that is not implicitly popped.
4233 It is possible that if an input dies in an insn, reload might
4234 use the input reg for an output reload. Consider this example:
4237 asm ("foo" : "=t" (a) : "f" (b));
4240 This asm says that input B is not popped by the asm, and that
4241 the asm pushes a result onto the reg-stack, i.e., the stack is one
4242 deeper after the asm than it was before. But, it is possible that
4243 reload will think that it can use the same reg for both the input and
4244 the output, if input B dies in this insn.
4246 If any input operand uses the @code{f} constraint, all output reg
4247 constraints must use the @code{&} earlyclobber.
4249 The asm above would be written as
4252 asm ("foo" : "=&t" (a) : "f" (b));
4256 Some operands need to be in particular places on the stack. All
4257 output operands fall in this category---there is no other way to
4258 know which regs the outputs appear in unless the user indicates
4259 this in the constraints.
4261 Output operands must specifically indicate which reg an output
4262 appears in after an asm. @code{=f} is not allowed: the operand
4263 constraints must select a class with a single reg.
4266 Output operands may not be ``inserted'' between existing stack regs.
4267 Since no 387 opcode uses a read/write operand, all output operands
4268 are dead before the asm_operands, and are pushed by the asm_operands.
4269 It makes no sense to push anywhere but the top of the reg-stack.
4271 Output operands must start at the top of the reg-stack: output
4272 operands may not ``skip'' a reg.
4275 Some asm statements may need extra stack space for internal
4276 calculations. This can be guaranteed by clobbering stack registers
4277 unrelated to the inputs and outputs.
4281 Here are a couple of reasonable asms to want to write. This asm
4282 takes one input, which is internally popped, and produces two outputs.
4285 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4288 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4289 and replaces them with one output. The user must code the @code{st(1)}
4290 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4293 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4299 @section Controlling Names Used in Assembler Code
4300 @cindex assembler names for identifiers
4301 @cindex names used in assembler code
4302 @cindex identifiers, names in assembler code
4304 You can specify the name to be used in the assembler code for a C
4305 function or variable by writing the @code{asm} (or @code{__asm__})
4306 keyword after the declarator as follows:
4309 int foo asm ("myfoo") = 2;
4313 This specifies that the name to be used for the variable @code{foo} in
4314 the assembler code should be @samp{myfoo} rather than the usual
4317 On systems where an underscore is normally prepended to the name of a C
4318 function or variable, this feature allows you to define names for the
4319 linker that do not start with an underscore.
4321 It does not make sense to use this feature with a non-static local
4322 variable since such variables do not have assembler names. If you are
4323 trying to put the variable in a particular register, see @ref{Explicit
4324 Reg Vars}. GCC presently accepts such code with a warning, but will
4325 probably be changed to issue an error, rather than a warning, in the
4328 You cannot use @code{asm} in this way in a function @emph{definition}; but
4329 you can get the same effect by writing a declaration for the function
4330 before its definition and putting @code{asm} there, like this:
4333 extern func () asm ("FUNC");
4340 It is up to you to make sure that the assembler names you choose do not
4341 conflict with any other assembler symbols. Also, you must not use a
4342 register name; that would produce completely invalid assembler code. GCC
4343 does not as yet have the ability to store static variables in registers.
4344 Perhaps that will be added.
4346 @node Explicit Reg Vars
4347 @section Variables in Specified Registers
4348 @cindex explicit register variables
4349 @cindex variables in specified registers
4350 @cindex specified registers
4351 @cindex registers, global allocation
4353 GNU C allows you to put a few global variables into specified hardware
4354 registers. You can also specify the register in which an ordinary
4355 register variable should be allocated.
4359 Global register variables reserve registers throughout the program.
4360 This may be useful in programs such as programming language
4361 interpreters which have a couple of global variables that are accessed
4365 Local register variables in specific registers do not reserve the
4366 registers. The compiler's data flow analysis is capable of determining
4367 where the specified registers contain live values, and where they are
4368 available for other uses. Stores into local register variables may be deleted
4369 when they appear to be dead according to dataflow analysis. References
4370 to local register variables may be deleted or moved or simplified.
4372 These local variables are sometimes convenient for use with the extended
4373 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4374 output of the assembler instruction directly into a particular register.
4375 (This will work provided the register you specify fits the constraints
4376 specified for that operand in the @code{asm}.)
4384 @node Global Reg Vars
4385 @subsection Defining Global Register Variables
4386 @cindex global register variables
4387 @cindex registers, global variables in
4389 You can define a global register variable in GNU C like this:
4392 register int *foo asm ("a5");
4396 Here @code{a5} is the name of the register which should be used. Choose a
4397 register which is normally saved and restored by function calls on your
4398 machine, so that library routines will not clobber it.
4400 Naturally the register name is cpu-dependent, so you would need to
4401 conditionalize your program according to cpu type. The register
4402 @code{a5} would be a good choice on a 68000 for a variable of pointer
4403 type. On machines with register windows, be sure to choose a ``global''
4404 register that is not affected magically by the function call mechanism.
4406 In addition, operating systems on one type of cpu may differ in how they
4407 name the registers; then you would need additional conditionals. For
4408 example, some 68000 operating systems call this register @code{%a5}.
4410 Eventually there may be a way of asking the compiler to choose a register
4411 automatically, but first we need to figure out how it should choose and
4412 how to enable you to guide the choice. No solution is evident.
4414 Defining a global register variable in a certain register reserves that
4415 register entirely for this use, at least within the current compilation.
4416 The register will not be allocated for any other purpose in the functions
4417 in the current compilation. The register will not be saved and restored by
4418 these functions. Stores into this register are never deleted even if they
4419 would appear to be dead, but references may be deleted or moved or
4422 It is not safe to access the global register variables from signal
4423 handlers, or from more than one thread of control, because the system
4424 library routines may temporarily use the register for other things (unless
4425 you recompile them specially for the task at hand).
4427 @cindex @code{qsort}, and global register variables
4428 It is not safe for one function that uses a global register variable to
4429 call another such function @code{foo} by way of a third function
4430 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4431 different source file in which the variable wasn't declared). This is
4432 because @code{lose} might save the register and put some other value there.
4433 For example, you can't expect a global register variable to be available in
4434 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4435 might have put something else in that register. (If you are prepared to
4436 recompile @code{qsort} with the same global register variable, you can
4437 solve this problem.)
4439 If you want to recompile @code{qsort} or other source files which do not
4440 actually use your global register variable, so that they will not use that
4441 register for any other purpose, then it suffices to specify the compiler
4442 option @option{-ffixed-@var{reg}}. You need not actually add a global
4443 register declaration to their source code.
4445 A function which can alter the value of a global register variable cannot
4446 safely be called from a function compiled without this variable, because it
4447 could clobber the value the caller expects to find there on return.
4448 Therefore, the function which is the entry point into the part of the
4449 program that uses the global register variable must explicitly save and
4450 restore the value which belongs to its caller.
4452 @cindex register variable after @code{longjmp}
4453 @cindex global register after @code{longjmp}
4454 @cindex value after @code{longjmp}
4457 On most machines, @code{longjmp} will restore to each global register
4458 variable the value it had at the time of the @code{setjmp}. On some
4459 machines, however, @code{longjmp} will not change the value of global
4460 register variables. To be portable, the function that called @code{setjmp}
4461 should make other arrangements to save the values of the global register
4462 variables, and to restore them in a @code{longjmp}. This way, the same
4463 thing will happen regardless of what @code{longjmp} does.
4465 All global register variable declarations must precede all function
4466 definitions. If such a declaration could appear after function
4467 definitions, the declaration would be too late to prevent the register from
4468 being used for other purposes in the preceding functions.
4470 Global register variables may not have initial values, because an
4471 executable file has no means to supply initial contents for a register.
4473 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4474 registers, but certain library functions, such as @code{getwd}, as well
4475 as the subroutines for division and remainder, modify g3 and g4. g1 and
4476 g2 are local temporaries.
4478 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4479 Of course, it will not do to use more than a few of those.
4481 @node Local Reg Vars
4482 @subsection Specifying Registers for Local Variables
4483 @cindex local variables, specifying registers
4484 @cindex specifying registers for local variables
4485 @cindex registers for local variables
4487 You can define a local register variable with a specified register
4491 register int *foo asm ("a5");
4495 Here @code{a5} is the name of the register which should be used. Note
4496 that this is the same syntax used for defining global register
4497 variables, but for a local variable it would appear within a function.
4499 Naturally the register name is cpu-dependent, but this is not a
4500 problem, since specific registers are most often useful with explicit
4501 assembler instructions (@pxref{Extended Asm}). Both of these things
4502 generally require that you conditionalize your program according to
4505 In addition, operating systems on one type of cpu may differ in how they
4506 name the registers; then you would need additional conditionals. For
4507 example, some 68000 operating systems call this register @code{%a5}.
4509 Defining such a register variable does not reserve the register; it
4510 remains available for other uses in places where flow control determines
4511 the variable's value is not live. However, these registers are made
4512 unavailable for use in the reload pass; excessive use of this feature
4513 leaves the compiler too few available registers to compile certain
4516 This option does not guarantee that GCC will generate code that has
4517 this variable in the register you specify at all times. You may not
4518 code an explicit reference to this register in an @code{asm} statement
4519 and assume it will always refer to this variable.
4521 Stores into local register variables may be deleted when they appear to be dead
4522 according to dataflow analysis. References to local register variables may
4523 be deleted or moved or simplified.
4525 @node Alternate Keywords
4526 @section Alternate Keywords
4527 @cindex alternate keywords
4528 @cindex keywords, alternate
4530 @option{-ansi} and the various @option{-std} options disable certain
4531 keywords. This causes trouble when you want to use GNU C extensions, or
4532 a general-purpose header file that should be usable by all programs,
4533 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4534 @code{inline} are not available in programs compiled with
4535 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4536 program compiled with @option{-std=c99}). The ISO C99 keyword
4537 @code{restrict} is only available when @option{-std=gnu99} (which will
4538 eventually be the default) or @option{-std=c99} (or the equivalent
4539 @option{-std=iso9899:1999}) is used.
4541 The way to solve these problems is to put @samp{__} at the beginning and
4542 end of each problematical keyword. For example, use @code{__asm__}
4543 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4545 Other C compilers won't accept these alternative keywords; if you want to
4546 compile with another compiler, you can define the alternate keywords as
4547 macros to replace them with the customary keywords. It looks like this:
4555 @findex __extension__
4557 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4559 prevent such warnings within one expression by writing
4560 @code{__extension__} before the expression. @code{__extension__} has no
4561 effect aside from this.
4563 @node Incomplete Enums
4564 @section Incomplete @code{enum} Types
4566 You can define an @code{enum} tag without specifying its possible values.
4567 This results in an incomplete type, much like what you get if you write
4568 @code{struct foo} without describing the elements. A later declaration
4569 which does specify the possible values completes the type.
4571 You can't allocate variables or storage using the type while it is
4572 incomplete. However, you can work with pointers to that type.
4574 This extension may not be very useful, but it makes the handling of
4575 @code{enum} more consistent with the way @code{struct} and @code{union}
4578 This extension is not supported by GNU C++.
4580 @node Function Names
4581 @section Function Names as Strings
4582 @cindex @code{__func__} identifier
4583 @cindex @code{__FUNCTION__} identifier
4584 @cindex @code{__PRETTY_FUNCTION__} identifier
4586 GCC provides three magic variables which hold the name of the current
4587 function, as a string. The first of these is @code{__func__}, which
4588 is part of the C99 standard:
4591 The identifier @code{__func__} is implicitly declared by the translator
4592 as if, immediately following the opening brace of each function
4593 definition, the declaration
4596 static const char __func__[] = "function-name";
4599 appeared, where function-name is the name of the lexically-enclosing
4600 function. This name is the unadorned name of the function.
4603 @code{__FUNCTION__} is another name for @code{__func__}. Older
4604 versions of GCC recognize only this name. However, it is not
4605 standardized. For maximum portability, we recommend you use
4606 @code{__func__}, but provide a fallback definition with the
4610 #if __STDC_VERSION__ < 199901L
4612 # define __func__ __FUNCTION__
4614 # define __func__ "<unknown>"
4619 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4620 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4621 the type signature of the function as well as its bare name. For
4622 example, this program:
4626 extern int printf (char *, ...);
4633 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4634 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4652 __PRETTY_FUNCTION__ = void a::sub(int)
4655 These identifiers are not preprocessor macros. In GCC 3.3 and
4656 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4657 were treated as string literals; they could be used to initialize
4658 @code{char} arrays, and they could be concatenated with other string
4659 literals. GCC 3.4 and later treat them as variables, like
4660 @code{__func__}. In C++, @code{__FUNCTION__} and
4661 @code{__PRETTY_FUNCTION__} have always been variables.
4663 @node Return Address
4664 @section Getting the Return or Frame Address of a Function
4666 These functions may be used to get information about the callers of a
4669 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4670 This function returns the return address of the current function, or of
4671 one of its callers. The @var{level} argument is number of frames to
4672 scan up the call stack. A value of @code{0} yields the return address
4673 of the current function, a value of @code{1} yields the return address
4674 of the caller of the current function, and so forth. When inlining
4675 the expected behavior is that the function will return the address of
4676 the function that will be returned to. To work around this behavior use
4677 the @code{noinline} function attribute.
4679 The @var{level} argument must be a constant integer.
4681 On some machines it may be impossible to determine the return address of
4682 any function other than the current one; in such cases, or when the top
4683 of the stack has been reached, this function will return @code{0} or a
4684 random value. In addition, @code{__builtin_frame_address} may be used
4685 to determine if the top of the stack has been reached.
4687 This function should only be used with a nonzero argument for debugging
4691 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4692 This function is similar to @code{__builtin_return_address}, but it
4693 returns the address of the function frame rather than the return address
4694 of the function. Calling @code{__builtin_frame_address} with a value of
4695 @code{0} yields the frame address of the current function, a value of
4696 @code{1} yields the frame address of the caller of the current function,
4699 The frame is the area on the stack which holds local variables and saved
4700 registers. The frame address is normally the address of the first word
4701 pushed on to the stack by the function. However, the exact definition
4702 depends upon the processor and the calling convention. If the processor
4703 has a dedicated frame pointer register, and the function has a frame,
4704 then @code{__builtin_frame_address} will return the value of the frame
4707 On some machines it may be impossible to determine the frame address of
4708 any function other than the current one; in such cases, or when the top
4709 of the stack has been reached, this function will return @code{0} if
4710 the first frame pointer is properly initialized by the startup code.
4712 This function should only be used with a nonzero argument for debugging
4716 @node Vector Extensions
4717 @section Using vector instructions through built-in functions
4719 On some targets, the instruction set contains SIMD vector instructions that
4720 operate on multiple values contained in one large register at the same time.
4721 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4724 The first step in using these extensions is to provide the necessary data
4725 types. This should be done using an appropriate @code{typedef}:
4728 typedef int v4si __attribute__ ((mode(V4SI)));
4731 The base type @code{int} is effectively ignored by the compiler, the
4732 actual properties of the new type @code{v4si} are defined by the
4733 @code{__attribute__}. It defines the machine mode to be used; for vector
4734 types these have the form @code{V@var{n}@var{B}}; @var{n} should be the
4735 number of elements in the vector, and @var{B} should be the base mode of the
4736 individual elements. The following can be used as base modes:
4740 An integer that is as wide as the smallest addressable unit, usually 8 bits.
4742 An integer, twice as wide as a QI mode integer, usually 16 bits.
4744 An integer, four times as wide as a QI mode integer, usually 32 bits.
4746 An integer, eight times as wide as a QI mode integer, usually 64 bits.
4748 A floating point value, as wide as a SI mode integer, usually 32 bits.
4750 A floating point value, as wide as a DI mode integer, usually 64 bits.
4753 Specifying a combination that is not valid for the current architecture
4754 will cause GCC to synthesize the instructions using a narrower mode.
4755 For example, if you specify a variable of type @code{V4SI} and your
4756 architecture does not allow for this specific SIMD type, GCC will
4757 produce code that uses 4 @code{SIs}.
4759 The types defined in this manner can be used with a subset of normal C
4760 operations. Currently, GCC will allow using the following operators
4761 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4763 The operations behave like C++ @code{valarrays}. Addition is defined as
4764 the addition of the corresponding elements of the operands. For
4765 example, in the code below, each of the 4 elements in @var{a} will be
4766 added to the corresponding 4 elements in @var{b} and the resulting
4767 vector will be stored in @var{c}.
4770 typedef int v4si __attribute__ ((mode(V4SI)));
4777 Subtraction, multiplication, division, and the logical operations
4778 operate in a similar manner. Likewise, the result of using the unary
4779 minus or complement operators on a vector type is a vector whose
4780 elements are the negative or complemented values of the corresponding
4781 elements in the operand.
4783 You can declare variables and use them in function calls and returns, as
4784 well as in assignments and some casts. You can specify a vector type as
4785 a return type for a function. Vector types can also be used as function
4786 arguments. It is possible to cast from one vector type to another,
4787 provided they are of the same size (in fact, you can also cast vectors
4788 to and from other datatypes of the same size).
4790 You cannot operate between vectors of different lengths or different
4791 signedness without a cast.
4793 A port that supports hardware vector operations, usually provides a set
4794 of built-in functions that can be used to operate on vectors. For
4795 example, a function to add two vectors and multiply the result by a
4796 third could look like this:
4799 v4si f (v4si a, v4si b, v4si c)
4801 v4si tmp = __builtin_addv4si (a, b);
4802 return __builtin_mulv4si (tmp, c);
4807 @node Other Builtins
4808 @section Other built-in functions provided by GCC
4809 @cindex built-in functions
4810 @findex __builtin_isgreater
4811 @findex __builtin_isgreaterequal
4812 @findex __builtin_isless
4813 @findex __builtin_islessequal
4814 @findex __builtin_islessgreater
4815 @findex __builtin_isunordered
4970 @findex fprintf_unlocked
4972 @findex fputs_unlocked
5057 @findex printf_unlocked
5083 @findex significandf
5084 @findex significandl
5146 GCC provides a large number of built-in functions other than the ones
5147 mentioned above. Some of these are for internal use in the processing
5148 of exceptions or variable-length argument lists and will not be
5149 documented here because they may change from time to time; we do not
5150 recommend general use of these functions.
5152 The remaining functions are provided for optimization purposes.
5154 @opindex fno-builtin
5155 GCC includes built-in versions of many of the functions in the standard
5156 C library. The versions prefixed with @code{__builtin_} will always be
5157 treated as having the same meaning as the C library function even if you
5158 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
5159 Many of these functions are only optimized in certain cases; if they are
5160 not optimized in a particular case, a call to the library function will
5165 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
5166 @option{-std=c99}), the functions
5167 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
5168 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
5169 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
5170 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
5171 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
5172 @code{index}, @code{j0f}, @code{j0l}, @code{j0}, @code{j1f}, @code{j1l},
5173 @code{j1}, @code{jnf}, @code{jnl}, @code{jn}, @code{mempcpy},
5174 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
5175 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
5176 @code{significandf}, @code{significandl}, @code{significand},
5177 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
5178 @code{strdup}, @code{strfmon}, @code{y0f}, @code{y0l}, @code{y0},
5179 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and @code{yn}
5180 may be handled as built-in functions.
5181 All these functions have corresponding versions
5182 prefixed with @code{__builtin_}, which may be used even in strict C89
5185 The ISO C99 functions
5186 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
5187 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
5188 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
5189 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
5190 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
5191 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
5192 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
5193 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
5194 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
5195 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
5196 @code{cimagl}, @code{cimag},
5197 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf},
5198 @code{copysignl}, @code{copysign}, @code{cpowf}, @code{cpowl},
5199 @code{cpow}, @code{cprojf}, @code{cprojl}, @code{cproj}, @code{crealf},
5200 @code{creall}, @code{creal}, @code{csinf}, @code{csinhf}, @code{csinhl},
5201 @code{csinh}, @code{csinl}, @code{csin}, @code{csqrtf}, @code{csqrtl},
5202 @code{csqrt}, @code{ctanf}, @code{ctanhf}, @code{ctanhl}, @code{ctanh},
5203 @code{ctanl}, @code{ctan}, @code{erfcf}, @code{erfcl}, @code{erfc},
5204 @code{erff}, @code{erfl}, @code{erf}, @code{exp2f}, @code{exp2l},
5205 @code{exp2}, @code{expm1f}, @code{expm1l}, @code{expm1}, @code{fdimf},
5206 @code{fdiml}, @code{fdim}, @code{fmaf}, @code{fmal}, @code{fmaxf},
5207 @code{fmaxl}, @code{fmax}, @code{fma}, @code{fminf}, @code{fminl},
5208 @code{fmin}, @code{hypotf}, @code{hypotl}, @code{hypot}, @code{ilogbf},
5209 @code{ilogbl}, @code{ilogb}, @code{imaxabs}, @code{lgammaf},
5210 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf},
5211 @code{llrintl}, @code{llrint}, @code{llroundf}, @code{llroundl},
5212 @code{llround}, @code{log1pf}, @code{log1pl}, @code{log1p},
5213 @code{log2f}, @code{log2l}, @code{log2}, @code{logbf}, @code{logbl},
5214 @code{logb}, @code{lrintf}, @code{lrintl}, @code{lrint}, @code{lroundf},
5215 @code{lroundl}, @code{lround}, @code{nearbyintf}, @code{nearbyintl},
5216 @code{nearbyint}, @code{nextafterf}, @code{nextafterl},
5217 @code{nextafter}, @code{nexttowardf}, @code{nexttowardl},
5218 @code{nexttoward}, @code{remainderf}, @code{remainderl},
5219 @code{remainder}, @code{remquof}, @code{remquol}, @code{remquo},
5220 @code{rintf}, @code{rintl}, @code{rint}, @code{roundf}, @code{roundl},
5221 @code{round}, @code{scalblnf}, @code{scalblnl}, @code{scalbln},
5222 @code{scalbnf}, @code{scalbnl}, @code{scalbn}, @code{snprintf},
5223 @code{tgammaf}, @code{tgammal}, @code{tgamma}, @code{truncf},
5224 @code{truncl}, @code{trunc}, @code{vfscanf}, @code{vscanf},
5225 @code{vsnprintf} and @code{vsscanf}
5226 are handled as built-in functions
5227 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5229 There are also built-in versions of the ISO C99 functions
5230 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5231 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5232 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5233 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5234 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5235 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5236 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5237 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5238 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5239 that are recognized in any mode since ISO C90 reserves these names for
5240 the purpose to which ISO C99 puts them. All these functions have
5241 corresponding versions prefixed with @code{__builtin_}.
5243 The ISO C90 functions
5244 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5245 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5246 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5247 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf}, @code{labs},
5248 @code{ldexp}, @code{log10}, @code{log}, @code{malloc}, @code{memcmp},
5249 @code{memcpy}, @code{memset}, @code{modf}, @code{pow}, @code{printf},
5250 @code{putchar}, @code{puts}, @code{scanf}, @code{sinh}, @code{sin},
5251 @code{snprintf}, @code{sprintf}, @code{sqrt}, @code{sscanf},
5252 @code{strcat}, @code{strchr}, @code{strcmp}, @code{strcpy},
5253 @code{strcspn}, @code{strlen}, @code{strncat}, @code{strncmp},
5254 @code{strncpy}, @code{strpbrk}, @code{strrchr}, @code{strspn},
5255 @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf}
5257 are all recognized as built-in functions unless
5258 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5259 is specified for an individual function). All of these functions have
5260 corresponding versions prefixed with @code{__builtin_}.
5262 GCC provides built-in versions of the ISO C99 floating point comparison
5263 macros that avoid raising exceptions for unordered operands. They have
5264 the same names as the standard macros ( @code{isgreater},
5265 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5266 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5267 prefixed. We intend for a library implementor to be able to simply
5268 @code{#define} each standard macro to its built-in equivalent.
5270 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
5272 You can use the built-in function @code{__builtin_types_compatible_p} to
5273 determine whether two types are the same.
5275 This built-in function returns 1 if the unqualified versions of the
5276 types @var{type1} and @var{type2} (which are types, not expressions) are
5277 compatible, 0 otherwise. The result of this built-in function can be
5278 used in integer constant expressions.
5280 This built-in function ignores top level qualifiers (e.g., @code{const},
5281 @code{volatile}). For example, @code{int} is equivalent to @code{const
5284 The type @code{int[]} and @code{int[5]} are compatible. On the other
5285 hand, @code{int} and @code{char *} are not compatible, even if the size
5286 of their types, on the particular architecture are the same. Also, the
5287 amount of pointer indirection is taken into account when determining
5288 similarity. Consequently, @code{short *} is not similar to
5289 @code{short **}. Furthermore, two types that are typedefed are
5290 considered compatible if their underlying types are compatible.
5292 An @code{enum} type is not considered to be compatible with another
5293 @code{enum} type even if both are compatible with the same integer
5294 type; this is what the C standard specifies.
5295 For example, @code{enum @{foo, bar@}} is not similar to
5296 @code{enum @{hot, dog@}}.
5298 You would typically use this function in code whose execution varies
5299 depending on the arguments' types. For example:
5305 if (__builtin_types_compatible_p (typeof (x), long double)) \
5306 tmp = foo_long_double (tmp); \
5307 else if (__builtin_types_compatible_p (typeof (x), double)) \
5308 tmp = foo_double (tmp); \
5309 else if (__builtin_types_compatible_p (typeof (x), float)) \
5310 tmp = foo_float (tmp); \
5317 @emph{Note:} This construct is only available for C.
5321 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
5323 You can use the built-in function @code{__builtin_choose_expr} to
5324 evaluate code depending on the value of a constant expression. This
5325 built-in function returns @var{exp1} if @var{const_exp}, which is a
5326 constant expression that must be able to be determined at compile time,
5327 is nonzero. Otherwise it returns 0.
5329 This built-in function is analogous to the @samp{? :} operator in C,
5330 except that the expression returned has its type unaltered by promotion
5331 rules. Also, the built-in function does not evaluate the expression
5332 that was not chosen. For example, if @var{const_exp} evaluates to true,
5333 @var{exp2} is not evaluated even if it has side-effects.
5335 This built-in function can return an lvalue if the chosen argument is an
5338 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5339 type. Similarly, if @var{exp2} is returned, its return type is the same
5346 __builtin_choose_expr ( \
5347 __builtin_types_compatible_p (typeof (x), double), \
5349 __builtin_choose_expr ( \
5350 __builtin_types_compatible_p (typeof (x), float), \
5352 /* @r{The void expression results in a compile-time error} \
5353 @r{when assigning the result to something.} */ \
5357 @emph{Note:} This construct is only available for C. Furthermore, the
5358 unused expression (@var{exp1} or @var{exp2} depending on the value of
5359 @var{const_exp}) may still generate syntax errors. This may change in
5364 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5365 You can use the built-in function @code{__builtin_constant_p} to
5366 determine if a value is known to be constant at compile-time and hence
5367 that GCC can perform constant-folding on expressions involving that
5368 value. The argument of the function is the value to test. The function
5369 returns the integer 1 if the argument is known to be a compile-time
5370 constant and 0 if it is not known to be a compile-time constant. A
5371 return of 0 does not indicate that the value is @emph{not} a constant,
5372 but merely that GCC cannot prove it is a constant with the specified
5373 value of the @option{-O} option.
5375 You would typically use this function in an embedded application where
5376 memory was a critical resource. If you have some complex calculation,
5377 you may want it to be folded if it involves constants, but need to call
5378 a function if it does not. For example:
5381 #define Scale_Value(X) \
5382 (__builtin_constant_p (X) \
5383 ? ((X) * SCALE + OFFSET) : Scale (X))
5386 You may use this built-in function in either a macro or an inline
5387 function. However, if you use it in an inlined function and pass an
5388 argument of the function as the argument to the built-in, GCC will
5389 never return 1 when you call the inline function with a string constant
5390 or compound literal (@pxref{Compound Literals}) and will not return 1
5391 when you pass a constant numeric value to the inline function unless you
5392 specify the @option{-O} option.
5394 You may also use @code{__builtin_constant_p} in initializers for static
5395 data. For instance, you can write
5398 static const int table[] = @{
5399 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5405 This is an acceptable initializer even if @var{EXPRESSION} is not a
5406 constant expression. GCC must be more conservative about evaluating the
5407 built-in in this case, because it has no opportunity to perform
5410 Previous versions of GCC did not accept this built-in in data
5411 initializers. The earliest version where it is completely safe is
5415 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5416 @opindex fprofile-arcs
5417 You may use @code{__builtin_expect} to provide the compiler with
5418 branch prediction information. In general, you should prefer to
5419 use actual profile feedback for this (@option{-fprofile-arcs}), as
5420 programmers are notoriously bad at predicting how their programs
5421 actually perform. However, there are applications in which this
5422 data is hard to collect.
5424 The return value is the value of @var{exp}, which should be an
5425 integral expression. The value of @var{c} must be a compile-time
5426 constant. The semantics of the built-in are that it is expected
5427 that @var{exp} == @var{c}. For example:
5430 if (__builtin_expect (x, 0))
5435 would indicate that we do not expect to call @code{foo}, since
5436 we expect @code{x} to be zero. Since you are limited to integral
5437 expressions for @var{exp}, you should use constructions such as
5440 if (__builtin_expect (ptr != NULL, 1))
5445 when testing pointer or floating-point values.
5448 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5449 This function is used to minimize cache-miss latency by moving data into
5450 a cache before it is accessed.
5451 You can insert calls to @code{__builtin_prefetch} into code for which
5452 you know addresses of data in memory that is likely to be accessed soon.
5453 If the target supports them, data prefetch instructions will be generated.
5454 If the prefetch is done early enough before the access then the data will
5455 be in the cache by the time it is accessed.
5457 The value of @var{addr} is the address of the memory to prefetch.
5458 There are two optional arguments, @var{rw} and @var{locality}.
5459 The value of @var{rw} is a compile-time constant one or zero; one
5460 means that the prefetch is preparing for a write to the memory address
5461 and zero, the default, means that the prefetch is preparing for a read.
5462 The value @var{locality} must be a compile-time constant integer between
5463 zero and three. A value of zero means that the data has no temporal
5464 locality, so it need not be left in the cache after the access. A value
5465 of three means that the data has a high degree of temporal locality and
5466 should be left in all levels of cache possible. Values of one and two
5467 mean, respectively, a low or moderate degree of temporal locality. The
5471 for (i = 0; i < n; i++)
5474 __builtin_prefetch (&a[i+j], 1, 1);
5475 __builtin_prefetch (&b[i+j], 0, 1);
5480 Data prefetch does not generate faults if @var{addr} is invalid, but
5481 the address expression itself must be valid. For example, a prefetch
5482 of @code{p->next} will not fault if @code{p->next} is not a valid
5483 address, but evaluation will fault if @code{p} is not a valid address.
5485 If the target does not support data prefetch, the address expression
5486 is evaluated if it includes side effects but no other code is generated
5487 and GCC does not issue a warning.
5490 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5491 Returns a positive infinity, if supported by the floating-point format,
5492 else @code{DBL_MAX}. This function is suitable for implementing the
5493 ISO C macro @code{HUGE_VAL}.
5496 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5497 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5500 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5501 Similar to @code{__builtin_huge_val}, except the return
5502 type is @code{long double}.
5505 @deftypefn {Built-in Function} double __builtin_inf (void)
5506 Similar to @code{__builtin_huge_val}, except a warning is generated
5507 if the target floating-point format does not support infinities.
5508 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5511 @deftypefn {Built-in Function} float __builtin_inff (void)
5512 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5515 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5516 Similar to @code{__builtin_inf}, except the return
5517 type is @code{long double}.
5520 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5521 This is an implementation of the ISO C99 function @code{nan}.
5523 Since ISO C99 defines this function in terms of @code{strtod}, which we
5524 do not implement, a description of the parsing is in order. The string
5525 is parsed as by @code{strtol}; that is, the base is recognized by
5526 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5527 in the significand such that the least significant bit of the number
5528 is at the least significant bit of the significand. The number is
5529 truncated to fit the significand field provided. The significand is
5530 forced to be a quiet NaN.
5532 This function, if given a string literal, is evaluated early enough
5533 that it is considered a compile-time constant.
5536 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
5537 Similar to @code{__builtin_nan}, except the return type is @code{float}.
5540 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
5541 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
5544 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
5545 Similar to @code{__builtin_nan}, except the significand is forced
5546 to be a signaling NaN. The @code{nans} function is proposed by
5547 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
5550 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
5551 Similar to @code{__builtin_nans}, except the return type is @code{float}.
5554 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
5555 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
5558 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
5559 Returns one plus the index of the least significant 1-bit of @var{x}, or
5560 if @var{x} is zero, returns zero.
5563 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
5564 Returns the number of leading 0-bits in @var{x}, starting at the most
5565 significant bit position. If @var{x} is 0, the result is undefined.
5568 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
5569 Returns the number of trailing 0-bits in @var{x}, starting at the least
5570 significant bit position. If @var{x} is 0, the result is undefined.
5573 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
5574 Returns the number of 1-bits in @var{x}.
5577 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
5578 Returns the parity of @var{x}, i.@:e. the number of 1-bits in @var{x}
5582 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
5583 Similar to @code{__builtin_ffs}, except the argument type is
5584 @code{unsigned long}.
5587 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
5588 Similar to @code{__builtin_clz}, except the argument type is
5589 @code{unsigned long}.
5592 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
5593 Similar to @code{__builtin_ctz}, except the argument type is
5594 @code{unsigned long}.
5597 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
5598 Similar to @code{__builtin_popcount}, except the argument type is
5599 @code{unsigned long}.
5602 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
5603 Similar to @code{__builtin_parity}, except the argument type is
5604 @code{unsigned long}.
5607 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
5608 Similar to @code{__builtin_ffs}, except the argument type is
5609 @code{unsigned long long}.
5612 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
5613 Similar to @code{__builtin_clz}, except the argument type is
5614 @code{unsigned long long}.
5617 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
5618 Similar to @code{__builtin_ctz}, except the argument type is
5619 @code{unsigned long long}.
5622 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
5623 Similar to @code{__builtin_popcount}, except the argument type is
5624 @code{unsigned long long}.
5627 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
5628 Similar to @code{__builtin_parity}, except the argument type is
5629 @code{unsigned long long}.
5633 @node Target Builtins
5634 @section Built-in Functions Specific to Particular Target Machines
5636 On some target machines, GCC supports many built-in functions specific
5637 to those machines. Generally these generate calls to specific machine
5638 instructions, but allow the compiler to schedule those calls.
5641 * Alpha Built-in Functions::
5642 * ARM Built-in Functions::
5643 * X86 Built-in Functions::
5644 * PowerPC AltiVec Built-in Functions::
5647 @node Alpha Built-in Functions
5648 @subsection Alpha Built-in Functions
5650 These built-in functions are available for the Alpha family of
5651 processors, depending on the command-line switches used.
5653 The following built-in functions are always available. They
5654 all generate the machine instruction that is part of the name.
5657 long __builtin_alpha_implver (void)
5658 long __builtin_alpha_rpcc (void)
5659 long __builtin_alpha_amask (long)
5660 long __builtin_alpha_cmpbge (long, long)
5661 long __builtin_alpha_extbl (long, long)
5662 long __builtin_alpha_extwl (long, long)
5663 long __builtin_alpha_extll (long, long)
5664 long __builtin_alpha_extql (long, long)
5665 long __builtin_alpha_extwh (long, long)
5666 long __builtin_alpha_extlh (long, long)
5667 long __builtin_alpha_extqh (long, long)
5668 long __builtin_alpha_insbl (long, long)
5669 long __builtin_alpha_inswl (long, long)
5670 long __builtin_alpha_insll (long, long)
5671 long __builtin_alpha_insql (long, long)
5672 long __builtin_alpha_inswh (long, long)
5673 long __builtin_alpha_inslh (long, long)
5674 long __builtin_alpha_insqh (long, long)
5675 long __builtin_alpha_mskbl (long, long)
5676 long __builtin_alpha_mskwl (long, long)
5677 long __builtin_alpha_mskll (long, long)
5678 long __builtin_alpha_mskql (long, long)
5679 long __builtin_alpha_mskwh (long, long)
5680 long __builtin_alpha_msklh (long, long)
5681 long __builtin_alpha_mskqh (long, long)
5682 long __builtin_alpha_umulh (long, long)
5683 long __builtin_alpha_zap (long, long)
5684 long __builtin_alpha_zapnot (long, long)
5687 The following built-in functions are always with @option{-mmax}
5688 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
5689 later. They all generate the machine instruction that is part
5693 long __builtin_alpha_pklb (long)
5694 long __builtin_alpha_pkwb (long)
5695 long __builtin_alpha_unpkbl (long)
5696 long __builtin_alpha_unpkbw (long)
5697 long __builtin_alpha_minub8 (long, long)
5698 long __builtin_alpha_minsb8 (long, long)
5699 long __builtin_alpha_minuw4 (long, long)
5700 long __builtin_alpha_minsw4 (long, long)
5701 long __builtin_alpha_maxub8 (long, long)
5702 long __builtin_alpha_maxsb8 (long, long)
5703 long __builtin_alpha_maxuw4 (long, long)
5704 long __builtin_alpha_maxsw4 (long, long)
5705 long __builtin_alpha_perr (long, long)
5708 The following built-in functions are always with @option{-mcix}
5709 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
5710 later. They all generate the machine instruction that is part
5714 long __builtin_alpha_cttz (long)
5715 long __builtin_alpha_ctlz (long)
5716 long __builtin_alpha_ctpop (long)
5719 The following builtins are available on systems that use the OSF/1
5720 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
5721 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
5722 @code{rdval} and @code{wrval}.
5725 void *__builtin_thread_pointer (void)
5726 void __builtin_set_thread_pointer (void *)
5729 @node ARM Built-in Functions
5730 @subsection ARM Built-in Functions
5732 These built-in functions are available for the ARM family of
5733 processors, when the @option{-mcpu=iwmmxt} switch is used:
5736 typedef int v2si __attribute__ ((vector_size (8)));
5737 typedef short v4hi __attribute__ ((vector_size (8)));
5738 typedef char v8qi __attribute__ ((vector_size (8)));
5740 int __builtin_arm_getwcx (int)
5741 void __builtin_arm_setwcx (int, int)
5742 int __builtin_arm_textrmsb (v8qi, int)
5743 int __builtin_arm_textrmsh (v4hi, int)
5744 int __builtin_arm_textrmsw (v2si, int)
5745 int __builtin_arm_textrmub (v8qi, int)
5746 int __builtin_arm_textrmuh (v4hi, int)
5747 int __builtin_arm_textrmuw (v2si, int)
5748 v8qi __builtin_arm_tinsrb (v8qi, int)
5749 v4hi __builtin_arm_tinsrh (v4hi, int)
5750 v2si __builtin_arm_tinsrw (v2si, int)
5751 long long __builtin_arm_tmia (long long, int, int)
5752 long long __builtin_arm_tmiabb (long long, int, int)
5753 long long __builtin_arm_tmiabt (long long, int, int)
5754 long long __builtin_arm_tmiaph (long long, int, int)
5755 long long __builtin_arm_tmiatb (long long, int, int)
5756 long long __builtin_arm_tmiatt (long long, int, int)
5757 int __builtin_arm_tmovmskb (v8qi)
5758 int __builtin_arm_tmovmskh (v4hi)
5759 int __builtin_arm_tmovmskw (v2si)
5760 long long __builtin_arm_waccb (v8qi)
5761 long long __builtin_arm_wacch (v4hi)
5762 long long __builtin_arm_waccw (v2si)
5763 v8qi __builtin_arm_waddb (v8qi, v8qi)
5764 v8qi __builtin_arm_waddbss (v8qi, v8qi)
5765 v8qi __builtin_arm_waddbus (v8qi, v8qi)
5766 v4hi __builtin_arm_waddh (v4hi, v4hi)
5767 v4hi __builtin_arm_waddhss (v4hi, v4hi)
5768 v4hi __builtin_arm_waddhus (v4hi, v4hi)
5769 v2si __builtin_arm_waddw (v2si, v2si)
5770 v2si __builtin_arm_waddwss (v2si, v2si)
5771 v2si __builtin_arm_waddwus (v2si, v2si)
5772 v8qi __builtin_arm_walign (v8qi, v8qi, int)
5773 long long __builtin_arm_wand(long long, long long)
5774 long long __builtin_arm_wandn (long long, long long)
5775 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
5776 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
5777 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
5778 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
5779 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
5780 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
5781 v2si __builtin_arm_wcmpeqw (v2si, v2si)
5782 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
5783 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
5784 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
5785 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
5786 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
5787 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
5788 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
5789 long long __builtin_arm_wmacsz (v4hi, v4hi)
5790 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
5791 long long __builtin_arm_wmacuz (v4hi, v4hi)
5792 v4hi __builtin_arm_wmadds (v4hi, v4hi)
5793 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
5794 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
5795 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
5796 v2si __builtin_arm_wmaxsw (v2si, v2si)
5797 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
5798 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
5799 v2si __builtin_arm_wmaxuw (v2si, v2si)
5800 v8qi __builtin_arm_wminsb (v8qi, v8qi)
5801 v4hi __builtin_arm_wminsh (v4hi, v4hi)
5802 v2si __builtin_arm_wminsw (v2si, v2si)
5803 v8qi __builtin_arm_wminub (v8qi, v8qi)
5804 v4hi __builtin_arm_wminuh (v4hi, v4hi)
5805 v2si __builtin_arm_wminuw (v2si, v2si)
5806 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
5807 v4hi __builtin_arm_wmulul (v4hi, v4hi)
5808 v4hi __builtin_arm_wmulum (v4hi, v4hi)
5809 long long __builtin_arm_wor (long long, long long)
5810 v2si __builtin_arm_wpackdss (long long, long long)
5811 v2si __builtin_arm_wpackdus (long long, long long)
5812 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
5813 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
5814 v4hi __builtin_arm_wpackwss (v2si, v2si)
5815 v4hi __builtin_arm_wpackwus (v2si, v2si)
5816 long long __builtin_arm_wrord (long long, long long)
5817 long long __builtin_arm_wrordi (long long, int)
5818 v4hi __builtin_arm_wrorh (v4hi, long long)
5819 v4hi __builtin_arm_wrorhi (v4hi, int)
5820 v2si __builtin_arm_wrorw (v2si, long long)
5821 v2si __builtin_arm_wrorwi (v2si, int)
5822 v2si __builtin_arm_wsadb (v8qi, v8qi)
5823 v2si __builtin_arm_wsadbz (v8qi, v8qi)
5824 v2si __builtin_arm_wsadh (v4hi, v4hi)
5825 v2si __builtin_arm_wsadhz (v4hi, v4hi)
5826 v4hi __builtin_arm_wshufh (v4hi, int)
5827 long long __builtin_arm_wslld (long long, long long)
5828 long long __builtin_arm_wslldi (long long, int)
5829 v4hi __builtin_arm_wsllh (v4hi, long long)
5830 v4hi __builtin_arm_wsllhi (v4hi, int)
5831 v2si __builtin_arm_wsllw (v2si, long long)
5832 v2si __builtin_arm_wsllwi (v2si, int)
5833 long long __builtin_arm_wsrad (long long, long long)
5834 long long __builtin_arm_wsradi (long long, int)
5835 v4hi __builtin_arm_wsrah (v4hi, long long)
5836 v4hi __builtin_arm_wsrahi (v4hi, int)
5837 v2si __builtin_arm_wsraw (v2si, long long)
5838 v2si __builtin_arm_wsrawi (v2si, int)
5839 long long __builtin_arm_wsrld (long long, long long)
5840 long long __builtin_arm_wsrldi (long long, int)
5841 v4hi __builtin_arm_wsrlh (v4hi, long long)
5842 v4hi __builtin_arm_wsrlhi (v4hi, int)
5843 v2si __builtin_arm_wsrlw (v2si, long long)
5844 v2si __builtin_arm_wsrlwi (v2si, int)
5845 v8qi __builtin_arm_wsubb (v8qi, v8qi)
5846 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
5847 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
5848 v4hi __builtin_arm_wsubh (v4hi, v4hi)
5849 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
5850 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
5851 v2si __builtin_arm_wsubw (v2si, v2si)
5852 v2si __builtin_arm_wsubwss (v2si, v2si)
5853 v2si __builtin_arm_wsubwus (v2si, v2si)
5854 v4hi __builtin_arm_wunpckehsb (v8qi)
5855 v2si __builtin_arm_wunpckehsh (v4hi)
5856 long long __builtin_arm_wunpckehsw (v2si)
5857 v4hi __builtin_arm_wunpckehub (v8qi)
5858 v2si __builtin_arm_wunpckehuh (v4hi)
5859 long long __builtin_arm_wunpckehuw (v2si)
5860 v4hi __builtin_arm_wunpckelsb (v8qi)
5861 v2si __builtin_arm_wunpckelsh (v4hi)
5862 long long __builtin_arm_wunpckelsw (v2si)
5863 v4hi __builtin_arm_wunpckelub (v8qi)
5864 v2si __builtin_arm_wunpckeluh (v4hi)
5865 long long __builtin_arm_wunpckeluw (v2si)
5866 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
5867 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
5868 v2si __builtin_arm_wunpckihw (v2si, v2si)
5869 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
5870 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
5871 v2si __builtin_arm_wunpckilw (v2si, v2si)
5872 long long __builtin_arm_wxor (long long, long long)
5873 long long __builtin_arm_wzero ()
5876 @node X86 Built-in Functions
5877 @subsection X86 Built-in Functions
5879 These built-in functions are available for the i386 and x86-64 family
5880 of computers, depending on the command-line switches used.
5882 The following machine modes are available for use with MMX built-in functions
5883 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
5884 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
5885 vector of eight 8-bit integers. Some of the built-in functions operate on
5886 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
5888 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
5889 of two 32-bit floating point values.
5891 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
5892 floating point values. Some instructions use a vector of four 32-bit
5893 integers, these use @code{V4SI}. Finally, some instructions operate on an
5894 entire vector register, interpreting it as a 128-bit integer, these use mode
5897 The following built-in functions are made available by @option{-mmmx}.
5898 All of them generate the machine instruction that is part of the name.
5901 v8qi __builtin_ia32_paddb (v8qi, v8qi)
5902 v4hi __builtin_ia32_paddw (v4hi, v4hi)
5903 v2si __builtin_ia32_paddd (v2si, v2si)
5904 v8qi __builtin_ia32_psubb (v8qi, v8qi)
5905 v4hi __builtin_ia32_psubw (v4hi, v4hi)
5906 v2si __builtin_ia32_psubd (v2si, v2si)
5907 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
5908 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
5909 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
5910 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
5911 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
5912 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
5913 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
5914 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
5915 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
5916 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
5917 di __builtin_ia32_pand (di, di)
5918 di __builtin_ia32_pandn (di,di)
5919 di __builtin_ia32_por (di, di)
5920 di __builtin_ia32_pxor (di, di)
5921 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
5922 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
5923 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
5924 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
5925 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
5926 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
5927 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
5928 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
5929 v2si __builtin_ia32_punpckhdq (v2si, v2si)
5930 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
5931 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
5932 v2si __builtin_ia32_punpckldq (v2si, v2si)
5933 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
5934 v4hi __builtin_ia32_packssdw (v2si, v2si)
5935 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
5938 The following built-in functions are made available either with
5939 @option{-msse}, or with a combination of @option{-m3dnow} and
5940 @option{-march=athlon}. All of them generate the machine
5941 instruction that is part of the name.
5944 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
5945 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
5946 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
5947 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
5948 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
5949 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
5950 v8qi __builtin_ia32_pminub (v8qi, v8qi)
5951 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
5952 int __builtin_ia32_pextrw (v4hi, int)
5953 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
5954 int __builtin_ia32_pmovmskb (v8qi)
5955 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
5956 void __builtin_ia32_movntq (di *, di)
5957 void __builtin_ia32_sfence (void)
5960 The following built-in functions are available when @option{-msse} is used.
5961 All of them generate the machine instruction that is part of the name.
5964 int __builtin_ia32_comieq (v4sf, v4sf)
5965 int __builtin_ia32_comineq (v4sf, v4sf)
5966 int __builtin_ia32_comilt (v4sf, v4sf)
5967 int __builtin_ia32_comile (v4sf, v4sf)
5968 int __builtin_ia32_comigt (v4sf, v4sf)
5969 int __builtin_ia32_comige (v4sf, v4sf)
5970 int __builtin_ia32_ucomieq (v4sf, v4sf)
5971 int __builtin_ia32_ucomineq (v4sf, v4sf)
5972 int __builtin_ia32_ucomilt (v4sf, v4sf)
5973 int __builtin_ia32_ucomile (v4sf, v4sf)
5974 int __builtin_ia32_ucomigt (v4sf, v4sf)
5975 int __builtin_ia32_ucomige (v4sf, v4sf)
5976 v4sf __builtin_ia32_addps (v4sf, v4sf)
5977 v4sf __builtin_ia32_subps (v4sf, v4sf)
5978 v4sf __builtin_ia32_mulps (v4sf, v4sf)
5979 v4sf __builtin_ia32_divps (v4sf, v4sf)
5980 v4sf __builtin_ia32_addss (v4sf, v4sf)
5981 v4sf __builtin_ia32_subss (v4sf, v4sf)
5982 v4sf __builtin_ia32_mulss (v4sf, v4sf)
5983 v4sf __builtin_ia32_divss (v4sf, v4sf)
5984 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
5985 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
5986 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
5987 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
5988 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
5989 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
5990 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
5991 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
5992 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
5993 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
5994 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
5995 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
5996 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
5997 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
5998 v4si __builtin_ia32_cmpless (v4sf, v4sf)
5999 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
6000 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
6001 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
6002 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
6003 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
6004 v4sf __builtin_ia32_maxps (v4sf, v4sf)
6005 v4sf __builtin_ia32_maxss (v4sf, v4sf)
6006 v4sf __builtin_ia32_minps (v4sf, v4sf)
6007 v4sf __builtin_ia32_minss (v4sf, v4sf)
6008 v4sf __builtin_ia32_andps (v4sf, v4sf)
6009 v4sf __builtin_ia32_andnps (v4sf, v4sf)
6010 v4sf __builtin_ia32_orps (v4sf, v4sf)
6011 v4sf __builtin_ia32_xorps (v4sf, v4sf)
6012 v4sf __builtin_ia32_movss (v4sf, v4sf)
6013 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
6014 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
6015 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
6016 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
6017 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
6018 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
6019 v2si __builtin_ia32_cvtps2pi (v4sf)
6020 int __builtin_ia32_cvtss2si (v4sf)
6021 v2si __builtin_ia32_cvttps2pi (v4sf)
6022 int __builtin_ia32_cvttss2si (v4sf)
6023 v4sf __builtin_ia32_rcpps (v4sf)
6024 v4sf __builtin_ia32_rsqrtps (v4sf)
6025 v4sf __builtin_ia32_sqrtps (v4sf)
6026 v4sf __builtin_ia32_rcpss (v4sf)
6027 v4sf __builtin_ia32_rsqrtss (v4sf)
6028 v4sf __builtin_ia32_sqrtss (v4sf)
6029 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
6030 void __builtin_ia32_movntps (float *, v4sf)
6031 int __builtin_ia32_movmskps (v4sf)
6034 The following built-in functions are available when @option{-msse} is used.
6037 @item v4sf __builtin_ia32_loadaps (float *)
6038 Generates the @code{movaps} machine instruction as a load from memory.
6039 @item void __builtin_ia32_storeaps (float *, v4sf)
6040 Generates the @code{movaps} machine instruction as a store to memory.
6041 @item v4sf __builtin_ia32_loadups (float *)
6042 Generates the @code{movups} machine instruction as a load from memory.
6043 @item void __builtin_ia32_storeups (float *, v4sf)
6044 Generates the @code{movups} machine instruction as a store to memory.
6045 @item v4sf __builtin_ia32_loadsss (float *)
6046 Generates the @code{movss} machine instruction as a load from memory.
6047 @item void __builtin_ia32_storess (float *, v4sf)
6048 Generates the @code{movss} machine instruction as a store to memory.
6049 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
6050 Generates the @code{movhps} machine instruction as a load from memory.
6051 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
6052 Generates the @code{movlps} machine instruction as a load from memory
6053 @item void __builtin_ia32_storehps (v4sf, v2si *)
6054 Generates the @code{movhps} machine instruction as a store to memory.
6055 @item void __builtin_ia32_storelps (v4sf, v2si *)
6056 Generates the @code{movlps} machine instruction as a store to memory.
6059 The following built-in functions are available when @option{-msse3} is used.
6060 All of them generate the machine instruction that is part of the name.
6063 v2df __builtin_ia32_addsubpd (v2df, v2df)
6064 v2df __builtin_ia32_addsubps (v2df, v2df)
6065 v2df __builtin_ia32_haddpd (v2df, v2df)
6066 v2df __builtin_ia32_haddps (v2df, v2df)
6067 v2df __builtin_ia32_hsubpd (v2df, v2df)
6068 v2df __builtin_ia32_hsubps (v2df, v2df)
6069 v16qi __builtin_ia32_lddqu (char const *)
6070 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
6071 v2df __builtin_ia32_movddup (v2df)
6072 v4sf __builtin_ia32_movshdup (v4sf)
6073 v4sf __builtin_ia32_movsldup (v4sf)
6074 void __builtin_ia32_mwait (unsigned int, unsigned int)
6077 The following built-in functions are available when @option{-msse3} is used.
6080 @item v2df __builtin_ia32_loadddup (double const *)
6081 Generates the @code{movddup} machine instruction as a load from memory.
6084 The following built-in functions are available when @option{-m3dnow} is used.
6085 All of them generate the machine instruction that is part of the name.
6088 void __builtin_ia32_femms (void)
6089 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
6090 v2si __builtin_ia32_pf2id (v2sf)
6091 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
6092 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
6093 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
6094 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
6095 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
6096 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
6097 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
6098 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
6099 v2sf __builtin_ia32_pfrcp (v2sf)
6100 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
6101 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
6102 v2sf __builtin_ia32_pfrsqrt (v2sf)
6103 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
6104 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
6105 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
6106 v2sf __builtin_ia32_pi2fd (v2si)
6107 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
6110 The following built-in functions are available when both @option{-m3dnow}
6111 and @option{-march=athlon} are used. All of them generate the machine
6112 instruction that is part of the name.
6115 v2si __builtin_ia32_pf2iw (v2sf)
6116 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
6117 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
6118 v2sf __builtin_ia32_pi2fw (v2si)
6119 v2sf __builtin_ia32_pswapdsf (v2sf)
6120 v2si __builtin_ia32_pswapdsi (v2si)
6123 @node PowerPC AltiVec Built-in Functions
6124 @subsection PowerPC AltiVec Built-in Functions
6126 GCC provides an interface for the PowerPC family of processors to access
6127 the AltiVec operations described in Motorola's AltiVec Programming
6128 Interface Manual. The interface is made available by including
6129 @code{<altivec.h>} and using @option{-maltivec} and
6130 @option{-mabi=altivec}. The interface supports the following vector
6134 vector unsigned char
6138 vector unsigned short
6149 GCC's implementation of the high-level language interface available from
6150 C and C++ code differs from Motorola's documentation in several ways.
6155 A vector constant is a list of constant expressions within curly braces.
6158 A vector initializer requires no cast if the vector constant is of the
6159 same type as the variable it is initializing.
6162 If @code{signed} or @code{unsigned} is omitted, the vector type defaults
6163 to @code{signed} for @code{vector int} or @code{vector short} and to
6164 @code{unsigned} for @code{vector char}.
6167 Compiling with @option{-maltivec} adds keywords @code{__vector},
6168 @code{__pixel}, and @code{__bool}. Macros @option{vector},
6169 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
6173 GCC allows using a @code{typedef} name as the type specifier for a
6177 For C, overloaded functions are implemented with macros so the following
6181 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
6184 Since @code{vec_add} is a macro, the vector constant in the example
6185 is treated as four separate arguments. Wrap the entire argument in
6186 parentheses for this to work.
6189 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
6190 Internally, GCC uses built-in functions to achieve the functionality in
6191 the aforementioned header file, but they are not supported and are
6192 subject to change without notice.
6194 The following interfaces are supported for the generic and specific
6195 AltiVec operations and the AltiVec predicates. In cases where there
6196 is a direct mapping between generic and specific operations, only the
6197 generic names are shown here, although the specific operations can also
6200 Arguments that are documented as @code{const int} require literal
6201 integral values within the range required for that operation.
6204 vector signed char vec_abs (vector signed char);
6205 vector signed short vec_abs (vector signed short);
6206 vector signed int vec_abs (vector signed int);
6207 vector float vec_abs (vector float);
6209 vector signed char vec_abss (vector signed char);
6210 vector signed short vec_abss (vector signed short);
6211 vector signed int vec_abss (vector signed int);
6213 vector signed char vec_add (vector bool char, vector signed char);
6214 vector signed char vec_add (vector signed char, vector bool char);
6215 vector signed char vec_add (vector signed char, vector signed char);
6216 vector unsigned char vec_add (vector bool char, vector unsigned char);
6217 vector unsigned char vec_add (vector unsigned char, vector bool char);
6218 vector unsigned char vec_add (vector unsigned char,
6219 vector unsigned char);
6220 vector signed short vec_add (vector bool short, vector signed short);
6221 vector signed short vec_add (vector signed short, vector bool short);
6222 vector signed short vec_add (vector signed short, vector signed short);
6223 vector unsigned short vec_add (vector bool short,
6224 vector unsigned short);
6225 vector unsigned short vec_add (vector unsigned short,
6227 vector unsigned short vec_add (vector unsigned short,
6228 vector unsigned short);
6229 vector signed int vec_add (vector bool int, vector signed int);
6230 vector signed int vec_add (vector signed int, vector bool int);
6231 vector signed int vec_add (vector signed int, vector signed int);
6232 vector unsigned int vec_add (vector bool int, vector unsigned int);
6233 vector unsigned int vec_add (vector unsigned int, vector bool int);
6234 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
6235 vector float vec_add (vector float, vector float);
6237 vector float vec_vaddfp (vector float, vector float);
6239 vector signed int vec_vadduwm (vector bool int, vector signed int);
6240 vector signed int vec_vadduwm (vector signed int, vector bool int);
6241 vector signed int vec_vadduwm (vector signed int, vector signed int);
6242 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
6243 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
6244 vector unsigned int vec_vadduwm (vector unsigned int,
6245 vector unsigned int);
6247 vector signed short vec_vadduhm (vector bool short,
6248 vector signed short);
6249 vector signed short vec_vadduhm (vector signed short,
6251 vector signed short vec_vadduhm (vector signed short,
6252 vector signed short);
6253 vector unsigned short vec_vadduhm (vector bool short,
6254 vector unsigned short);
6255 vector unsigned short vec_vadduhm (vector unsigned short,
6257 vector unsigned short vec_vadduhm (vector unsigned short,
6258 vector unsigned short);
6260 vector signed char vec_vaddubm (vector bool char, vector signed char);
6261 vector signed char vec_vaddubm (vector signed char, vector bool char);
6262 vector signed char vec_vaddubm (vector signed char, vector signed char);
6263 vector unsigned char vec_vaddubm (vector bool char,
6264 vector unsigned char);
6265 vector unsigned char vec_vaddubm (vector unsigned char,
6267 vector unsigned char vec_vaddubm (vector unsigned char,
6268 vector unsigned char);
6270 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
6272 vector unsigned char vec_adds (vector bool char, vector unsigned char);
6273 vector unsigned char vec_adds (vector unsigned char, vector bool char);
6274 vector unsigned char vec_adds (vector unsigned char,
6275 vector unsigned char);
6276 vector signed char vec_adds (vector bool char, vector signed char);
6277 vector signed char vec_adds (vector signed char, vector bool char);
6278 vector signed char vec_adds (vector signed char, vector signed char);
6279 vector unsigned short vec_adds (vector bool short,
6280 vector unsigned short);
6281 vector unsigned short vec_adds (vector unsigned short,
6283 vector unsigned short vec_adds (vector unsigned short,
6284 vector unsigned short);
6285 vector signed short vec_adds (vector bool short, vector signed short);
6286 vector signed short vec_adds (vector signed short, vector bool short);
6287 vector signed short vec_adds (vector signed short, vector signed short);
6288 vector unsigned int vec_adds (vector bool int, vector unsigned int);
6289 vector unsigned int vec_adds (vector unsigned int, vector bool int);
6290 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
6291 vector signed int vec_adds (vector bool int, vector signed int);
6292 vector signed int vec_adds (vector signed int, vector bool int);
6293 vector signed int vec_adds (vector signed int, vector signed int);
6295 vector signed int vec_vaddsws (vector bool int, vector signed int);
6296 vector signed int vec_vaddsws (vector signed int, vector bool int);
6297 vector signed int vec_vaddsws (vector signed int, vector signed int);
6299 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
6300 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
6301 vector unsigned int vec_vadduws (vector unsigned int,
6302 vector unsigned int);
6304 vector signed short vec_vaddshs (vector bool short,
6305 vector signed short);
6306 vector signed short vec_vaddshs (vector signed short,
6308 vector signed short vec_vaddshs (vector signed short,
6309 vector signed short);
6311 vector unsigned short vec_vadduhs (vector bool short,
6312 vector unsigned short);
6313 vector unsigned short vec_vadduhs (vector unsigned short,
6315 vector unsigned short vec_vadduhs (vector unsigned short,
6316 vector unsigned short);
6318 vector signed char vec_vaddsbs (vector bool char, vector signed char);
6319 vector signed char vec_vaddsbs (vector signed char, vector bool char);
6320 vector signed char vec_vaddsbs (vector signed char, vector signed char);
6322 vector unsigned char vec_vaddubs (vector bool char,
6323 vector unsigned char);
6324 vector unsigned char vec_vaddubs (vector unsigned char,
6326 vector unsigned char vec_vaddubs (vector unsigned char,
6327 vector unsigned char);
6329 vector float vec_and (vector float, vector float);
6330 vector float vec_and (vector float, vector bool int);
6331 vector float vec_and (vector bool int, vector float);
6332 vector bool int vec_and (vector bool int, vector bool int);
6333 vector signed int vec_and (vector bool int, vector signed int);
6334 vector signed int vec_and (vector signed int, vector bool int);
6335 vector signed int vec_and (vector signed int, vector signed int);
6336 vector unsigned int vec_and (vector bool int, vector unsigned int);
6337 vector unsigned int vec_and (vector unsigned int, vector bool int);
6338 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
6339 vector bool short vec_and (vector bool short, vector bool short);
6340 vector signed short vec_and (vector bool short, vector signed short);
6341 vector signed short vec_and (vector signed short, vector bool short);
6342 vector signed short vec_and (vector signed short, vector signed short);
6343 vector unsigned short vec_and (vector bool short,
6344 vector unsigned short);
6345 vector unsigned short vec_and (vector unsigned short,
6347 vector unsigned short vec_and (vector unsigned short,
6348 vector unsigned short);
6349 vector signed char vec_and (vector bool char, vector signed char);
6350 vector bool char vec_and (vector bool char, vector bool char);
6351 vector signed char vec_and (vector signed char, vector bool char);
6352 vector signed char vec_and (vector signed char, vector signed char);
6353 vector unsigned char vec_and (vector bool char, vector unsigned char);
6354 vector unsigned char vec_and (vector unsigned char, vector bool char);
6355 vector unsigned char vec_and (vector unsigned char,
6356 vector unsigned char);
6358 vector float vec_andc (vector float, vector float);
6359 vector float vec_andc (vector float, vector bool int);
6360 vector float vec_andc (vector bool int, vector float);
6361 vector bool int vec_andc (vector bool int, vector bool int);
6362 vector signed int vec_andc (vector bool int, vector signed int);
6363 vector signed int vec_andc (vector signed int, vector bool int);
6364 vector signed int vec_andc (vector signed int, vector signed int);
6365 vector unsigned int vec_andc (vector bool int, vector unsigned int);
6366 vector unsigned int vec_andc (vector unsigned int, vector bool int);
6367 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
6368 vector bool short vec_andc (vector bool short, vector bool short);
6369 vector signed short vec_andc (vector bool short, vector signed short);
6370 vector signed short vec_andc (vector signed short, vector bool short);
6371 vector signed short vec_andc (vector signed short, vector signed short);
6372 vector unsigned short vec_andc (vector bool short,
6373 vector unsigned short);
6374 vector unsigned short vec_andc (vector unsigned short,
6376 vector unsigned short vec_andc (vector unsigned short,
6377 vector unsigned short);
6378 vector signed char vec_andc (vector bool char, vector signed char);
6379 vector bool char vec_andc (vector bool char, vector bool char);
6380 vector signed char vec_andc (vector signed char, vector bool char);
6381 vector signed char vec_andc (vector signed char, vector signed char);
6382 vector unsigned char vec_andc (vector bool char, vector unsigned char);
6383 vector unsigned char vec_andc (vector unsigned char, vector bool char);
6384 vector unsigned char vec_andc (vector unsigned char,
6385 vector unsigned char);
6387 vector unsigned char vec_avg (vector unsigned char,
6388 vector unsigned char);
6389 vector signed char vec_avg (vector signed char, vector signed char);
6390 vector unsigned short vec_avg (vector unsigned short,
6391 vector unsigned short);
6392 vector signed short vec_avg (vector signed short, vector signed short);
6393 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
6394 vector signed int vec_avg (vector signed int, vector signed int);
6396 vector signed int vec_vavgsw (vector signed int, vector signed int);
6398 vector unsigned int vec_vavguw (vector unsigned int,
6399 vector unsigned int);
6401 vector signed short vec_vavgsh (vector signed short,
6402 vector signed short);
6404 vector unsigned short vec_vavguh (vector unsigned short,
6405 vector unsigned short);
6407 vector signed char vec_vavgsb (vector signed char, vector signed char);
6409 vector unsigned char vec_vavgub (vector unsigned char,
6410 vector unsigned char);
6412 vector float vec_ceil (vector float);
6414 vector signed int vec_cmpb (vector float, vector float);
6416 vector bool char vec_cmpeq (vector signed char, vector signed char);
6417 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
6418 vector bool short vec_cmpeq (vector signed short, vector signed short);
6419 vector bool short vec_cmpeq (vector unsigned short,
6420 vector unsigned short);
6421 vector bool int vec_cmpeq (vector signed int, vector signed int);
6422 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
6423 vector bool int vec_cmpeq (vector float, vector float);
6425 vector bool int vec_vcmpeqfp (vector float, vector float);
6427 vector bool int vec_vcmpequw (vector signed int, vector signed int);
6428 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
6430 vector bool short vec_vcmpequh (vector signed short,
6431 vector signed short);
6432 vector bool short vec_vcmpequh (vector unsigned short,
6433 vector unsigned short);
6435 vector bool char vec_vcmpequb (vector signed char, vector signed char);
6436 vector bool char vec_vcmpequb (vector unsigned char,
6437 vector unsigned char);
6439 vector bool int vec_cmpge (vector float, vector float);
6441 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
6442 vector bool char vec_cmpgt (vector signed char, vector signed char);
6443 vector bool short vec_cmpgt (vector unsigned short,
6444 vector unsigned short);
6445 vector bool short vec_cmpgt (vector signed short, vector signed short);
6446 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
6447 vector bool int vec_cmpgt (vector signed int, vector signed int);
6448 vector bool int vec_cmpgt (vector float, vector float);
6450 vector bool int vec_vcmpgtfp (vector float, vector float);
6452 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
6454 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
6456 vector bool short vec_vcmpgtsh (vector signed short,
6457 vector signed short);
6459 vector bool short vec_vcmpgtuh (vector unsigned short,
6460 vector unsigned short);
6462 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
6464 vector bool char vec_vcmpgtub (vector unsigned char,
6465 vector unsigned char);
6467 vector bool int vec_cmple (vector float, vector float);
6469 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
6470 vector bool char vec_cmplt (vector signed char, vector signed char);
6471 vector bool short vec_cmplt (vector unsigned short,
6472 vector unsigned short);
6473 vector bool short vec_cmplt (vector signed short, vector signed short);
6474 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
6475 vector bool int vec_cmplt (vector signed int, vector signed int);
6476 vector bool int vec_cmplt (vector float, vector float);
6478 vector float vec_ctf (vector unsigned int, const int);
6479 vector float vec_ctf (vector signed int, const int);
6481 vector float vec_vcfsx (vector signed int, const int);
6483 vector float vec_vcfux (vector unsigned int, const int);
6485 vector signed int vec_cts (vector float, const int);
6487 vector unsigned int vec_ctu (vector float, const int);
6489 void vec_dss (const int);
6491 void vec_dssall (void);
6493 void vec_dst (const vector unsigned char *, int, const int);
6494 void vec_dst (const vector signed char *, int, const int);
6495 void vec_dst (const vector bool char *, int, const int);
6496 void vec_dst (const vector unsigned short *, int, const int);
6497 void vec_dst (const vector signed short *, int, const int);
6498 void vec_dst (const vector bool short *, int, const int);
6499 void vec_dst (const vector pixel *, int, const int);
6500 void vec_dst (const vector unsigned int *, int, const int);
6501 void vec_dst (const vector signed int *, int, const int);
6502 void vec_dst (const vector bool int *, int, const int);
6503 void vec_dst (const vector float *, int, const int);
6504 void vec_dst (const unsigned char *, int, const int);
6505 void vec_dst (const signed char *, int, const int);
6506 void vec_dst (const unsigned short *, int, const int);
6507 void vec_dst (const short *, int, const int);
6508 void vec_dst (const unsigned int *, int, const int);
6509 void vec_dst (const int *, int, const int);
6510 void vec_dst (const unsigned long *, int, const int);
6511 void vec_dst (const long *, int, const int);
6512 void vec_dst (const float *, int, const int);
6514 void vec_dstst (const vector unsigned char *, int, const int);
6515 void vec_dstst (const vector signed char *, int, const int);
6516 void vec_dstst (const vector bool char *, int, const int);
6517 void vec_dstst (const vector unsigned short *, int, const int);
6518 void vec_dstst (const vector signed short *, int, const int);
6519 void vec_dstst (const vector bool short *, int, const int);
6520 void vec_dstst (const vector pixel *, int, const int);
6521 void vec_dstst (const vector unsigned int *, int, const int);
6522 void vec_dstst (const vector signed int *, int, const int);
6523 void vec_dstst (const vector bool int *, int, const int);
6524 void vec_dstst (const vector float *, int, const int);
6525 void vec_dstst (const unsigned char *, int, const int);
6526 void vec_dstst (const signed char *, int, const int);
6527 void vec_dstst (const unsigned short *, int, const int);
6528 void vec_dstst (const short *, int, const int);
6529 void vec_dstst (const unsigned int *, int, const int);
6530 void vec_dstst (const int *, int, const int);
6531 void vec_dstst (const unsigned long *, int, const int);
6532 void vec_dstst (const long *, int, const int);
6533 void vec_dstst (const float *, int, const int);
6535 void vec_dststt (const vector unsigned char *, int, const int);
6536 void vec_dststt (const vector signed char *, int, const int);
6537 void vec_dststt (const vector bool char *, int, const int);
6538 void vec_dststt (const vector unsigned short *, int, const int);
6539 void vec_dststt (const vector signed short *, int, const int);
6540 void vec_dststt (const vector bool short *, int, const int);
6541 void vec_dststt (const vector pixel *, int, const int);
6542 void vec_dststt (const vector unsigned int *, int, const int);
6543 void vec_dststt (const vector signed int *, int, const int);
6544 void vec_dststt (const vector bool int *, int, const int);
6545 void vec_dststt (const vector float *, int, const int);
6546 void vec_dststt (const unsigned char *, int, const int);
6547 void vec_dststt (const signed char *, int, const int);
6548 void vec_dststt (const unsigned short *, int, const int);
6549 void vec_dststt (const short *, int, const int);
6550 void vec_dststt (const unsigned int *, int, const int);
6551 void vec_dststt (const int *, int, const int);
6552 void vec_dststt (const unsigned long *, int, const int);
6553 void vec_dststt (const long *, int, const int);
6554 void vec_dststt (const float *, int, const int);
6556 void vec_dstt (const vector unsigned char *, int, const int);
6557 void vec_dstt (const vector signed char *, int, const int);
6558 void vec_dstt (const vector bool char *, int, const int);
6559 void vec_dstt (const vector unsigned short *, int, const int);
6560 void vec_dstt (const vector signed short *, int, const int);
6561 void vec_dstt (const vector bool short *, int, const int);
6562 void vec_dstt (const vector pixel *, int, const int);
6563 void vec_dstt (const vector unsigned int *, int, const int);
6564 void vec_dstt (const vector signed int *, int, const int);
6565 void vec_dstt (const vector bool int *, int, const int);
6566 void vec_dstt (const vector float *, int, const int);
6567 void vec_dstt (const unsigned char *, int, const int);
6568 void vec_dstt (const signed char *, int, const int);
6569 void vec_dstt (const unsigned short *, int, const int);
6570 void vec_dstt (const short *, int, const int);
6571 void vec_dstt (const unsigned int *, int, const int);
6572 void vec_dstt (const int *, int, const int);
6573 void vec_dstt (const unsigned long *, int, const int);
6574 void vec_dstt (const long *, int, const int);
6575 void vec_dstt (const float *, int, const int);
6577 vector float vec_expte (vector float);
6579 vector float vec_floor (vector float);
6581 vector float vec_ld (int, const vector float *);
6582 vector float vec_ld (int, const float *);
6583 vector bool int vec_ld (int, const vector bool int *);
6584 vector signed int vec_ld (int, const vector signed int *);
6585 vector signed int vec_ld (int, const int *);
6586 vector signed int vec_ld (int, const long *);
6587 vector unsigned int vec_ld (int, const vector unsigned int *);
6588 vector unsigned int vec_ld (int, const unsigned int *);
6589 vector unsigned int vec_ld (int, const unsigned long *);
6590 vector bool short vec_ld (int, const vector bool short *);
6591 vector pixel vec_ld (int, const vector pixel *);
6592 vector signed short vec_ld (int, const vector signed short *);
6593 vector signed short vec_ld (int, const short *);
6594 vector unsigned short vec_ld (int, const vector unsigned short *);
6595 vector unsigned short vec_ld (int, const unsigned short *);
6596 vector bool char vec_ld (int, const vector bool char *);
6597 vector signed char vec_ld (int, const vector signed char *);
6598 vector signed char vec_ld (int, const signed char *);
6599 vector unsigned char vec_ld (int, const vector unsigned char *);
6600 vector unsigned char vec_ld (int, const unsigned char *);
6602 vector signed char vec_lde (int, const signed char *);
6603 vector unsigned char vec_lde (int, const unsigned char *);
6604 vector signed short vec_lde (int, const short *);
6605 vector unsigned short vec_lde (int, const unsigned short *);
6606 vector float vec_lde (int, const float *);
6607 vector signed int vec_lde (int, const int *);
6608 vector unsigned int vec_lde (int, const unsigned int *);
6609 vector signed int vec_lde (int, const long *);
6610 vector unsigned int vec_lde (int, const unsigned long *);
6612 vector float vec_lvewx (int, float *);
6613 vector signed int vec_lvewx (int, int *);
6614 vector unsigned int vec_lvewx (int, unsigned int *);
6615 vector signed int vec_lvewx (int, long *);
6616 vector unsigned int vec_lvewx (int, unsigned long *);
6618 vector signed short vec_lvehx (int, short *);
6619 vector unsigned short vec_lvehx (int, unsigned short *);
6621 vector signed char vec_lvebx (int, char *);
6622 vector unsigned char vec_lvebx (int, unsigned char *);
6624 vector float vec_ldl (int, const vector float *);
6625 vector float vec_ldl (int, const float *);
6626 vector bool int vec_ldl (int, const vector bool int *);
6627 vector signed int vec_ldl (int, const vector signed int *);
6628 vector signed int vec_ldl (int, const int *);
6629 vector signed int vec_ldl (int, const long *);
6630 vector unsigned int vec_ldl (int, const vector unsigned int *);
6631 vector unsigned int vec_ldl (int, const unsigned int *);
6632 vector unsigned int vec_ldl (int, const unsigned long *);
6633 vector bool short vec_ldl (int, const vector bool short *);
6634 vector pixel vec_ldl (int, const vector pixel *);
6635 vector signed short vec_ldl (int, const vector signed short *);
6636 vector signed short vec_ldl (int, const short *);
6637 vector unsigned short vec_ldl (int, const vector unsigned short *);
6638 vector unsigned short vec_ldl (int, const unsigned short *);
6639 vector bool char vec_ldl (int, const vector bool char *);
6640 vector signed char vec_ldl (int, const vector signed char *);
6641 vector signed char vec_ldl (int, const signed char *);
6642 vector unsigned char vec_ldl (int, const vector unsigned char *);
6643 vector unsigned char vec_ldl (int, const unsigned char *);
6645 vector float vec_loge (vector float);
6647 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
6648 vector unsigned char vec_lvsl (int, const volatile signed char *);
6649 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
6650 vector unsigned char vec_lvsl (int, const volatile short *);
6651 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
6652 vector unsigned char vec_lvsl (int, const volatile int *);
6653 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
6654 vector unsigned char vec_lvsl (int, const volatile long *);
6655 vector unsigned char vec_lvsl (int, const volatile float *);
6657 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
6658 vector unsigned char vec_lvsr (int, const volatile signed char *);
6659 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
6660 vector unsigned char vec_lvsr (int, const volatile short *);
6661 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
6662 vector unsigned char vec_lvsr (int, const volatile int *);
6663 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
6664 vector unsigned char vec_lvsr (int, const volatile long *);
6665 vector unsigned char vec_lvsr (int, const volatile float *);
6667 vector float vec_madd (vector float, vector float, vector float);
6669 vector signed short vec_madds (vector signed short,
6670 vector signed short,
6671 vector signed short);
6673 vector unsigned char vec_max (vector bool char, vector unsigned char);
6674 vector unsigned char vec_max (vector unsigned char, vector bool char);
6675 vector unsigned char vec_max (vector unsigned char,
6676 vector unsigned char);
6677 vector signed char vec_max (vector bool char, vector signed char);
6678 vector signed char vec_max (vector signed char, vector bool char);
6679 vector signed char vec_max (vector signed char, vector signed char);
6680 vector unsigned short vec_max (vector bool short,
6681 vector unsigned short);
6682 vector unsigned short vec_max (vector unsigned short,
6684 vector unsigned short vec_max (vector unsigned short,
6685 vector unsigned short);
6686 vector signed short vec_max (vector bool short, vector signed short);
6687 vector signed short vec_max (vector signed short, vector bool short);
6688 vector signed short vec_max (vector signed short, vector signed short);
6689 vector unsigned int vec_max (vector bool int, vector unsigned int);
6690 vector unsigned int vec_max (vector unsigned int, vector bool int);
6691 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
6692 vector signed int vec_max (vector bool int, vector signed int);
6693 vector signed int vec_max (vector signed int, vector bool int);
6694 vector signed int vec_max (vector signed int, vector signed int);
6695 vector float vec_max (vector float, vector float);
6697 vector float vec_vmaxfp (vector float, vector float);
6699 vector signed int vec_vmaxsw (vector bool int, vector signed int);
6700 vector signed int vec_vmaxsw (vector signed int, vector bool int);
6701 vector signed int vec_vmaxsw (vector signed int, vector signed int);
6703 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
6704 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
6705 vector unsigned int vec_vmaxuw (vector unsigned int,
6706 vector unsigned int);
6708 vector signed short vec_vmaxsh (vector bool short, vector signed short);
6709 vector signed short vec_vmaxsh (vector signed short, vector bool short);
6710 vector signed short vec_vmaxsh (vector signed short,
6711 vector signed short);
6713 vector unsigned short vec_vmaxuh (vector bool short,
6714 vector unsigned short);
6715 vector unsigned short vec_vmaxuh (vector unsigned short,
6717 vector unsigned short vec_vmaxuh (vector unsigned short,
6718 vector unsigned short);
6720 vector signed char vec_vmaxsb (vector bool char, vector signed char);
6721 vector signed char vec_vmaxsb (vector signed char, vector bool char);
6722 vector signed char vec_vmaxsb (vector signed char, vector signed char);
6724 vector unsigned char vec_vmaxub (vector bool char,
6725 vector unsigned char);
6726 vector unsigned char vec_vmaxub (vector unsigned char,
6728 vector unsigned char vec_vmaxub (vector unsigned char,
6729 vector unsigned char);
6731 vector bool char vec_mergeh (vector bool char, vector bool char);
6732 vector signed char vec_mergeh (vector signed char, vector signed char);
6733 vector unsigned char vec_mergeh (vector unsigned char,
6734 vector unsigned char);
6735 vector bool short vec_mergeh (vector bool short, vector bool short);
6736 vector pixel vec_mergeh (vector pixel, vector pixel);
6737 vector signed short vec_mergeh (vector signed short,
6738 vector signed short);
6739 vector unsigned short vec_mergeh (vector unsigned short,
6740 vector unsigned short);
6741 vector float vec_mergeh (vector float, vector float);
6742 vector bool int vec_mergeh (vector bool int, vector bool int);
6743 vector signed int vec_mergeh (vector signed int, vector signed int);
6744 vector unsigned int vec_mergeh (vector unsigned int,
6745 vector unsigned int);
6747 vector float vec_vmrghw (vector float, vector float);
6748 vector bool int vec_vmrghw (vector bool int, vector bool int);
6749 vector signed int vec_vmrghw (vector signed int, vector signed int);
6750 vector unsigned int vec_vmrghw (vector unsigned int,
6751 vector unsigned int);
6753 vector bool short vec_vmrghh (vector bool short, vector bool short);
6754 vector signed short vec_vmrghh (vector signed short,
6755 vector signed short);
6756 vector unsigned short vec_vmrghh (vector unsigned short,
6757 vector unsigned short);
6758 vector pixel vec_vmrghh (vector pixel, vector pixel);
6760 vector bool char vec_vmrghb (vector bool char, vector bool char);
6761 vector signed char vec_vmrghb (vector signed char, vector signed char);
6762 vector unsigned char vec_vmrghb (vector unsigned char,
6763 vector unsigned char);
6765 vector bool char vec_mergel (vector bool char, vector bool char);
6766 vector signed char vec_mergel (vector signed char, vector signed char);
6767 vector unsigned char vec_mergel (vector unsigned char,
6768 vector unsigned char);
6769 vector bool short vec_mergel (vector bool short, vector bool short);
6770 vector pixel vec_mergel (vector pixel, vector pixel);
6771 vector signed short vec_mergel (vector signed short,
6772 vector signed short);
6773 vector unsigned short vec_mergel (vector unsigned short,
6774 vector unsigned short);
6775 vector float vec_mergel (vector float, vector float);
6776 vector bool int vec_mergel (vector bool int, vector bool int);
6777 vector signed int vec_mergel (vector signed int, vector signed int);
6778 vector unsigned int vec_mergel (vector unsigned int,
6779 vector unsigned int);
6781 vector float vec_vmrglw (vector float, vector float);
6782 vector signed int vec_vmrglw (vector signed int, vector signed int);
6783 vector unsigned int vec_vmrglw (vector unsigned int,
6784 vector unsigned int);
6785 vector bool int vec_vmrglw (vector bool int, vector bool int);
6787 vector bool short vec_vmrglh (vector bool short, vector bool short);
6788 vector signed short vec_vmrglh (vector signed short,
6789 vector signed short);
6790 vector unsigned short vec_vmrglh (vector unsigned short,
6791 vector unsigned short);
6792 vector pixel vec_vmrglh (vector pixel, vector pixel);
6794 vector bool char vec_vmrglb (vector bool char, vector bool char);
6795 vector signed char vec_vmrglb (vector signed char, vector signed char);
6796 vector unsigned char vec_vmrglb (vector unsigned char,
6797 vector unsigned char);
6799 vector unsigned short vec_mfvscr (void);
6801 vector unsigned char vec_min (vector bool char, vector unsigned char);
6802 vector unsigned char vec_min (vector unsigned char, vector bool char);
6803 vector unsigned char vec_min (vector unsigned char,
6804 vector unsigned char);
6805 vector signed char vec_min (vector bool char, vector signed char);
6806 vector signed char vec_min (vector signed char, vector bool char);
6807 vector signed char vec_min (vector signed char, vector signed char);
6808 vector unsigned short vec_min (vector bool short,
6809 vector unsigned short);
6810 vector unsigned short vec_min (vector unsigned short,
6812 vector unsigned short vec_min (vector unsigned short,
6813 vector unsigned short);
6814 vector signed short vec_min (vector bool short, vector signed short);
6815 vector signed short vec_min (vector signed short, vector bool short);
6816 vector signed short vec_min (vector signed short, vector signed short);
6817 vector unsigned int vec_min (vector bool int, vector unsigned int);
6818 vector unsigned int vec_min (vector unsigned int, vector bool int);
6819 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
6820 vector signed int vec_min (vector bool int, vector signed int);
6821 vector signed int vec_min (vector signed int, vector bool int);
6822 vector signed int vec_min (vector signed int, vector signed int);
6823 vector float vec_min (vector float, vector float);
6825 vector float vec_vminfp (vector float, vector float);
6827 vector signed int vec_vminsw (vector bool int, vector signed int);
6828 vector signed int vec_vminsw (vector signed int, vector bool int);
6829 vector signed int vec_vminsw (vector signed int, vector signed int);
6831 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
6832 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
6833 vector unsigned int vec_vminuw (vector unsigned int,
6834 vector unsigned int);
6836 vector signed short vec_vminsh (vector bool short, vector signed short);
6837 vector signed short vec_vminsh (vector signed short, vector bool short);
6838 vector signed short vec_vminsh (vector signed short,
6839 vector signed short);
6841 vector unsigned short vec_vminuh (vector bool short,
6842 vector unsigned short);
6843 vector unsigned short vec_vminuh (vector unsigned short,
6845 vector unsigned short vec_vminuh (vector unsigned short,
6846 vector unsigned short);
6848 vector signed char vec_vminsb (vector bool char, vector signed char);
6849 vector signed char vec_vminsb (vector signed char, vector bool char);
6850 vector signed char vec_vminsb (vector signed char, vector signed char);
6852 vector unsigned char vec_vminub (vector bool char,
6853 vector unsigned char);
6854 vector unsigned char vec_vminub (vector unsigned char,
6856 vector unsigned char vec_vminub (vector unsigned char,
6857 vector unsigned char);
6859 vector signed short vec_mladd (vector signed short,
6860 vector signed short,
6861 vector signed short);
6862 vector signed short vec_mladd (vector signed short,
6863 vector unsigned short,
6864 vector unsigned short);
6865 vector signed short vec_mladd (vector unsigned short,
6866 vector signed short,
6867 vector signed short);
6868 vector unsigned short vec_mladd (vector unsigned short,
6869 vector unsigned short,
6870 vector unsigned short);
6872 vector signed short vec_mradds (vector signed short,
6873 vector signed short,
6874 vector signed short);
6876 vector unsigned int vec_msum (vector unsigned char,
6877 vector unsigned char,
6878 vector unsigned int);
6879 vector signed int vec_msum (vector signed char,
6880 vector unsigned char,
6882 vector unsigned int vec_msum (vector unsigned short,
6883 vector unsigned short,
6884 vector unsigned int);
6885 vector signed int vec_msum (vector signed short,
6886 vector signed short,
6889 vector signed int vec_vmsumshm (vector signed short,
6890 vector signed short,
6893 vector unsigned int vec_vmsumuhm (vector unsigned short,
6894 vector unsigned short,
6895 vector unsigned int);
6897 vector signed int vec_vmsummbm (vector signed char,
6898 vector unsigned char,
6901 vector unsigned int vec_vmsumubm (vector unsigned char,
6902 vector unsigned char,
6903 vector unsigned int);
6905 vector unsigned int vec_msums (vector unsigned short,
6906 vector unsigned short,
6907 vector unsigned int);
6908 vector signed int vec_msums (vector signed short,
6909 vector signed short,
6912 vector signed int vec_vmsumshs (vector signed short,
6913 vector signed short,
6916 vector unsigned int vec_vmsumuhs (vector unsigned short,
6917 vector unsigned short,
6918 vector unsigned int);
6920 void vec_mtvscr (vector signed int);
6921 void vec_mtvscr (vector unsigned int);
6922 void vec_mtvscr (vector bool int);
6923 void vec_mtvscr (vector signed short);
6924 void vec_mtvscr (vector unsigned short);
6925 void vec_mtvscr (vector bool short);
6926 void vec_mtvscr (vector pixel);
6927 void vec_mtvscr (vector signed char);
6928 void vec_mtvscr (vector unsigned char);
6929 void vec_mtvscr (vector bool char);
6931 vector unsigned short vec_mule (vector unsigned char,
6932 vector unsigned char);
6933 vector signed short vec_mule (vector signed char,
6934 vector signed char);
6935 vector unsigned int vec_mule (vector unsigned short,
6936 vector unsigned short);
6937 vector signed int vec_mule (vector signed short, vector signed short);
6939 vector signed int vec_vmulesh (vector signed short,
6940 vector signed short);
6942 vector unsigned int vec_vmuleuh (vector unsigned short,
6943 vector unsigned short);
6945 vector signed short vec_vmulesb (vector signed char,
6946 vector signed char);
6948 vector unsigned short vec_vmuleub (vector unsigned char,
6949 vector unsigned char);
6951 vector unsigned short vec_mulo (vector unsigned char,
6952 vector unsigned char);
6953 vector signed short vec_mulo (vector signed char, vector signed char);
6954 vector unsigned int vec_mulo (vector unsigned short,
6955 vector unsigned short);
6956 vector signed int vec_mulo (vector signed short, vector signed short);
6958 vector signed int vec_vmulosh (vector signed short,
6959 vector signed short);
6961 vector unsigned int vec_vmulouh (vector unsigned short,
6962 vector unsigned short);
6964 vector signed short vec_vmulosb (vector signed char,
6965 vector signed char);
6967 vector unsigned short vec_vmuloub (vector unsigned char,
6968 vector unsigned char);
6970 vector float vec_nmsub (vector float, vector float, vector float);
6972 vector float vec_nor (vector float, vector float);
6973 vector signed int vec_nor (vector signed int, vector signed int);
6974 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
6975 vector bool int vec_nor (vector bool int, vector bool int);
6976 vector signed short vec_nor (vector signed short, vector signed short);
6977 vector unsigned short vec_nor (vector unsigned short,
6978 vector unsigned short);
6979 vector bool short vec_nor (vector bool short, vector bool short);
6980 vector signed char vec_nor (vector signed char, vector signed char);
6981 vector unsigned char vec_nor (vector unsigned char,
6982 vector unsigned char);
6983 vector bool char vec_nor (vector bool char, vector bool char);
6985 vector float vec_or (vector float, vector float);
6986 vector float vec_or (vector float, vector bool int);
6987 vector float vec_or (vector bool int, vector float);
6988 vector bool int vec_or (vector bool int, vector bool int);
6989 vector signed int vec_or (vector bool int, vector signed int);
6990 vector signed int vec_or (vector signed int, vector bool int);
6991 vector signed int vec_or (vector signed int, vector signed int);
6992 vector unsigned int vec_or (vector bool int, vector unsigned int);
6993 vector unsigned int vec_or (vector unsigned int, vector bool int);
6994 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
6995 vector bool short vec_or (vector bool short, vector bool short);
6996 vector signed short vec_or (vector bool short, vector signed short);
6997 vector signed short vec_or (vector signed short, vector bool short);
6998 vector signed short vec_or (vector signed short, vector signed short);
6999 vector unsigned short vec_or (vector bool short, vector unsigned short);
7000 vector unsigned short vec_or (vector unsigned short, vector bool short);
7001 vector unsigned short vec_or (vector unsigned short,
7002 vector unsigned short);
7003 vector signed char vec_or (vector bool char, vector signed char);
7004 vector bool char vec_or (vector bool char, vector bool char);
7005 vector signed char vec_or (vector signed char, vector bool char);
7006 vector signed char vec_or (vector signed char, vector signed char);
7007 vector unsigned char vec_or (vector bool char, vector unsigned char);
7008 vector unsigned char vec_or (vector unsigned char, vector bool char);
7009 vector unsigned char vec_or (vector unsigned char,
7010 vector unsigned char);
7012 vector signed char vec_pack (vector signed short, vector signed short);
7013 vector unsigned char vec_pack (vector unsigned short,
7014 vector unsigned short);
7015 vector bool char vec_pack (vector bool short, vector bool short);
7016 vector signed short vec_pack (vector signed int, vector signed int);
7017 vector unsigned short vec_pack (vector unsigned int,
7018 vector unsigned int);
7019 vector bool short vec_pack (vector bool int, vector bool int);
7021 vector bool short vec_vpkuwum (vector bool int, vector bool int);
7022 vector signed short vec_vpkuwum (vector signed int, vector signed int);
7023 vector unsigned short vec_vpkuwum (vector unsigned int,
7024 vector unsigned int);
7026 vector bool char vec_vpkuhum (vector bool short, vector bool short);
7027 vector signed char vec_vpkuhum (vector signed short,
7028 vector signed short);
7029 vector unsigned char vec_vpkuhum (vector unsigned short,
7030 vector unsigned short);
7032 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
7034 vector unsigned char vec_packs (vector unsigned short,
7035 vector unsigned short);
7036 vector signed char vec_packs (vector signed short, vector signed short);
7037 vector unsigned short vec_packs (vector unsigned int,
7038 vector unsigned int);
7039 vector signed short vec_packs (vector signed int, vector signed int);
7041 vector signed short vec_vpkswss (vector signed int, vector signed int);
7043 vector unsigned short vec_vpkuwus (vector unsigned int,
7044 vector unsigned int);
7046 vector signed char vec_vpkshss (vector signed short,
7047 vector signed short);
7049 vector unsigned char vec_vpkuhus (vector unsigned short,
7050 vector unsigned short);
7052 vector unsigned char vec_packsu (vector unsigned short,
7053 vector unsigned short);
7054 vector unsigned char vec_packsu (vector signed short,
7055 vector signed short);
7056 vector unsigned short vec_packsu (vector unsigned int,
7057 vector unsigned int);
7058 vector unsigned short vec_packsu (vector signed int, vector signed int);
7060 vector unsigned short vec_vpkswus (vector signed int,
7063 vector unsigned char vec_vpkshus (vector signed short,
7064 vector signed short);
7066 vector float vec_perm (vector float,
7068 vector unsigned char);
7069 vector signed int vec_perm (vector signed int,
7071 vector unsigned char);
7072 vector unsigned int vec_perm (vector unsigned int,
7073 vector unsigned int,
7074 vector unsigned char);
7075 vector bool int vec_perm (vector bool int,
7077 vector unsigned char);
7078 vector signed short vec_perm (vector signed short,
7079 vector signed short,
7080 vector unsigned char);
7081 vector unsigned short vec_perm (vector unsigned short,
7082 vector unsigned short,
7083 vector unsigned char);
7084 vector bool short vec_perm (vector bool short,
7086 vector unsigned char);
7087 vector pixel vec_perm (vector pixel,
7089 vector unsigned char);
7090 vector signed char vec_perm (vector signed char,
7092 vector unsigned char);
7093 vector unsigned char vec_perm (vector unsigned char,
7094 vector unsigned char,
7095 vector unsigned char);
7096 vector bool char vec_perm (vector bool char,
7098 vector unsigned char);
7100 vector float vec_re (vector float);
7102 vector signed char vec_rl (vector signed char,
7103 vector unsigned char);
7104 vector unsigned char vec_rl (vector unsigned char,
7105 vector unsigned char);
7106 vector signed short vec_rl (vector signed short, vector unsigned short);
7107 vector unsigned short vec_rl (vector unsigned short,
7108 vector unsigned short);
7109 vector signed int vec_rl (vector signed int, vector unsigned int);
7110 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
7112 vector signed int vec_vrlw (vector signed int, vector unsigned int);
7113 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
7115 vector signed short vec_vrlh (vector signed short,
7116 vector unsigned short);
7117 vector unsigned short vec_vrlh (vector unsigned short,
7118 vector unsigned short);
7120 vector signed char vec_vrlb (vector signed char, vector unsigned char);
7121 vector unsigned char vec_vrlb (vector unsigned char,
7122 vector unsigned char);
7124 vector float vec_round (vector float);
7126 vector float vec_rsqrte (vector float);
7128 vector float vec_sel (vector float, vector float, vector bool int);
7129 vector float vec_sel (vector float, vector float, vector unsigned int);
7130 vector signed int vec_sel (vector signed int,
7133 vector signed int vec_sel (vector signed int,
7135 vector unsigned int);
7136 vector unsigned int vec_sel (vector unsigned int,
7137 vector unsigned int,
7139 vector unsigned int vec_sel (vector unsigned int,
7140 vector unsigned int,
7141 vector unsigned int);
7142 vector bool int vec_sel (vector bool int,
7145 vector bool int vec_sel (vector bool int,
7147 vector unsigned int);
7148 vector signed short vec_sel (vector signed short,
7149 vector signed short,
7151 vector signed short vec_sel (vector signed short,
7152 vector signed short,
7153 vector unsigned short);
7154 vector unsigned short vec_sel (vector unsigned short,
7155 vector unsigned short,
7157 vector unsigned short vec_sel (vector unsigned short,
7158 vector unsigned short,
7159 vector unsigned short);
7160 vector bool short vec_sel (vector bool short,
7163 vector bool short vec_sel (vector bool short,
7165 vector unsigned short);
7166 vector signed char vec_sel (vector signed char,
7169 vector signed char vec_sel (vector signed char,
7171 vector unsigned char);
7172 vector unsigned char vec_sel (vector unsigned char,
7173 vector unsigned char,
7175 vector unsigned char vec_sel (vector unsigned char,
7176 vector unsigned char,
7177 vector unsigned char);
7178 vector bool char vec_sel (vector bool char,
7181 vector bool char vec_sel (vector bool char,
7183 vector unsigned char);
7185 vector signed char vec_sl (vector signed char,
7186 vector unsigned char);
7187 vector unsigned char vec_sl (vector unsigned char,
7188 vector unsigned char);
7189 vector signed short vec_sl (vector signed short, vector unsigned short);
7190 vector unsigned short vec_sl (vector unsigned short,
7191 vector unsigned short);
7192 vector signed int vec_sl (vector signed int, vector unsigned int);
7193 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
7195 vector signed int vec_vslw (vector signed int, vector unsigned int);
7196 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
7198 vector signed short vec_vslh (vector signed short,
7199 vector unsigned short);
7200 vector unsigned short vec_vslh (vector unsigned short,
7201 vector unsigned short);
7203 vector signed char vec_vslb (vector signed char, vector unsigned char);
7204 vector unsigned char vec_vslb (vector unsigned char,
7205 vector unsigned char);
7207 vector float vec_sld (vector float, vector float, const int);
7208 vector signed int vec_sld (vector signed int,
7211 vector unsigned int vec_sld (vector unsigned int,
7212 vector unsigned int,
7214 vector bool int vec_sld (vector bool int,
7217 vector signed short vec_sld (vector signed short,
7218 vector signed short,
7220 vector unsigned short vec_sld (vector unsigned short,
7221 vector unsigned short,
7223 vector bool short vec_sld (vector bool short,
7226 vector pixel vec_sld (vector pixel,
7229 vector signed char vec_sld (vector signed char,
7232 vector unsigned char vec_sld (vector unsigned char,
7233 vector unsigned char,
7235 vector bool char vec_sld (vector bool char,
7239 vector signed int vec_sll (vector signed int,
7240 vector unsigned int);
7241 vector signed int vec_sll (vector signed int,
7242 vector unsigned short);
7243 vector signed int vec_sll (vector signed int,
7244 vector unsigned char);
7245 vector unsigned int vec_sll (vector unsigned int,
7246 vector unsigned int);
7247 vector unsigned int vec_sll (vector unsigned int,
7248 vector unsigned short);
7249 vector unsigned int vec_sll (vector unsigned int,
7250 vector unsigned char);
7251 vector bool int vec_sll (vector bool int,
7252 vector unsigned int);
7253 vector bool int vec_sll (vector bool int,
7254 vector unsigned short);
7255 vector bool int vec_sll (vector bool int,
7256 vector unsigned char);
7257 vector signed short vec_sll (vector signed short,
7258 vector unsigned int);
7259 vector signed short vec_sll (vector signed short,
7260 vector unsigned short);
7261 vector signed short vec_sll (vector signed short,
7262 vector unsigned char);
7263 vector unsigned short vec_sll (vector unsigned short,
7264 vector unsigned int);
7265 vector unsigned short vec_sll (vector unsigned short,
7266 vector unsigned short);
7267 vector unsigned short vec_sll (vector unsigned short,
7268 vector unsigned char);
7269 vector bool short vec_sll (vector bool short, vector unsigned int);
7270 vector bool short vec_sll (vector bool short, vector unsigned short);
7271 vector bool short vec_sll (vector bool short, vector unsigned char);
7272 vector pixel vec_sll (vector pixel, vector unsigned int);
7273 vector pixel vec_sll (vector pixel, vector unsigned short);
7274 vector pixel vec_sll (vector pixel, vector unsigned char);
7275 vector signed char vec_sll (vector signed char, vector unsigned int);
7276 vector signed char vec_sll (vector signed char, vector unsigned short);
7277 vector signed char vec_sll (vector signed char, vector unsigned char);
7278 vector unsigned char vec_sll (vector unsigned char,
7279 vector unsigned int);
7280 vector unsigned char vec_sll (vector unsigned char,
7281 vector unsigned short);
7282 vector unsigned char vec_sll (vector unsigned char,
7283 vector unsigned char);
7284 vector bool char vec_sll (vector bool char, vector unsigned int);
7285 vector bool char vec_sll (vector bool char, vector unsigned short);
7286 vector bool char vec_sll (vector bool char, vector unsigned char);
7288 vector float vec_slo (vector float, vector signed char);
7289 vector float vec_slo (vector float, vector unsigned char);
7290 vector signed int vec_slo (vector signed int, vector signed char);
7291 vector signed int vec_slo (vector signed int, vector unsigned char);
7292 vector unsigned int vec_slo (vector unsigned int, vector signed char);
7293 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
7294 vector signed short vec_slo (vector signed short, vector signed char);
7295 vector signed short vec_slo (vector signed short, vector unsigned char);
7296 vector unsigned short vec_slo (vector unsigned short,
7297 vector signed char);
7298 vector unsigned short vec_slo (vector unsigned short,
7299 vector unsigned char);
7300 vector pixel vec_slo (vector pixel, vector signed char);
7301 vector pixel vec_slo (vector pixel, vector unsigned char);
7302 vector signed char vec_slo (vector signed char, vector signed char);
7303 vector signed char vec_slo (vector signed char, vector unsigned char);
7304 vector unsigned char vec_slo (vector unsigned char, vector signed char);
7305 vector unsigned char vec_slo (vector unsigned char,
7306 vector unsigned char);
7308 vector signed char vec_splat (vector signed char, const int);
7309 vector unsigned char vec_splat (vector unsigned char, const int);
7310 vector bool char vec_splat (vector bool char, const int);
7311 vector signed short vec_splat (vector signed short, const int);
7312 vector unsigned short vec_splat (vector unsigned short, const int);
7313 vector bool short vec_splat (vector bool short, const int);
7314 vector pixel vec_splat (vector pixel, const int);
7315 vector float vec_splat (vector float, const int);
7316 vector signed int vec_splat (vector signed int, const int);
7317 vector unsigned int vec_splat (vector unsigned int, const int);
7318 vector bool int vec_splat (vector bool int, const int);
7320 vector float vec_vspltw (vector float, const int);
7321 vector signed int vec_vspltw (vector signed int, const int);
7322 vector unsigned int vec_vspltw (vector unsigned int, const int);
7323 vector bool int vec_vspltw (vector bool int, const int);
7325 vector bool short vec_vsplth (vector bool short, const int);
7326 vector signed short vec_vsplth (vector signed short, const int);
7327 vector unsigned short vec_vsplth (vector unsigned short, const int);
7328 vector pixel vec_vsplth (vector pixel, const int);
7330 vector signed char vec_vspltb (vector signed char, const int);
7331 vector unsigned char vec_vspltb (vector unsigned char, const int);
7332 vector bool char vec_vspltb (vector bool char, const int);
7334 vector signed char vec_splat_s8 (const int);
7336 vector signed short vec_splat_s16 (const int);
7338 vector signed int vec_splat_s32 (const int);
7340 vector unsigned char vec_splat_u8 (const int);
7342 vector unsigned short vec_splat_u16 (const int);
7344 vector unsigned int vec_splat_u32 (const int);
7346 vector signed char vec_sr (vector signed char, vector unsigned char);
7347 vector unsigned char vec_sr (vector unsigned char,
7348 vector unsigned char);
7349 vector signed short vec_sr (vector signed short,
7350 vector unsigned short);
7351 vector unsigned short vec_sr (vector unsigned short,
7352 vector unsigned short);
7353 vector signed int vec_sr (vector signed int, vector unsigned int);
7354 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
7356 vector signed int vec_vsrw (vector signed int, vector unsigned int);
7357 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
7359 vector signed short vec_vsrh (vector signed short,
7360 vector unsigned short);
7361 vector unsigned short vec_vsrh (vector unsigned short,
7362 vector unsigned short);
7364 vector signed char vec_vsrb (vector signed char, vector unsigned char);
7365 vector unsigned char vec_vsrb (vector unsigned char,
7366 vector unsigned char);
7368 vector signed char vec_sra (vector signed char, vector unsigned char);
7369 vector unsigned char vec_sra (vector unsigned char,
7370 vector unsigned char);
7371 vector signed short vec_sra (vector signed short,
7372 vector unsigned short);
7373 vector unsigned short vec_sra (vector unsigned short,
7374 vector unsigned short);
7375 vector signed int vec_sra (vector signed int, vector unsigned int);
7376 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
7378 vector signed int vec_vsraw (vector signed int, vector unsigned int);
7379 vector unsigned int vec_vsraw (vector unsigned int,
7380 vector unsigned int);
7382 vector signed short vec_vsrah (vector signed short,
7383 vector unsigned short);
7384 vector unsigned short vec_vsrah (vector unsigned short,
7385 vector unsigned short);
7387 vector signed char vec_vsrab (vector signed char, vector unsigned char);
7388 vector unsigned char vec_vsrab (vector unsigned char,
7389 vector unsigned char);
7391 vector signed int vec_srl (vector signed int, vector unsigned int);
7392 vector signed int vec_srl (vector signed int, vector unsigned short);
7393 vector signed int vec_srl (vector signed int, vector unsigned char);
7394 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
7395 vector unsigned int vec_srl (vector unsigned int,
7396 vector unsigned short);
7397 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
7398 vector bool int vec_srl (vector bool int, vector unsigned int);
7399 vector bool int vec_srl (vector bool int, vector unsigned short);
7400 vector bool int vec_srl (vector bool int, vector unsigned char);
7401 vector signed short vec_srl (vector signed short, vector unsigned int);
7402 vector signed short vec_srl (vector signed short,
7403 vector unsigned short);
7404 vector signed short vec_srl (vector signed short, vector unsigned char);
7405 vector unsigned short vec_srl (vector unsigned short,
7406 vector unsigned int);
7407 vector unsigned short vec_srl (vector unsigned short,
7408 vector unsigned short);
7409 vector unsigned short vec_srl (vector unsigned short,
7410 vector unsigned char);
7411 vector bool short vec_srl (vector bool short, vector unsigned int);
7412 vector bool short vec_srl (vector bool short, vector unsigned short);
7413 vector bool short vec_srl (vector bool short, vector unsigned char);
7414 vector pixel vec_srl (vector pixel, vector unsigned int);
7415 vector pixel vec_srl (vector pixel, vector unsigned short);
7416 vector pixel vec_srl (vector pixel, vector unsigned char);
7417 vector signed char vec_srl (vector signed char, vector unsigned int);
7418 vector signed char vec_srl (vector signed char, vector unsigned short);
7419 vector signed char vec_srl (vector signed char, vector unsigned char);
7420 vector unsigned char vec_srl (vector unsigned char,
7421 vector unsigned int);
7422 vector unsigned char vec_srl (vector unsigned char,
7423 vector unsigned short);
7424 vector unsigned char vec_srl (vector unsigned char,
7425 vector unsigned char);
7426 vector bool char vec_srl (vector bool char, vector unsigned int);
7427 vector bool char vec_srl (vector bool char, vector unsigned short);
7428 vector bool char vec_srl (vector bool char, vector unsigned char);
7430 vector float vec_sro (vector float, vector signed char);
7431 vector float vec_sro (vector float, vector unsigned char);
7432 vector signed int vec_sro (vector signed int, vector signed char);
7433 vector signed int vec_sro (vector signed int, vector unsigned char);
7434 vector unsigned int vec_sro (vector unsigned int, vector signed char);
7435 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
7436 vector signed short vec_sro (vector signed short, vector signed char);
7437 vector signed short vec_sro (vector signed short, vector unsigned char);
7438 vector unsigned short vec_sro (vector unsigned short,
7439 vector signed char);
7440 vector unsigned short vec_sro (vector unsigned short,
7441 vector unsigned char);
7442 vector pixel vec_sro (vector pixel, vector signed char);
7443 vector pixel vec_sro (vector pixel, vector unsigned char);
7444 vector signed char vec_sro (vector signed char, vector signed char);
7445 vector signed char vec_sro (vector signed char, vector unsigned char);
7446 vector unsigned char vec_sro (vector unsigned char, vector signed char);
7447 vector unsigned char vec_sro (vector unsigned char,
7448 vector unsigned char);
7450 void vec_st (vector float, int, vector float *);
7451 void vec_st (vector float, int, float *);
7452 void vec_st (vector signed int, int, vector signed int *);
7453 void vec_st (vector signed int, int, int *);
7454 void vec_st (vector unsigned int, int, vector unsigned int *);
7455 void vec_st (vector unsigned int, int, unsigned int *);
7456 void vec_st (vector bool int, int, vector bool int *);
7457 void vec_st (vector bool int, int, unsigned int *);
7458 void vec_st (vector bool int, int, int *);
7459 void vec_st (vector signed short, int, vector signed short *);
7460 void vec_st (vector signed short, int, short *);
7461 void vec_st (vector unsigned short, int, vector unsigned short *);
7462 void vec_st (vector unsigned short, int, unsigned short *);
7463 void vec_st (vector bool short, int, vector bool short *);
7464 void vec_st (vector bool short, int, unsigned short *);
7465 void vec_st (vector pixel, int, vector pixel *);
7466 void vec_st (vector pixel, int, unsigned short *);
7467 void vec_st (vector pixel, int, short *);
7468 void vec_st (vector bool short, int, short *);
7469 void vec_st (vector signed char, int, vector signed char *);
7470 void vec_st (vector signed char, int, signed char *);
7471 void vec_st (vector unsigned char, int, vector unsigned char *);
7472 void vec_st (vector unsigned char, int, unsigned char *);
7473 void vec_st (vector bool char, int, vector bool char *);
7474 void vec_st (vector bool char, int, unsigned char *);
7475 void vec_st (vector bool char, int, signed char *);
7477 void vec_ste (vector signed char, int, signed char *);
7478 void vec_ste (vector unsigned char, int, unsigned char *);
7479 void vec_ste (vector bool char, int, signed char *);
7480 void vec_ste (vector bool char, int, unsigned char *);
7481 void vec_ste (vector signed short, int, short *);
7482 void vec_ste (vector unsigned short, int, unsigned short *);
7483 void vec_ste (vector bool short, int, short *);
7484 void vec_ste (vector bool short, int, unsigned short *);
7485 void vec_ste (vector pixel, int, short *);
7486 void vec_ste (vector pixel, int, unsigned short *);
7487 void vec_ste (vector float, int, float *);
7488 void vec_ste (vector signed int, int, int *);
7489 void vec_ste (vector unsigned int, int, unsigned int *);
7490 void vec_ste (vector bool int, int, int *);
7491 void vec_ste (vector bool int, int, unsigned int *);
7493 void vec_stvewx (vector float, int, float *);
7494 void vec_stvewx (vector signed int, int, int *);
7495 void vec_stvewx (vector unsigned int, int, unsigned int *);
7496 void vec_stvewx (vector bool int, int, int *);
7497 void vec_stvewx (vector bool int, int, unsigned int *);
7499 void vec_stvehx (vector signed short, int, short *);
7500 void vec_stvehx (vector unsigned short, int, unsigned short *);
7501 void vec_stvehx (vector bool short, int, short *);
7502 void vec_stvehx (vector bool short, int, unsigned short *);
7503 void vec_stvehx (vector pixel, int, short *);
7504 void vec_stvehx (vector pixel, int, unsigned short *);
7506 void vec_stvebx (vector signed char, int, signed char *);
7507 void vec_stvebx (vector unsigned char, int, unsigned char *);
7508 void vec_stvebx (vector bool char, int, signed char *);
7509 void vec_stvebx (vector bool char, int, unsigned char *);
7511 void vec_stl (vector float, int, vector float *);
7512 void vec_stl (vector float, int, float *);
7513 void vec_stl (vector signed int, int, vector signed int *);
7514 void vec_stl (vector signed int, int, int *);
7515 void vec_stl (vector unsigned int, int, vector unsigned int *);
7516 void vec_stl (vector unsigned int, int, unsigned int *);
7517 void vec_stl (vector bool int, int, vector bool int *);
7518 void vec_stl (vector bool int, int, unsigned int *);
7519 void vec_stl (vector bool int, int, int *);
7520 void vec_stl (vector signed short, int, vector signed short *);
7521 void vec_stl (vector signed short, int, short *);
7522 void vec_stl (vector unsigned short, int, vector unsigned short *);
7523 void vec_stl (vector unsigned short, int, unsigned short *);
7524 void vec_stl (vector bool short, int, vector bool short *);
7525 void vec_stl (vector bool short, int, unsigned short *);
7526 void vec_stl (vector bool short, int, short *);
7527 void vec_stl (vector pixel, int, vector pixel *);
7528 void vec_stl (vector pixel, int, unsigned short *);
7529 void vec_stl (vector pixel, int, short *);
7530 void vec_stl (vector signed char, int, vector signed char *);
7531 void vec_stl (vector signed char, int, signed char *);
7532 void vec_stl (vector unsigned char, int, vector unsigned char *);
7533 void vec_stl (vector unsigned char, int, unsigned char *);
7534 void vec_stl (vector bool char, int, vector bool char *);
7535 void vec_stl (vector bool char, int, unsigned char *);
7536 void vec_stl (vector bool char, int, signed char *);
7538 vector signed char vec_sub (vector bool char, vector signed char);
7539 vector signed char vec_sub (vector signed char, vector bool char);
7540 vector signed char vec_sub (vector signed char, vector signed char);
7541 vector unsigned char vec_sub (vector bool char, vector unsigned char);
7542 vector unsigned char vec_sub (vector unsigned char, vector bool char);
7543 vector unsigned char vec_sub (vector unsigned char,
7544 vector unsigned char);
7545 vector signed short vec_sub (vector bool short, vector signed short);
7546 vector signed short vec_sub (vector signed short, vector bool short);
7547 vector signed short vec_sub (vector signed short, vector signed short);
7548 vector unsigned short vec_sub (vector bool short,
7549 vector unsigned short);
7550 vector unsigned short vec_sub (vector unsigned short,
7552 vector unsigned short vec_sub (vector unsigned short,
7553 vector unsigned short);
7554 vector signed int vec_sub (vector bool int, vector signed int);
7555 vector signed int vec_sub (vector signed int, vector bool int);
7556 vector signed int vec_sub (vector signed int, vector signed int);
7557 vector unsigned int vec_sub (vector bool int, vector unsigned int);
7558 vector unsigned int vec_sub (vector unsigned int, vector bool int);
7559 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
7560 vector float vec_sub (vector float, vector float);
7562 vector float vec_vsubfp (vector float, vector float);
7564 vector signed int vec_vsubuwm (vector bool int, vector signed int);
7565 vector signed int vec_vsubuwm (vector signed int, vector bool int);
7566 vector signed int vec_vsubuwm (vector signed int, vector signed int);
7567 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
7568 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
7569 vector unsigned int vec_vsubuwm (vector unsigned int,
7570 vector unsigned int);
7572 vector signed short vec_vsubuhm (vector bool short,
7573 vector signed short);
7574 vector signed short vec_vsubuhm (vector signed short,
7576 vector signed short vec_vsubuhm (vector signed short,
7577 vector signed short);
7578 vector unsigned short vec_vsubuhm (vector bool short,
7579 vector unsigned short);
7580 vector unsigned short vec_vsubuhm (vector unsigned short,
7582 vector unsigned short vec_vsubuhm (vector unsigned short,
7583 vector unsigned short);
7585 vector signed char vec_vsububm (vector bool char, vector signed char);
7586 vector signed char vec_vsububm (vector signed char, vector bool char);
7587 vector signed char vec_vsububm (vector signed char, vector signed char);
7588 vector unsigned char vec_vsububm (vector bool char,
7589 vector unsigned char);
7590 vector unsigned char vec_vsububm (vector unsigned char,
7592 vector unsigned char vec_vsububm (vector unsigned char,
7593 vector unsigned char);
7595 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
7597 vector unsigned char vec_subs (vector bool char, vector unsigned char);
7598 vector unsigned char vec_subs (vector unsigned char, vector bool char);
7599 vector unsigned char vec_subs (vector unsigned char,
7600 vector unsigned char);
7601 vector signed char vec_subs (vector bool char, vector signed char);
7602 vector signed char vec_subs (vector signed char, vector bool char);
7603 vector signed char vec_subs (vector signed char, vector signed char);
7604 vector unsigned short vec_subs (vector bool short,
7605 vector unsigned short);
7606 vector unsigned short vec_subs (vector unsigned short,
7608 vector unsigned short vec_subs (vector unsigned short,
7609 vector unsigned short);
7610 vector signed short vec_subs (vector bool short, vector signed short);
7611 vector signed short vec_subs (vector signed short, vector bool short);
7612 vector signed short vec_subs (vector signed short, vector signed short);
7613 vector unsigned int vec_subs (vector bool int, vector unsigned int);
7614 vector unsigned int vec_subs (vector unsigned int, vector bool int);
7615 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
7616 vector signed int vec_subs (vector bool int, vector signed int);
7617 vector signed int vec_subs (vector signed int, vector bool int);
7618 vector signed int vec_subs (vector signed int, vector signed int);
7620 vector signed int vec_vsubsws (vector bool int, vector signed int);
7621 vector signed int vec_vsubsws (vector signed int, vector bool int);
7622 vector signed int vec_vsubsws (vector signed int, vector signed int);
7624 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
7625 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
7626 vector unsigned int vec_vsubuws (vector unsigned int,
7627 vector unsigned int);
7629 vector signed short vec_vsubshs (vector bool short,
7630 vector signed short);
7631 vector signed short vec_vsubshs (vector signed short,
7633 vector signed short vec_vsubshs (vector signed short,
7634 vector signed short);
7636 vector unsigned short vec_vsubuhs (vector bool short,
7637 vector unsigned short);
7638 vector unsigned short vec_vsubuhs (vector unsigned short,
7640 vector unsigned short vec_vsubuhs (vector unsigned short,
7641 vector unsigned short);
7643 vector signed char vec_vsubsbs (vector bool char, vector signed char);
7644 vector signed char vec_vsubsbs (vector signed char, vector bool char);
7645 vector signed char vec_vsubsbs (vector signed char, vector signed char);
7647 vector unsigned char vec_vsububs (vector bool char,
7648 vector unsigned char);
7649 vector unsigned char vec_vsububs (vector unsigned char,
7651 vector unsigned char vec_vsububs (vector unsigned char,
7652 vector unsigned char);
7654 vector unsigned int vec_sum4s (vector unsigned char,
7655 vector unsigned int);
7656 vector signed int vec_sum4s (vector signed char, vector signed int);
7657 vector signed int vec_sum4s (vector signed short, vector signed int);
7659 vector signed int vec_vsum4shs (vector signed short, vector signed int);
7661 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
7663 vector unsigned int vec_vsum4ubs (vector unsigned char,
7664 vector unsigned int);
7666 vector signed int vec_sum2s (vector signed int, vector signed int);
7668 vector signed int vec_sums (vector signed int, vector signed int);
7670 vector float vec_trunc (vector float);
7672 vector signed short vec_unpackh (vector signed char);
7673 vector bool short vec_unpackh (vector bool char);
7674 vector signed int vec_unpackh (vector signed short);
7675 vector bool int vec_unpackh (vector bool short);
7676 vector unsigned int vec_unpackh (vector pixel);
7678 vector bool int vec_vupkhsh (vector bool short);
7679 vector signed int vec_vupkhsh (vector signed short);
7681 vector unsigned int vec_vupkhpx (vector pixel);
7683 vector bool short vec_vupkhsb (vector bool char);
7684 vector signed short vec_vupkhsb (vector signed char);
7686 vector signed short vec_unpackl (vector signed char);
7687 vector bool short vec_unpackl (vector bool char);
7688 vector unsigned int vec_unpackl (vector pixel);
7689 vector signed int vec_unpackl (vector signed short);
7690 vector bool int vec_unpackl (vector bool short);
7692 vector unsigned int vec_vupklpx (vector pixel);
7694 vector bool int vec_vupklsh (vector bool short);
7695 vector signed int vec_vupklsh (vector signed short);
7697 vector bool short vec_vupklsb (vector bool char);
7698 vector signed short vec_vupklsb (vector signed char);
7700 vector float vec_xor (vector float, vector float);
7701 vector float vec_xor (vector float, vector bool int);
7702 vector float vec_xor (vector bool int, vector float);
7703 vector bool int vec_xor (vector bool int, vector bool int);
7704 vector signed int vec_xor (vector bool int, vector signed int);
7705 vector signed int vec_xor (vector signed int, vector bool int);
7706 vector signed int vec_xor (vector signed int, vector signed int);
7707 vector unsigned int vec_xor (vector bool int, vector unsigned int);
7708 vector unsigned int vec_xor (vector unsigned int, vector bool int);
7709 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
7710 vector bool short vec_xor (vector bool short, vector bool short);
7711 vector signed short vec_xor (vector bool short, vector signed short);
7712 vector signed short vec_xor (vector signed short, vector bool short);
7713 vector signed short vec_xor (vector signed short, vector signed short);
7714 vector unsigned short vec_xor (vector bool short,
7715 vector unsigned short);
7716 vector unsigned short vec_xor (vector unsigned short,
7718 vector unsigned short vec_xor (vector unsigned short,
7719 vector unsigned short);
7720 vector signed char vec_xor (vector bool char, vector signed char);
7721 vector bool char vec_xor (vector bool char, vector bool char);
7722 vector signed char vec_xor (vector signed char, vector bool char);
7723 vector signed char vec_xor (vector signed char, vector signed char);
7724 vector unsigned char vec_xor (vector bool char, vector unsigned char);
7725 vector unsigned char vec_xor (vector unsigned char, vector bool char);
7726 vector unsigned char vec_xor (vector unsigned char,
7727 vector unsigned char);
7729 int vec_all_eq (vector signed char, vector bool char);
7730 int vec_all_eq (vector signed char, vector signed char);
7731 int vec_all_eq (vector unsigned char, vector bool char);
7732 int vec_all_eq (vector unsigned char, vector unsigned char);
7733 int vec_all_eq (vector bool char, vector bool char);
7734 int vec_all_eq (vector bool char, vector unsigned char);
7735 int vec_all_eq (vector bool char, vector signed char);
7736 int vec_all_eq (vector signed short, vector bool short);
7737 int vec_all_eq (vector signed short, vector signed short);
7738 int vec_all_eq (vector unsigned short, vector bool short);
7739 int vec_all_eq (vector unsigned short, vector unsigned short);
7740 int vec_all_eq (vector bool short, vector bool short);
7741 int vec_all_eq (vector bool short, vector unsigned short);
7742 int vec_all_eq (vector bool short, vector signed short);
7743 int vec_all_eq (vector pixel, vector pixel);
7744 int vec_all_eq (vector signed int, vector bool int);
7745 int vec_all_eq (vector signed int, vector signed int);
7746 int vec_all_eq (vector unsigned int, vector bool int);
7747 int vec_all_eq (vector unsigned int, vector unsigned int);
7748 int vec_all_eq (vector bool int, vector bool int);
7749 int vec_all_eq (vector bool int, vector unsigned int);
7750 int vec_all_eq (vector bool int, vector signed int);
7751 int vec_all_eq (vector float, vector float);
7753 int vec_all_ge (vector bool char, vector unsigned char);
7754 int vec_all_ge (vector unsigned char, vector bool char);
7755 int vec_all_ge (vector unsigned char, vector unsigned char);
7756 int vec_all_ge (vector bool char, vector signed char);
7757 int vec_all_ge (vector signed char, vector bool char);
7758 int vec_all_ge (vector signed char, vector signed char);
7759 int vec_all_ge (vector bool short, vector unsigned short);
7760 int vec_all_ge (vector unsigned short, vector bool short);
7761 int vec_all_ge (vector unsigned short, vector unsigned short);
7762 int vec_all_ge (vector signed short, vector signed short);
7763 int vec_all_ge (vector bool short, vector signed short);
7764 int vec_all_ge (vector signed short, vector bool short);
7765 int vec_all_ge (vector bool int, vector unsigned int);
7766 int vec_all_ge (vector unsigned int, vector bool int);
7767 int vec_all_ge (vector unsigned int, vector unsigned int);
7768 int vec_all_ge (vector bool int, vector signed int);
7769 int vec_all_ge (vector signed int, vector bool int);
7770 int vec_all_ge (vector signed int, vector signed int);
7771 int vec_all_ge (vector float, vector float);
7773 int vec_all_gt (vector bool char, vector unsigned char);
7774 int vec_all_gt (vector unsigned char, vector bool char);
7775 int vec_all_gt (vector unsigned char, vector unsigned char);
7776 int vec_all_gt (vector bool char, vector signed char);
7777 int vec_all_gt (vector signed char, vector bool char);
7778 int vec_all_gt (vector signed char, vector signed char);
7779 int vec_all_gt (vector bool short, vector unsigned short);
7780 int vec_all_gt (vector unsigned short, vector bool short);
7781 int vec_all_gt (vector unsigned short, vector unsigned short);
7782 int vec_all_gt (vector bool short, vector signed short);
7783 int vec_all_gt (vector signed short, vector bool short);
7784 int vec_all_gt (vector signed short, vector signed short);
7785 int vec_all_gt (vector bool int, vector unsigned int);
7786 int vec_all_gt (vector unsigned int, vector bool int);
7787 int vec_all_gt (vector unsigned int, vector unsigned int);
7788 int vec_all_gt (vector bool int, vector signed int);
7789 int vec_all_gt (vector signed int, vector bool int);
7790 int vec_all_gt (vector signed int, vector signed int);
7791 int vec_all_gt (vector float, vector float);
7793 int vec_all_in (vector float, vector float);
7795 int vec_all_le (vector bool char, vector unsigned char);
7796 int vec_all_le (vector unsigned char, vector bool char);
7797 int vec_all_le (vector unsigned char, vector unsigned char);
7798 int vec_all_le (vector bool char, vector signed char);
7799 int vec_all_le (vector signed char, vector bool char);
7800 int vec_all_le (vector signed char, vector signed char);
7801 int vec_all_le (vector bool short, vector unsigned short);
7802 int vec_all_le (vector unsigned short, vector bool short);
7803 int vec_all_le (vector unsigned short, vector unsigned short);
7804 int vec_all_le (vector bool short, vector signed short);
7805 int vec_all_le (vector signed short, vector bool short);
7806 int vec_all_le (vector signed short, vector signed short);
7807 int vec_all_le (vector bool int, vector unsigned int);
7808 int vec_all_le (vector unsigned int, vector bool int);
7809 int vec_all_le (vector unsigned int, vector unsigned int);
7810 int vec_all_le (vector bool int, vector signed int);
7811 int vec_all_le (vector signed int, vector bool int);
7812 int vec_all_le (vector signed int, vector signed int);
7813 int vec_all_le (vector float, vector float);
7815 int vec_all_lt (vector bool char, vector unsigned char);
7816 int vec_all_lt (vector unsigned char, vector bool char);
7817 int vec_all_lt (vector unsigned char, vector unsigned char);
7818 int vec_all_lt (vector bool char, vector signed char);
7819 int vec_all_lt (vector signed char, vector bool char);
7820 int vec_all_lt (vector signed char, vector signed char);
7821 int vec_all_lt (vector bool short, vector unsigned short);
7822 int vec_all_lt (vector unsigned short, vector bool short);
7823 int vec_all_lt (vector unsigned short, vector unsigned short);
7824 int vec_all_lt (vector bool short, vector signed short);
7825 int vec_all_lt (vector signed short, vector bool short);
7826 int vec_all_lt (vector signed short, vector signed short);
7827 int vec_all_lt (vector bool int, vector unsigned int);
7828 int vec_all_lt (vector unsigned int, vector bool int);
7829 int vec_all_lt (vector unsigned int, vector unsigned int);
7830 int vec_all_lt (vector bool int, vector signed int);
7831 int vec_all_lt (vector signed int, vector bool int);
7832 int vec_all_lt (vector signed int, vector signed int);
7833 int vec_all_lt (vector float, vector float);
7835 int vec_all_nan (vector float);
7837 int vec_all_ne (vector signed char, vector bool char);
7838 int vec_all_ne (vector signed char, vector signed char);
7839 int vec_all_ne (vector unsigned char, vector bool char);
7840 int vec_all_ne (vector unsigned char, vector unsigned char);
7841 int vec_all_ne (vector bool char, vector bool char);
7842 int vec_all_ne (vector bool char, vector unsigned char);
7843 int vec_all_ne (vector bool char, vector signed char);
7844 int vec_all_ne (vector signed short, vector bool short);
7845 int vec_all_ne (vector signed short, vector signed short);
7846 int vec_all_ne (vector unsigned short, vector bool short);
7847 int vec_all_ne (vector unsigned short, vector unsigned short);
7848 int vec_all_ne (vector bool short, vector bool short);
7849 int vec_all_ne (vector bool short, vector unsigned short);
7850 int vec_all_ne (vector bool short, vector signed short);
7851 int vec_all_ne (vector pixel, vector pixel);
7852 int vec_all_ne (vector signed int, vector bool int);
7853 int vec_all_ne (vector signed int, vector signed int);
7854 int vec_all_ne (vector unsigned int, vector bool int);
7855 int vec_all_ne (vector unsigned int, vector unsigned int);
7856 int vec_all_ne (vector bool int, vector bool int);
7857 int vec_all_ne (vector bool int, vector unsigned int);
7858 int vec_all_ne (vector bool int, vector signed int);
7859 int vec_all_ne (vector float, vector float);
7861 int vec_all_nge (vector float, vector float);
7863 int vec_all_ngt (vector float, vector float);
7865 int vec_all_nle (vector float, vector float);
7867 int vec_all_nlt (vector float, vector float);
7869 int vec_all_numeric (vector float);
7871 int vec_any_eq (vector signed char, vector bool char);
7872 int vec_any_eq (vector signed char, vector signed char);
7873 int vec_any_eq (vector unsigned char, vector bool char);
7874 int vec_any_eq (vector unsigned char, vector unsigned char);
7875 int vec_any_eq (vector bool char, vector bool char);
7876 int vec_any_eq (vector bool char, vector unsigned char);
7877 int vec_any_eq (vector bool char, vector signed char);
7878 int vec_any_eq (vector signed short, vector bool short);
7879 int vec_any_eq (vector signed short, vector signed short);
7880 int vec_any_eq (vector unsigned short, vector bool short);
7881 int vec_any_eq (vector unsigned short, vector unsigned short);
7882 int vec_any_eq (vector bool short, vector bool short);
7883 int vec_any_eq (vector bool short, vector unsigned short);
7884 int vec_any_eq (vector bool short, vector signed short);
7885 int vec_any_eq (vector pixel, vector pixel);
7886 int vec_any_eq (vector signed int, vector bool int);
7887 int vec_any_eq (vector signed int, vector signed int);
7888 int vec_any_eq (vector unsigned int, vector bool int);
7889 int vec_any_eq (vector unsigned int, vector unsigned int);
7890 int vec_any_eq (vector bool int, vector bool int);
7891 int vec_any_eq (vector bool int, vector unsigned int);
7892 int vec_any_eq (vector bool int, vector signed int);
7893 int vec_any_eq (vector float, vector float);
7895 int vec_any_ge (vector signed char, vector bool char);
7896 int vec_any_ge (vector unsigned char, vector bool char);
7897 int vec_any_ge (vector unsigned char, vector unsigned char);
7898 int vec_any_ge (vector signed char, vector signed char);
7899 int vec_any_ge (vector bool char, vector unsigned char);
7900 int vec_any_ge (vector bool char, vector signed char);
7901 int vec_any_ge (vector unsigned short, vector bool short);
7902 int vec_any_ge (vector unsigned short, vector unsigned short);
7903 int vec_any_ge (vector signed short, vector signed short);
7904 int vec_any_ge (vector signed short, vector bool short);
7905 int vec_any_ge (vector bool short, vector unsigned short);
7906 int vec_any_ge (vector bool short, vector signed short);
7907 int vec_any_ge (vector signed int, vector bool int);
7908 int vec_any_ge (vector unsigned int, vector bool int);
7909 int vec_any_ge (vector unsigned int, vector unsigned int);
7910 int vec_any_ge (vector signed int, vector signed int);
7911 int vec_any_ge (vector bool int, vector unsigned int);
7912 int vec_any_ge (vector bool int, vector signed int);
7913 int vec_any_ge (vector float, vector float);
7915 int vec_any_gt (vector bool char, vector unsigned char);
7916 int vec_any_gt (vector unsigned char, vector bool char);
7917 int vec_any_gt (vector unsigned char, vector unsigned char);
7918 int vec_any_gt (vector bool char, vector signed char);
7919 int vec_any_gt (vector signed char, vector bool char);
7920 int vec_any_gt (vector signed char, vector signed char);
7921 int vec_any_gt (vector bool short, vector unsigned short);
7922 int vec_any_gt (vector unsigned short, vector bool short);
7923 int vec_any_gt (vector unsigned short, vector unsigned short);
7924 int vec_any_gt (vector bool short, vector signed short);
7925 int vec_any_gt (vector signed short, vector bool short);
7926 int vec_any_gt (vector signed short, vector signed short);
7927 int vec_any_gt (vector bool int, vector unsigned int);
7928 int vec_any_gt (vector unsigned int, vector bool int);
7929 int vec_any_gt (vector unsigned int, vector unsigned int);
7930 int vec_any_gt (vector bool int, vector signed int);
7931 int vec_any_gt (vector signed int, vector bool int);
7932 int vec_any_gt (vector signed int, vector signed int);
7933 int vec_any_gt (vector float, vector float);
7935 int vec_any_le (vector bool char, vector unsigned char);
7936 int vec_any_le (vector unsigned char, vector bool char);
7937 int vec_any_le (vector unsigned char, vector unsigned char);
7938 int vec_any_le (vector bool char, vector signed char);
7939 int vec_any_le (vector signed char, vector bool char);
7940 int vec_any_le (vector signed char, vector signed char);
7941 int vec_any_le (vector bool short, vector unsigned short);
7942 int vec_any_le (vector unsigned short, vector bool short);
7943 int vec_any_le (vector unsigned short, vector unsigned short);
7944 int vec_any_le (vector bool short, vector signed short);
7945 int vec_any_le (vector signed short, vector bool short);
7946 int vec_any_le (vector signed short, vector signed short);
7947 int vec_any_le (vector bool int, vector unsigned int);
7948 int vec_any_le (vector unsigned int, vector bool int);
7949 int vec_any_le (vector unsigned int, vector unsigned int);
7950 int vec_any_le (vector bool int, vector signed int);
7951 int vec_any_le (vector signed int, vector bool int);
7952 int vec_any_le (vector signed int, vector signed int);
7953 int vec_any_le (vector float, vector float);
7955 int vec_any_lt (vector bool char, vector unsigned char);
7956 int vec_any_lt (vector unsigned char, vector bool char);
7957 int vec_any_lt (vector unsigned char, vector unsigned char);
7958 int vec_any_lt (vector bool char, vector signed char);
7959 int vec_any_lt (vector signed char, vector bool char);
7960 int vec_any_lt (vector signed char, vector signed char);
7961 int vec_any_lt (vector bool short, vector unsigned short);
7962 int vec_any_lt (vector unsigned short, vector bool short);
7963 int vec_any_lt (vector unsigned short, vector unsigned short);
7964 int vec_any_lt (vector bool short, vector signed short);
7965 int vec_any_lt (vector signed short, vector bool short);
7966 int vec_any_lt (vector signed short, vector signed short);
7967 int vec_any_lt (vector bool int, vector unsigned int);
7968 int vec_any_lt (vector unsigned int, vector bool int);
7969 int vec_any_lt (vector unsigned int, vector unsigned int);
7970 int vec_any_lt (vector bool int, vector signed int);
7971 int vec_any_lt (vector signed int, vector bool int);
7972 int vec_any_lt (vector signed int, vector signed int);
7973 int vec_any_lt (vector float, vector float);
7975 int vec_any_nan (vector float);
7977 int vec_any_ne (vector signed char, vector bool char);
7978 int vec_any_ne (vector signed char, vector signed char);
7979 int vec_any_ne (vector unsigned char, vector bool char);
7980 int vec_any_ne (vector unsigned char, vector unsigned char);
7981 int vec_any_ne (vector bool char, vector bool char);
7982 int vec_any_ne (vector bool char, vector unsigned char);
7983 int vec_any_ne (vector bool char, vector signed char);
7984 int vec_any_ne (vector signed short, vector bool short);
7985 int vec_any_ne (vector signed short, vector signed short);
7986 int vec_any_ne (vector unsigned short, vector bool short);
7987 int vec_any_ne (vector unsigned short, vector unsigned short);
7988 int vec_any_ne (vector bool short, vector bool short);
7989 int vec_any_ne (vector bool short, vector unsigned short);
7990 int vec_any_ne (vector bool short, vector signed short);
7991 int vec_any_ne (vector pixel, vector pixel);
7992 int vec_any_ne (vector signed int, vector bool int);
7993 int vec_any_ne (vector signed int, vector signed int);
7994 int vec_any_ne (vector unsigned int, vector bool int);
7995 int vec_any_ne (vector unsigned int, vector unsigned int);
7996 int vec_any_ne (vector bool int, vector bool int);
7997 int vec_any_ne (vector bool int, vector unsigned int);
7998 int vec_any_ne (vector bool int, vector signed int);
7999 int vec_any_ne (vector float, vector float);
8001 int vec_any_nge (vector float, vector float);
8003 int vec_any_ngt (vector float, vector float);
8005 int vec_any_nle (vector float, vector float);
8007 int vec_any_nlt (vector float, vector float);
8009 int vec_any_numeric (vector float);
8011 int vec_any_out (vector float, vector float);
8015 @section Pragmas Accepted by GCC
8019 GCC supports several types of pragmas, primarily in order to compile
8020 code originally written for other compilers. Note that in general
8021 we do not recommend the use of pragmas; @xref{Function Attributes},
8022 for further explanation.
8026 * RS/6000 and PowerPC Pragmas::
8033 @subsection ARM Pragmas
8035 The ARM target defines pragmas for controlling the default addition of
8036 @code{long_call} and @code{short_call} attributes to functions.
8037 @xref{Function Attributes}, for information about the effects of these
8042 @cindex pragma, long_calls
8043 Set all subsequent functions to have the @code{long_call} attribute.
8046 @cindex pragma, no_long_calls
8047 Set all subsequent functions to have the @code{short_call} attribute.
8049 @item long_calls_off
8050 @cindex pragma, long_calls_off
8051 Do not affect the @code{long_call} or @code{short_call} attributes of
8052 subsequent functions.
8055 @node RS/6000 and PowerPC Pragmas
8056 @subsection RS/6000 and PowerPC Pragmas
8058 The RS/6000 and PowerPC targets define one pragma for controlling
8059 whether or not the @code{longcall} attribute is added to function
8060 declarations by default. This pragma overrides the @option{-mlongcall}
8061 option, but not the @code{longcall} and @code{shortcall} attributes.
8062 @xref{RS/6000 and PowerPC Options}, for more information about when long
8063 calls are and are not necessary.
8067 @cindex pragma, longcall
8068 Apply the @code{longcall} attribute to all subsequent function
8072 Do not apply the @code{longcall} attribute to subsequent function
8076 @c Describe c4x pragmas here.
8077 @c Describe h8300 pragmas here.
8078 @c Describe i370 pragmas here.
8079 @c Describe i960 pragmas here.
8080 @c Describe sh pragmas here.
8081 @c Describe v850 pragmas here.
8083 @node Darwin Pragmas
8084 @subsection Darwin Pragmas
8086 The following pragmas are available for all architectures running the
8087 Darwin operating system. These are useful for compatibility with other
8091 @item mark @var{tokens}@dots{}
8092 @cindex pragma, mark
8093 This pragma is accepted, but has no effect.
8095 @item options align=@var{alignment}
8096 @cindex pragma, options align
8097 This pragma sets the alignment of fields in structures. The values of
8098 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
8099 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
8100 properly; to restore the previous setting, use @code{reset} for the
8103 @item segment @var{tokens}@dots{}
8104 @cindex pragma, segment
8105 This pragma is accepted, but has no effect.
8107 @item unused (@var{var} [, @var{var}]@dots{})
8108 @cindex pragma, unused
8109 This pragma declares variables to be possibly unused. GCC will not
8110 produce warnings for the listed variables. The effect is similar to
8111 that of the @code{unused} attribute, except that this pragma may appear
8112 anywhere within the variables' scopes.
8115 @node Solaris Pragmas
8116 @subsection Solaris Pragmas
8118 For compatibility with the SunPRO compiler, the following pragma
8122 @item redefine_extname @var{oldname} @var{newname}
8123 @cindex pragma, redefine_extname
8125 This pragma gives the C function @var{oldname} the assembler label
8126 @var{newname}. The pragma must appear before the function declaration.
8127 This pragma is equivalent to the asm labels extension (@pxref{Asm
8128 Labels}). The preprocessor defines @code{__PRAGMA_REDEFINE_EXTNAME}
8129 if the pragma is available.
8133 @subsection Tru64 Pragmas
8135 For compatibility with the Compaq C compiler, the following pragma
8139 @item extern_prefix @var{string}
8140 @cindex pragma, extern_prefix
8142 This pragma renames all subsequent function and variable declarations
8143 such that @var{string} is prepended to the name. This effect may be
8144 terminated by using another @code{extern_prefix} pragma with the
8147 This pragma is similar in intent to to the asm labels extension
8148 (@pxref{Asm Labels}) in that the system programmer wants to change
8149 the assembly-level ABI without changing the source-level API. The
8150 preprocessor defines @code{__PRAGMA_EXTERN_PREFIX} if the pragma is
8154 @node Unnamed Fields
8155 @section Unnamed struct/union fields within structs/unions.
8159 For compatibility with other compilers, GCC allows you to define
8160 a structure or union that contains, as fields, structures and unions
8161 without names. For example:
8174 In this example, the user would be able to access members of the unnamed
8175 union with code like @samp{foo.b}. Note that only unnamed structs and
8176 unions are allowed, you may not have, for example, an unnamed
8179 You must never create such structures that cause ambiguous field definitions.
8180 For example, this structure:
8191 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
8192 Such constructs are not supported and must be avoided. In the future,
8193 such constructs may be detected and treated as compilation errors.
8196 @section Thread-Local Storage
8197 @cindex Thread-Local Storage
8198 @cindex @acronym{TLS}
8201 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
8202 are allocated such that there is one instance of the variable per extant
8203 thread. The run-time model GCC uses to implement this originates
8204 in the IA-64 processor-specific ABI, but has since been migrated
8205 to other processors as well. It requires significant support from
8206 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
8207 system libraries (@file{libc.so} and @file{libpthread.so}), so it
8208 is not available everywhere.
8210 At the user level, the extension is visible with a new storage
8211 class keyword: @code{__thread}. For example:
8215 extern __thread struct state s;
8216 static __thread char *p;
8219 The @code{__thread} specifier may be used alone, with the @code{extern}
8220 or @code{static} specifiers, but with no other storage class specifier.
8221 When used with @code{extern} or @code{static}, @code{__thread} must appear
8222 immediately after the other storage class specifier.
8224 The @code{__thread} specifier may be applied to any global, file-scoped
8225 static, function-scoped static, or static data member of a class. It may
8226 not be applied to block-scoped automatic or non-static data member.
8228 When the address-of operator is applied to a thread-local variable, it is
8229 evaluated at run-time and returns the address of the current thread's
8230 instance of that variable. An address so obtained may be used by any
8231 thread. When a thread terminates, any pointers to thread-local variables
8232 in that thread become invalid.
8234 No static initialization may refer to the address of a thread-local variable.
8236 In C++, if an initializer is present for a thread-local variable, it must
8237 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
8240 See @uref{http://people.redhat.com/drepper/tls.pdf,
8241 ELF Handling For Thread-Local Storage} for a detailed explanation of
8242 the four thread-local storage addressing models, and how the run-time
8243 is expected to function.
8246 * C99 Thread-Local Edits::
8247 * C++98 Thread-Local Edits::
8250 @node C99 Thread-Local Edits
8251 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
8253 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
8254 that document the exact semantics of the language extension.
8258 @cite{5.1.2 Execution environments}
8260 Add new text after paragraph 1
8263 Within either execution environment, a @dfn{thread} is a flow of
8264 control within a program. It is implementation defined whether
8265 or not there may be more than one thread associated with a program.
8266 It is implementation defined how threads beyond the first are
8267 created, the name and type of the function called at thread
8268 startup, and how threads may be terminated. However, objects
8269 with thread storage duration shall be initialized before thread
8274 @cite{6.2.4 Storage durations of objects}
8276 Add new text before paragraph 3
8279 An object whose identifier is declared with the storage-class
8280 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
8281 Its lifetime is the entire execution of the thread, and its
8282 stored value is initialized only once, prior to thread startup.
8286 @cite{6.4.1 Keywords}
8288 Add @code{__thread}.
8291 @cite{6.7.1 Storage-class specifiers}
8293 Add @code{__thread} to the list of storage class specifiers in
8296 Change paragraph 2 to
8299 With the exception of @code{__thread}, at most one storage-class
8300 specifier may be given [@dots{}]. The @code{__thread} specifier may
8301 be used alone, or immediately following @code{extern} or
8305 Add new text after paragraph 6
8308 The declaration of an identifier for a variable that has
8309 block scope that specifies @code{__thread} shall also
8310 specify either @code{extern} or @code{static}.
8312 The @code{__thread} specifier shall be used only with
8317 @node C++98 Thread-Local Edits
8318 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
8320 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
8321 that document the exact semantics of the language extension.
8325 @b{[intro.execution]}
8327 New text after paragraph 4
8330 A @dfn{thread} is a flow of control within the abstract machine.
8331 It is implementation defined whether or not there may be more than
8335 New text after paragraph 7
8338 It is unspecified whether additional action must be taken to
8339 ensure when and whether side effects are visible to other threads.
8345 Add @code{__thread}.
8348 @b{[basic.start.main]}
8350 Add after paragraph 5
8353 The thread that begins execution at the @code{main} function is called
8354 the @dfn{main thread}. It is implementation defined how functions
8355 beginning threads other than the main thread are designated or typed.
8356 A function so designated, as well as the @code{main} function, is called
8357 a @dfn{thread startup function}. It is implementation defined what
8358 happens if a thread startup function returns. It is implementation
8359 defined what happens to other threads when any thread calls @code{exit}.
8363 @b{[basic.start.init]}
8365 Add after paragraph 4
8368 The storage for an object of thread storage duration shall be
8369 statically initialized before the first statement of the thread startup
8370 function. An object of thread storage duration shall not require
8371 dynamic initialization.
8375 @b{[basic.start.term]}
8377 Add after paragraph 3
8380 The type of an object with thread storage duration shall not have a
8381 non-trivial destructor, nor shall it be an array type whose elements
8382 (directly or indirectly) have non-trivial destructors.
8388 Add ``thread storage duration'' to the list in paragraph 1.
8393 Thread, static, and automatic storage durations are associated with
8394 objects introduced by declarations [@dots{}].
8397 Add @code{__thread} to the list of specifiers in paragraph 3.
8400 @b{[basic.stc.thread]}
8402 New section before @b{[basic.stc.static]}
8405 The keyword @code{__thread} applied to a non-local object gives the
8406 object thread storage duration.
8408 A local variable or class data member declared both @code{static}
8409 and @code{__thread} gives the variable or member thread storage
8414 @b{[basic.stc.static]}
8419 All objects which have neither thread storage duration, dynamic
8420 storage duration nor are local [@dots{}].
8426 Add @code{__thread} to the list in paragraph 1.
8431 With the exception of @code{__thread}, at most one
8432 @var{storage-class-specifier} shall appear in a given
8433 @var{decl-specifier-seq}. The @code{__thread} specifier may
8434 be used alone, or immediately following the @code{extern} or
8435 @code{static} specifiers. [@dots{}]
8438 Add after paragraph 5
8441 The @code{__thread} specifier can be applied only to the names of objects
8442 and to anonymous unions.
8448 Add after paragraph 6
8451 Non-@code{static} members shall not be @code{__thread}.
8455 @node C++ Extensions
8456 @chapter Extensions to the C++ Language
8457 @cindex extensions, C++ language
8458 @cindex C++ language extensions
8460 The GNU compiler provides these extensions to the C++ language (and you
8461 can also use most of the C language extensions in your C++ programs). If you
8462 want to write code that checks whether these features are available, you can
8463 test for the GNU compiler the same way as for C programs: check for a
8464 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
8465 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
8466 Predefined Macros,cpp,The GNU C Preprocessor}).
8469 * Min and Max:: C++ Minimum and maximum operators.
8470 * Volatiles:: What constitutes an access to a volatile object.
8471 * Restricted Pointers:: C99 restricted pointers and references.
8472 * Vague Linkage:: Where G++ puts inlines, vtables and such.
8473 * C++ Interface:: You can use a single C++ header file for both
8474 declarations and definitions.
8475 * Template Instantiation:: Methods for ensuring that exactly one copy of
8476 each needed template instantiation is emitted.
8477 * Bound member functions:: You can extract a function pointer to the
8478 method denoted by a @samp{->*} or @samp{.*} expression.
8479 * C++ Attributes:: Variable, function, and type attributes for C++ only.
8480 * Strong Using:: Strong using-directives for namespace composition.
8481 * Offsetof:: Special syntax for implementing @code{offsetof}.
8482 * Java Exceptions:: Tweaking exception handling to work with Java.
8483 * Deprecated Features:: Things will disappear from g++.
8484 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
8488 @section Minimum and Maximum Operators in C++
8490 It is very convenient to have operators which return the ``minimum'' or the
8491 ``maximum'' of two arguments. In GNU C++ (but not in GNU C),
8494 @item @var{a} <? @var{b}
8496 @cindex minimum operator
8497 is the @dfn{minimum}, returning the smaller of the numeric values
8498 @var{a} and @var{b};
8500 @item @var{a} >? @var{b}
8502 @cindex maximum operator
8503 is the @dfn{maximum}, returning the larger of the numeric values @var{a}
8507 These operations are not primitive in ordinary C++, since you can
8508 use a macro to return the minimum of two things in C++, as in the
8512 #define MIN(X,Y) ((X) < (Y) ? : (X) : (Y))
8516 You might then use @w{@samp{int min = MIN (i, j);}} to set @var{min} to
8517 the minimum value of variables @var{i} and @var{j}.
8519 However, side effects in @code{X} or @code{Y} may cause unintended
8520 behavior. For example, @code{MIN (i++, j++)} will fail, incrementing
8521 the smaller counter twice. The GNU C @code{typeof} extension allows you
8522 to write safe macros that avoid this kind of problem (@pxref{Typeof}).
8523 However, writing @code{MIN} and @code{MAX} as macros also forces you to
8524 use function-call notation for a fundamental arithmetic operation.
8525 Using GNU C++ extensions, you can write @w{@samp{int min = i <? j;}}
8528 Since @code{<?} and @code{>?} are built into the compiler, they properly
8529 handle expressions with side-effects; @w{@samp{int min = i++ <? j++;}}
8533 @section When is a Volatile Object Accessed?
8534 @cindex accessing volatiles
8535 @cindex volatile read
8536 @cindex volatile write
8537 @cindex volatile access
8539 Both the C and C++ standard have the concept of volatile objects. These
8540 are normally accessed by pointers and used for accessing hardware. The
8541 standards encourage compilers to refrain from optimizations
8542 concerning accesses to volatile objects that it might perform on
8543 non-volatile objects. The C standard leaves it implementation defined
8544 as to what constitutes a volatile access. The C++ standard omits to
8545 specify this, except to say that C++ should behave in a similar manner
8546 to C with respect to volatiles, where possible. The minimum either
8547 standard specifies is that at a sequence point all previous accesses to
8548 volatile objects have stabilized and no subsequent accesses have
8549 occurred. Thus an implementation is free to reorder and combine
8550 volatile accesses which occur between sequence points, but cannot do so
8551 for accesses across a sequence point. The use of volatiles does not
8552 allow you to violate the restriction on updating objects multiple times
8553 within a sequence point.
8555 In most expressions, it is intuitively obvious what is a read and what is
8556 a write. For instance
8559 volatile int *dst = @var{somevalue};
8560 volatile int *src = @var{someothervalue};
8565 will cause a read of the volatile object pointed to by @var{src} and stores the
8566 value into the volatile object pointed to by @var{dst}. There is no
8567 guarantee that these reads and writes are atomic, especially for objects
8568 larger than @code{int}.
8570 Less obvious expressions are where something which looks like an access
8571 is used in a void context. An example would be,
8574 volatile int *src = @var{somevalue};
8578 With C, such expressions are rvalues, and as rvalues cause a read of
8579 the object, GCC interprets this as a read of the volatile being pointed
8580 to. The C++ standard specifies that such expressions do not undergo
8581 lvalue to rvalue conversion, and that the type of the dereferenced
8582 object may be incomplete. The C++ standard does not specify explicitly
8583 that it is this lvalue to rvalue conversion which is responsible for
8584 causing an access. However, there is reason to believe that it is,
8585 because otherwise certain simple expressions become undefined. However,
8586 because it would surprise most programmers, G++ treats dereferencing a
8587 pointer to volatile object of complete type in a void context as a read
8588 of the object. When the object has incomplete type, G++ issues a
8593 struct T @{int m;@};
8594 volatile S *ptr1 = @var{somevalue};
8595 volatile T *ptr2 = @var{somevalue};
8600 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
8601 causes a read of the object pointed to. If you wish to force an error on
8602 the first case, you must force a conversion to rvalue with, for instance
8603 a static cast, @code{static_cast<S>(*ptr1)}.
8605 When using a reference to volatile, G++ does not treat equivalent
8606 expressions as accesses to volatiles, but instead issues a warning that
8607 no volatile is accessed. The rationale for this is that otherwise it
8608 becomes difficult to determine where volatile access occur, and not
8609 possible to ignore the return value from functions returning volatile
8610 references. Again, if you wish to force a read, cast the reference to
8613 @node Restricted Pointers
8614 @section Restricting Pointer Aliasing
8615 @cindex restricted pointers
8616 @cindex restricted references
8617 @cindex restricted this pointer
8619 As with the C front end, G++ understands the C99 feature of restricted pointers,
8620 specified with the @code{__restrict__}, or @code{__restrict} type
8621 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
8622 language flag, @code{restrict} is not a keyword in C++.
8624 In addition to allowing restricted pointers, you can specify restricted
8625 references, which indicate that the reference is not aliased in the local
8629 void fn (int *__restrict__ rptr, int &__restrict__ rref)
8636 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
8637 @var{rref} refers to a (different) unaliased integer.
8639 You may also specify whether a member function's @var{this} pointer is
8640 unaliased by using @code{__restrict__} as a member function qualifier.
8643 void T::fn () __restrict__
8650 Within the body of @code{T::fn}, @var{this} will have the effective
8651 definition @code{T *__restrict__ const this}. Notice that the
8652 interpretation of a @code{__restrict__} member function qualifier is
8653 different to that of @code{const} or @code{volatile} qualifier, in that it
8654 is applied to the pointer rather than the object. This is consistent with
8655 other compilers which implement restricted pointers.
8657 As with all outermost parameter qualifiers, @code{__restrict__} is
8658 ignored in function definition matching. This means you only need to
8659 specify @code{__restrict__} in a function definition, rather than
8660 in a function prototype as well.
8663 @section Vague Linkage
8664 @cindex vague linkage
8666 There are several constructs in C++ which require space in the object
8667 file but are not clearly tied to a single translation unit. We say that
8668 these constructs have ``vague linkage''. Typically such constructs are
8669 emitted wherever they are needed, though sometimes we can be more
8673 @item Inline Functions
8674 Inline functions are typically defined in a header file which can be
8675 included in many different compilations. Hopefully they can usually be
8676 inlined, but sometimes an out-of-line copy is necessary, if the address
8677 of the function is taken or if inlining fails. In general, we emit an
8678 out-of-line copy in all translation units where one is needed. As an
8679 exception, we only emit inline virtual functions with the vtable, since
8680 it will always require a copy.
8682 Local static variables and string constants used in an inline function
8683 are also considered to have vague linkage, since they must be shared
8684 between all inlined and out-of-line instances of the function.
8688 C++ virtual functions are implemented in most compilers using a lookup
8689 table, known as a vtable. The vtable contains pointers to the virtual
8690 functions provided by a class, and each object of the class contains a
8691 pointer to its vtable (or vtables, in some multiple-inheritance
8692 situations). If the class declares any non-inline, non-pure virtual
8693 functions, the first one is chosen as the ``key method'' for the class,
8694 and the vtable is only emitted in the translation unit where the key
8697 @emph{Note:} If the chosen key method is later defined as inline, the
8698 vtable will still be emitted in every translation unit which defines it.
8699 Make sure that any inline virtuals are declared inline in the class
8700 body, even if they are not defined there.
8702 @item type_info objects
8705 C++ requires information about types to be written out in order to
8706 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
8707 For polymorphic classes (classes with virtual functions), the type_info
8708 object is written out along with the vtable so that @samp{dynamic_cast}
8709 can determine the dynamic type of a class object at runtime. For all
8710 other types, we write out the type_info object when it is used: when
8711 applying @samp{typeid} to an expression, throwing an object, or
8712 referring to a type in a catch clause or exception specification.
8714 @item Template Instantiations
8715 Most everything in this section also applies to template instantiations,
8716 but there are other options as well.
8717 @xref{Template Instantiation,,Where's the Template?}.
8721 When used with GNU ld version 2.8 or later on an ELF system such as
8722 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
8723 these constructs will be discarded at link time. This is known as
8726 On targets that don't support COMDAT, but do support weak symbols, GCC
8727 will use them. This way one copy will override all the others, but
8728 the unused copies will still take up space in the executable.
8730 For targets which do not support either COMDAT or weak symbols,
8731 most entities with vague linkage will be emitted as local symbols to
8732 avoid duplicate definition errors from the linker. This will not happen
8733 for local statics in inlines, however, as having multiple copies will
8734 almost certainly break things.
8736 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
8737 another way to control placement of these constructs.
8740 @section #pragma interface and implementation
8742 @cindex interface and implementation headers, C++
8743 @cindex C++ interface and implementation headers
8744 @cindex pragmas, interface and implementation
8746 @code{#pragma interface} and @code{#pragma implementation} provide the
8747 user with a way of explicitly directing the compiler to emit entities
8748 with vague linkage (and debugging information) in a particular
8751 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
8752 most cases, because of COMDAT support and the ``key method'' heuristic
8753 mentioned in @ref{Vague Linkage}. Using them can actually cause your
8754 program to grow due to unnecesary out-of-line copies of inline
8755 functions. Currently the only benefit of these @code{#pragma}s is
8756 reduced duplication of debugging information, and that should be
8757 addressed soon on DWARF 2 targets with the use of COMDAT sections.
8760 @item #pragma interface
8761 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
8762 @kindex #pragma interface
8763 Use this directive in @emph{header files} that define object classes, to save
8764 space in most of the object files that use those classes. Normally,
8765 local copies of certain information (backup copies of inline member
8766 functions, debugging information, and the internal tables that implement
8767 virtual functions) must be kept in each object file that includes class
8768 definitions. You can use this pragma to avoid such duplication. When a
8769 header file containing @samp{#pragma interface} is included in a
8770 compilation, this auxiliary information will not be generated (unless
8771 the main input source file itself uses @samp{#pragma implementation}).
8772 Instead, the object files will contain references to be resolved at link
8775 The second form of this directive is useful for the case where you have
8776 multiple headers with the same name in different directories. If you
8777 use this form, you must specify the same string to @samp{#pragma
8780 @item #pragma implementation
8781 @itemx #pragma implementation "@var{objects}.h"
8782 @kindex #pragma implementation
8783 Use this pragma in a @emph{main input file}, when you want full output from
8784 included header files to be generated (and made globally visible). The
8785 included header file, in turn, should use @samp{#pragma interface}.
8786 Backup copies of inline member functions, debugging information, and the
8787 internal tables used to implement virtual functions are all generated in
8788 implementation files.
8790 @cindex implied @code{#pragma implementation}
8791 @cindex @code{#pragma implementation}, implied
8792 @cindex naming convention, implementation headers
8793 If you use @samp{#pragma implementation} with no argument, it applies to
8794 an include file with the same basename@footnote{A file's @dfn{basename}
8795 was the name stripped of all leading path information and of trailing
8796 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
8797 file. For example, in @file{allclass.cc}, giving just
8798 @samp{#pragma implementation}
8799 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
8801 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
8802 an implementation file whenever you would include it from
8803 @file{allclass.cc} even if you never specified @samp{#pragma
8804 implementation}. This was deemed to be more trouble than it was worth,
8805 however, and disabled.
8807 Use the string argument if you want a single implementation file to
8808 include code from multiple header files. (You must also use
8809 @samp{#include} to include the header file; @samp{#pragma
8810 implementation} only specifies how to use the file---it doesn't actually
8813 There is no way to split up the contents of a single header file into
8814 multiple implementation files.
8817 @cindex inlining and C++ pragmas
8818 @cindex C++ pragmas, effect on inlining
8819 @cindex pragmas in C++, effect on inlining
8820 @samp{#pragma implementation} and @samp{#pragma interface} also have an
8821 effect on function inlining.
8823 If you define a class in a header file marked with @samp{#pragma
8824 interface}, the effect on an inline function defined in that class is
8825 similar to an explicit @code{extern} declaration---the compiler emits
8826 no code at all to define an independent version of the function. Its
8827 definition is used only for inlining with its callers.
8829 @opindex fno-implement-inlines
8830 Conversely, when you include the same header file in a main source file
8831 that declares it as @samp{#pragma implementation}, the compiler emits
8832 code for the function itself; this defines a version of the function
8833 that can be found via pointers (or by callers compiled without
8834 inlining). If all calls to the function can be inlined, you can avoid
8835 emitting the function by compiling with @option{-fno-implement-inlines}.
8836 If any calls were not inlined, you will get linker errors.
8838 @node Template Instantiation
8839 @section Where's the Template?
8840 @cindex template instantiation
8842 C++ templates are the first language feature to require more
8843 intelligence from the environment than one usually finds on a UNIX
8844 system. Somehow the compiler and linker have to make sure that each
8845 template instance occurs exactly once in the executable if it is needed,
8846 and not at all otherwise. There are two basic approaches to this
8847 problem, which are referred to as the Borland model and the Cfront model.
8851 Borland C++ solved the template instantiation problem by adding the code
8852 equivalent of common blocks to their linker; the compiler emits template
8853 instances in each translation unit that uses them, and the linker
8854 collapses them together. The advantage of this model is that the linker
8855 only has to consider the object files themselves; there is no external
8856 complexity to worry about. This disadvantage is that compilation time
8857 is increased because the template code is being compiled repeatedly.
8858 Code written for this model tends to include definitions of all
8859 templates in the header file, since they must be seen to be
8863 The AT&T C++ translator, Cfront, solved the template instantiation
8864 problem by creating the notion of a template repository, an
8865 automatically maintained place where template instances are stored. A
8866 more modern version of the repository works as follows: As individual
8867 object files are built, the compiler places any template definitions and
8868 instantiations encountered in the repository. At link time, the link
8869 wrapper adds in the objects in the repository and compiles any needed
8870 instances that were not previously emitted. The advantages of this
8871 model are more optimal compilation speed and the ability to use the
8872 system linker; to implement the Borland model a compiler vendor also
8873 needs to replace the linker. The disadvantages are vastly increased
8874 complexity, and thus potential for error; for some code this can be
8875 just as transparent, but in practice it can been very difficult to build
8876 multiple programs in one directory and one program in multiple
8877 directories. Code written for this model tends to separate definitions
8878 of non-inline member templates into a separate file, which should be
8879 compiled separately.
8882 When used with GNU ld version 2.8 or later on an ELF system such as
8883 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
8884 Borland model. On other systems, G++ implements neither automatic
8887 A future version of G++ will support a hybrid model whereby the compiler
8888 will emit any instantiations for which the template definition is
8889 included in the compile, and store template definitions and
8890 instantiation context information into the object file for the rest.
8891 The link wrapper will extract that information as necessary and invoke
8892 the compiler to produce the remaining instantiations. The linker will
8893 then combine duplicate instantiations.
8895 In the mean time, you have the following options for dealing with
8896 template instantiations:
8901 Compile your template-using code with @option{-frepo}. The compiler will
8902 generate files with the extension @samp{.rpo} listing all of the
8903 template instantiations used in the corresponding object files which
8904 could be instantiated there; the link wrapper, @samp{collect2}, will
8905 then update the @samp{.rpo} files to tell the compiler where to place
8906 those instantiations and rebuild any affected object files. The
8907 link-time overhead is negligible after the first pass, as the compiler
8908 will continue to place the instantiations in the same files.
8910 This is your best option for application code written for the Borland
8911 model, as it will just work. Code written for the Cfront model will
8912 need to be modified so that the template definitions are available at
8913 one or more points of instantiation; usually this is as simple as adding
8914 @code{#include <tmethods.cc>} to the end of each template header.
8916 For library code, if you want the library to provide all of the template
8917 instantiations it needs, just try to link all of its object files
8918 together; the link will fail, but cause the instantiations to be
8919 generated as a side effect. Be warned, however, that this may cause
8920 conflicts if multiple libraries try to provide the same instantiations.
8921 For greater control, use explicit instantiation as described in the next
8925 @opindex fno-implicit-templates
8926 Compile your code with @option{-fno-implicit-templates} to disable the
8927 implicit generation of template instances, and explicitly instantiate
8928 all the ones you use. This approach requires more knowledge of exactly
8929 which instances you need than do the others, but it's less
8930 mysterious and allows greater control. You can scatter the explicit
8931 instantiations throughout your program, perhaps putting them in the
8932 translation units where the instances are used or the translation units
8933 that define the templates themselves; you can put all of the explicit
8934 instantiations you need into one big file; or you can create small files
8941 template class Foo<int>;
8942 template ostream& operator <<
8943 (ostream&, const Foo<int>&);
8946 for each of the instances you need, and create a template instantiation
8949 If you are using Cfront-model code, you can probably get away with not
8950 using @option{-fno-implicit-templates} when compiling files that don't
8951 @samp{#include} the member template definitions.
8953 If you use one big file to do the instantiations, you may want to
8954 compile it without @option{-fno-implicit-templates} so you get all of the
8955 instances required by your explicit instantiations (but not by any
8956 other files) without having to specify them as well.
8958 G++ has extended the template instantiation syntax given in the ISO
8959 standard to allow forward declaration of explicit instantiations
8960 (with @code{extern}), instantiation of the compiler support data for a
8961 template class (i.e.@: the vtable) without instantiating any of its
8962 members (with @code{inline}), and instantiation of only the static data
8963 members of a template class, without the support data or member
8964 functions (with (@code{static}):
8967 extern template int max (int, int);
8968 inline template class Foo<int>;
8969 static template class Foo<int>;
8973 Do nothing. Pretend G++ does implement automatic instantiation
8974 management. Code written for the Borland model will work fine, but
8975 each translation unit will contain instances of each of the templates it
8976 uses. In a large program, this can lead to an unacceptable amount of code
8980 @node Bound member functions
8981 @section Extracting the function pointer from a bound pointer to member function
8983 @cindex pointer to member function
8984 @cindex bound pointer to member function
8986 In C++, pointer to member functions (PMFs) are implemented using a wide
8987 pointer of sorts to handle all the possible call mechanisms; the PMF
8988 needs to store information about how to adjust the @samp{this} pointer,
8989 and if the function pointed to is virtual, where to find the vtable, and
8990 where in the vtable to look for the member function. If you are using
8991 PMFs in an inner loop, you should really reconsider that decision. If
8992 that is not an option, you can extract the pointer to the function that
8993 would be called for a given object/PMF pair and call it directly inside
8994 the inner loop, to save a bit of time.
8996 Note that you will still be paying the penalty for the call through a
8997 function pointer; on most modern architectures, such a call defeats the
8998 branch prediction features of the CPU@. This is also true of normal
8999 virtual function calls.
9001 The syntax for this extension is
9005 extern int (A::*fp)();
9006 typedef int (*fptr)(A *);
9008 fptr p = (fptr)(a.*fp);
9011 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
9012 no object is needed to obtain the address of the function. They can be
9013 converted to function pointers directly:
9016 fptr p1 = (fptr)(&A::foo);
9019 @opindex Wno-pmf-conversions
9020 You must specify @option{-Wno-pmf-conversions} to use this extension.
9022 @node C++ Attributes
9023 @section C++-Specific Variable, Function, and Type Attributes
9025 Some attributes only make sense for C++ programs.
9028 @item init_priority (@var{priority})
9029 @cindex init_priority attribute
9032 In Standard C++, objects defined at namespace scope are guaranteed to be
9033 initialized in an order in strict accordance with that of their definitions
9034 @emph{in a given translation unit}. No guarantee is made for initializations
9035 across translation units. However, GNU C++ allows users to control the
9036 order of initialization of objects defined at namespace scope with the
9037 @code{init_priority} attribute by specifying a relative @var{priority},
9038 a constant integral expression currently bounded between 101 and 65535
9039 inclusive. Lower numbers indicate a higher priority.
9041 In the following example, @code{A} would normally be created before
9042 @code{B}, but the @code{init_priority} attribute has reversed that order:
9045 Some_Class A __attribute__ ((init_priority (2000)));
9046 Some_Class B __attribute__ ((init_priority (543)));
9050 Note that the particular values of @var{priority} do not matter; only their
9053 @item java_interface
9054 @cindex java_interface attribute
9056 This type attribute informs C++ that the class is a Java interface. It may
9057 only be applied to classes declared within an @code{extern "Java"} block.
9058 Calls to methods declared in this interface will be dispatched using GCJ's
9059 interface table mechanism, instead of regular virtual table dispatch.
9063 See also @xref{Strong Using}.
9066 @section Strong Using
9068 @strong{Caution:} The semantics of this extension are not fully
9069 defined. Users should refrain from using this extension as its
9070 semantics may change subtly over time. It is possible that this
9071 extension wil be removed in future versions of G++.
9073 A using-directive with @code{__attribute ((strong))} is stronger
9074 than a normal using-directive in two ways:
9078 Templates from the used namespace can be specialized as though they were members of the using namespace.
9081 The using namespace is considered an associated namespace of all
9082 templates in the used namespace for purposes of argument-dependent
9086 This is useful for composing a namespace transparently from
9087 implementation namespaces. For example:
9092 template <class T> struct A @{ @};
9094 using namespace debug __attribute ((__strong__));
9095 template <> struct A<int> @{ @}; // ok to specialize
9097 template <class T> void f (A<T>);
9102 f (std::A<float>()); // lookup finds std::f
9110 G++ uses a syntactic extension to implement the @code{offsetof} macro.
9115 __offsetof__ (expression)
9118 is equivalent to the parenthesized expression, except that the
9119 expression is considered an integral constant expression even if it
9120 contains certain operators that are not normally permitted in an
9121 integral constant expression. Users should never use
9122 @code{__offsetof__} directly; the only valid use of
9123 @code{__offsetof__} is to implement the @code{offsetof} macro in
9126 @node Java Exceptions
9127 @section Java Exceptions
9129 The Java language uses a slightly different exception handling model
9130 from C++. Normally, GNU C++ will automatically detect when you are
9131 writing C++ code that uses Java exceptions, and handle them
9132 appropriately. However, if C++ code only needs to execute destructors
9133 when Java exceptions are thrown through it, GCC will guess incorrectly.
9134 Sample problematic code is:
9137 struct S @{ ~S(); @};
9138 extern void bar(); // is written in Java, and may throw exceptions
9147 The usual effect of an incorrect guess is a link failure, complaining of
9148 a missing routine called @samp{__gxx_personality_v0}.
9150 You can inform the compiler that Java exceptions are to be used in a
9151 translation unit, irrespective of what it might think, by writing
9152 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
9153 @samp{#pragma} must appear before any functions that throw or catch
9154 exceptions, or run destructors when exceptions are thrown through them.
9156 You cannot mix Java and C++ exceptions in the same translation unit. It
9157 is believed to be safe to throw a C++ exception from one file through
9158 another file compiled for the Java exception model, or vice versa, but
9159 there may be bugs in this area.
9161 @node Deprecated Features
9162 @section Deprecated Features
9164 In the past, the GNU C++ compiler was extended to experiment with new
9165 features, at a time when the C++ language was still evolving. Now that
9166 the C++ standard is complete, some of those features are superseded by
9167 superior alternatives. Using the old features might cause a warning in
9168 some cases that the feature will be dropped in the future. In other
9169 cases, the feature might be gone already.
9171 While the list below is not exhaustive, it documents some of the options
9172 that are now deprecated:
9175 @item -fexternal-templates
9176 @itemx -falt-external-templates
9177 These are two of the many ways for G++ to implement template
9178 instantiation. @xref{Template Instantiation}. The C++ standard clearly
9179 defines how template definitions have to be organized across
9180 implementation units. G++ has an implicit instantiation mechanism that
9181 should work just fine for standard-conforming code.
9183 @item -fstrict-prototype
9184 @itemx -fno-strict-prototype
9185 Previously it was possible to use an empty prototype parameter list to
9186 indicate an unspecified number of parameters (like C), rather than no
9187 parameters, as C++ demands. This feature has been removed, except where
9188 it is required for backwards compatibility @xref{Backwards Compatibility}.
9191 The named return value extension has been deprecated, and is now
9194 The use of initializer lists with new expressions has been deprecated,
9195 and is now removed from G++.
9197 Floating and complex non-type template parameters have been deprecated,
9198 and are now removed from G++.
9200 The implicit typename extension has been deprecated and is now
9203 The use of default arguments in function pointers, function typedefs and
9204 and other places where they are not permitted by the standard is
9205 deprecated and will be removed from a future version of G++.
9207 @node Backwards Compatibility
9208 @section Backwards Compatibility
9209 @cindex Backwards Compatibility
9210 @cindex ARM [Annotated C++ Reference Manual]
9212 Now that there is a definitive ISO standard C++, G++ has a specification
9213 to adhere to. The C++ language evolved over time, and features that
9214 used to be acceptable in previous drafts of the standard, such as the ARM
9215 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
9216 compilation of C++ written to such drafts, G++ contains some backwards
9217 compatibilities. @emph{All such backwards compatibility features are
9218 liable to disappear in future versions of G++.} They should be considered
9219 deprecated @xref{Deprecated Features}.
9223 If a variable is declared at for scope, it used to remain in scope until
9224 the end of the scope which contained the for statement (rather than just
9225 within the for scope). G++ retains this, but issues a warning, if such a
9226 variable is accessed outside the for scope.
9228 @item Implicit C language
9229 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
9230 scope to set the language. On such systems, all header files are
9231 implicitly scoped inside a C language scope. Also, an empty prototype
9232 @code{()} will be treated as an unspecified number of arguments, rather
9233 than no arguments, as C++ demands.