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 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
433 * Long Long:: Double-word integers---@code{long long int}.
434 * Complex:: Data types for complex numbers.
435 * Hex Floats:: Hexadecimal floating-point constants.
436 * Zero Length:: Zero-length arrays.
437 * Variable Length:: Arrays whose length is computed at run time.
438 * Empty Structures:: Structures with no members.
439 * Variadic Macros:: Macros with a variable number of arguments.
440 * Escaped Newlines:: Slightly looser rules for escaped newlines.
441 * Subscripting:: Any array can be subscripted, even if not an lvalue.
442 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
443 * Initializers:: Non-constant initializers.
444 * Compound Literals:: Compound literals give structures, unions
446 * Designated Inits:: Labeling elements of initializers.
447 * Cast to Union:: Casting to union type from any member of the union.
448 * Case Ranges:: `case 1 ... 9' and such.
449 * Mixed Declarations:: Mixing declarations and code.
450 * Function Attributes:: Declaring that functions have no side effects,
451 or that they can never return.
452 * Attribute Syntax:: Formal syntax for attributes.
453 * Function Prototypes:: Prototype declarations and old-style definitions.
454 * C++ Comments:: C++ comments are recognized.
455 * Dollar Signs:: Dollar sign is allowed in identifiers.
456 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
457 * Variable Attributes:: Specifying attributes of variables.
458 * Type Attributes:: Specifying attributes of types.
459 * Alignment:: Inquiring about the alignment of a type or variable.
460 * Inline:: Defining inline functions (as fast as macros).
461 * Extended Asm:: Assembler instructions with C expressions as operands.
462 (With them you can define ``built-in'' functions.)
463 * Constraints:: Constraints for asm operands
464 * Asm Labels:: Specifying the assembler name to use for a C symbol.
465 * Explicit Reg Vars:: Defining variables residing in specified registers.
466 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
467 * Incomplete Enums:: @code{enum foo;}, with details to follow.
468 * Function Names:: Printable strings which are the name of the current
470 * Return Address:: Getting the return or frame address of a function.
471 * Vector Extensions:: Using vector instructions through built-in functions.
472 * Offsetof:: Special syntax for implementing @code{offsetof}.
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 Conditionals with Omitted Operands
1065 @cindex conditional expressions, extensions
1066 @cindex omitted middle-operands
1067 @cindex middle-operands, omitted
1068 @cindex extensions, @code{?:}
1069 @cindex @code{?:} extensions
1071 The middle operand in a conditional expression may be omitted. Then
1072 if the first operand is nonzero, its value is the value of the conditional
1075 Therefore, the expression
1082 has the value of @code{x} if that is nonzero; otherwise, the value of
1085 This example is perfectly equivalent to
1091 @cindex side effect in ?:
1092 @cindex ?: side effect
1094 In this simple case, the ability to omit the middle operand is not
1095 especially useful. When it becomes useful is when the first operand does,
1096 or may (if it is a macro argument), contain a side effect. Then repeating
1097 the operand in the middle would perform the side effect twice. Omitting
1098 the middle operand uses the value already computed without the undesirable
1099 effects of recomputing it.
1102 @section Double-Word Integers
1103 @cindex @code{long long} data types
1104 @cindex double-word arithmetic
1105 @cindex multiprecision arithmetic
1106 @cindex @code{LL} integer suffix
1107 @cindex @code{ULL} integer suffix
1109 ISO C99 supports data types for integers that are at least 64 bits wide,
1110 and as an extension GCC supports them in C89 mode and in C++.
1111 Simply write @code{long long int} for a signed integer, or
1112 @code{unsigned long long int} for an unsigned integer. To make an
1113 integer constant of type @code{long long int}, add the suffix @samp{LL}
1114 to the integer. To make an integer constant of type @code{unsigned long
1115 long int}, add the suffix @samp{ULL} to the integer.
1117 You can use these types in arithmetic like any other integer types.
1118 Addition, subtraction, and bitwise boolean operations on these types
1119 are open-coded on all types of machines. Multiplication is open-coded
1120 if the machine supports fullword-to-doubleword a widening multiply
1121 instruction. Division and shifts are open-coded only on machines that
1122 provide special support. The operations that are not open-coded use
1123 special library routines that come with GCC@.
1125 There may be pitfalls when you use @code{long long} types for function
1126 arguments, unless you declare function prototypes. If a function
1127 expects type @code{int} for its argument, and you pass a value of type
1128 @code{long long int}, confusion will result because the caller and the
1129 subroutine will disagree about the number of bytes for the argument.
1130 Likewise, if the function expects @code{long long int} and you pass
1131 @code{int}. The best way to avoid such problems is to use prototypes.
1134 @section Complex Numbers
1135 @cindex complex numbers
1136 @cindex @code{_Complex} keyword
1137 @cindex @code{__complex__} keyword
1139 ISO C99 supports complex floating data types, and as an extension GCC
1140 supports them in C89 mode and in C++, and supports complex integer data
1141 types which are not part of ISO C99. You can declare complex types
1142 using the keyword @code{_Complex}. As an extension, the older GNU
1143 keyword @code{__complex__} is also supported.
1145 For example, @samp{_Complex double x;} declares @code{x} as a
1146 variable whose real part and imaginary part are both of type
1147 @code{double}. @samp{_Complex short int y;} declares @code{y} to
1148 have real and imaginary parts of type @code{short int}; this is not
1149 likely to be useful, but it shows that the set of complex types is
1152 To write a constant with a complex data type, use the suffix @samp{i} or
1153 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
1154 has type @code{_Complex float} and @code{3i} has type
1155 @code{_Complex int}. Such a constant always has a pure imaginary
1156 value, but you can form any complex value you like by adding one to a
1157 real constant. This is a GNU extension; if you have an ISO C99
1158 conforming C library (such as GNU libc), and want to construct complex
1159 constants of floating type, you should include @code{<complex.h>} and
1160 use the macros @code{I} or @code{_Complex_I} instead.
1162 @cindex @code{__real__} keyword
1163 @cindex @code{__imag__} keyword
1164 To extract the real part of a complex-valued expression @var{exp}, write
1165 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
1166 extract the imaginary part. This is a GNU extension; for values of
1167 floating type, you should use the ISO C99 functions @code{crealf},
1168 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
1169 @code{cimagl}, declared in @code{<complex.h>} and also provided as
1170 built-in functions by GCC@.
1172 @cindex complex conjugation
1173 The operator @samp{~} performs complex conjugation when used on a value
1174 with a complex type. This is a GNU extension; for values of
1175 floating type, you should use the ISO C99 functions @code{conjf},
1176 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
1177 provided as built-in functions by GCC@.
1179 GCC can allocate complex automatic variables in a noncontiguous
1180 fashion; it's even possible for the real part to be in a register while
1181 the imaginary part is on the stack (or vice-versa). Only the DWARF2
1182 debug info format can represent this, so use of DWARF2 is recommended.
1183 If you are using the stabs debug info format, GCC describes a noncontiguous
1184 complex variable as if it were two separate variables of noncomplex type.
1185 If the variable's actual name is @code{foo}, the two fictitious
1186 variables are named @code{foo$real} and @code{foo$imag}. You can
1187 examine and set these two fictitious variables with your debugger.
1193 ISO C99 supports floating-point numbers written not only in the usual
1194 decimal notation, such as @code{1.55e1}, but also numbers such as
1195 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
1196 supports this in C89 mode (except in some cases when strictly
1197 conforming) and in C++. In that format the
1198 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
1199 mandatory. The exponent is a decimal number that indicates the power of
1200 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
1207 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
1208 is the same as @code{1.55e1}.
1210 Unlike for floating-point numbers in the decimal notation the exponent
1211 is always required in the hexadecimal notation. Otherwise the compiler
1212 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
1213 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
1214 extension for floating-point constants of type @code{float}.
1217 @section Arrays of Length Zero
1218 @cindex arrays of length zero
1219 @cindex zero-length arrays
1220 @cindex length-zero arrays
1221 @cindex flexible array members
1223 Zero-length arrays are allowed in GNU C@. They are very useful as the
1224 last element of a structure which is really a header for a variable-length
1233 struct line *thisline = (struct line *)
1234 malloc (sizeof (struct line) + this_length);
1235 thisline->length = this_length;
1238 In ISO C90, you would have to give @code{contents} a length of 1, which
1239 means either you waste space or complicate the argument to @code{malloc}.
1241 In ISO C99, you would use a @dfn{flexible array member}, which is
1242 slightly different in syntax and semantics:
1246 Flexible array members are written as @code{contents[]} without
1250 Flexible array members have incomplete type, and so the @code{sizeof}
1251 operator may not be applied. As a quirk of the original implementation
1252 of zero-length arrays, @code{sizeof} evaluates to zero.
1255 Flexible array members may only appear as the last member of a
1256 @code{struct} that is otherwise non-empty.
1259 A structure containing a flexible array member, or a union containing
1260 such a structure (possibly recursively), may not be a member of a
1261 structure or an element of an array. (However, these uses are
1262 permitted by GCC as extensions.)
1265 GCC versions before 3.0 allowed zero-length arrays to be statically
1266 initialized, as if they were flexible arrays. In addition to those
1267 cases that were useful, it also allowed initializations in situations
1268 that would corrupt later data. Non-empty initialization of zero-length
1269 arrays is now treated like any case where there are more initializer
1270 elements than the array holds, in that a suitable warning about "excess
1271 elements in array" is given, and the excess elements (all of them, in
1272 this case) are ignored.
1274 Instead GCC allows static initialization of flexible array members.
1275 This is equivalent to defining a new structure containing the original
1276 structure followed by an array of sufficient size to contain the data.
1277 I.e.@: in the following, @code{f1} is constructed as if it were declared
1283 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
1286 struct f1 f1; int data[3];
1287 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1291 The convenience of this extension is that @code{f1} has the desired
1292 type, eliminating the need to consistently refer to @code{f2.f1}.
1294 This has symmetry with normal static arrays, in that an array of
1295 unknown size is also written with @code{[]}.
1297 Of course, this extension only makes sense if the extra data comes at
1298 the end of a top-level object, as otherwise we would be overwriting
1299 data at subsequent offsets. To avoid undue complication and confusion
1300 with initialization of deeply nested arrays, we simply disallow any
1301 non-empty initialization except when the structure is the top-level
1302 object. For example:
1305 struct foo @{ int x; int y[]; @};
1306 struct bar @{ struct foo z; @};
1308 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1309 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1310 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1311 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1314 @node Empty Structures
1315 @section Structures With No Members
1316 @cindex empty structures
1317 @cindex zero-size structures
1319 GCC permits a C structure to have no members:
1326 The structure will have size zero. In C++, empty structures are part
1327 of the language. G++ treats empty structures as if they had a single
1328 member of type @code{char}.
1330 @node Variable Length
1331 @section Arrays of Variable Length
1332 @cindex variable-length arrays
1333 @cindex arrays of variable length
1336 Variable-length automatic arrays are allowed in ISO C99, and as an
1337 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1338 implementation of variable-length arrays does not yet conform in detail
1339 to the ISO C99 standard.) These arrays are
1340 declared like any other automatic arrays, but with a length that is not
1341 a constant expression. The storage is allocated at the point of
1342 declaration and deallocated when the brace-level is exited. For
1347 concat_fopen (char *s1, char *s2, char *mode)
1349 char str[strlen (s1) + strlen (s2) + 1];
1352 return fopen (str, mode);
1356 @cindex scope of a variable length array
1357 @cindex variable-length array scope
1358 @cindex deallocating variable length arrays
1359 Jumping or breaking out of the scope of the array name deallocates the
1360 storage. Jumping into the scope is not allowed; you get an error
1363 @cindex @code{alloca} vs variable-length arrays
1364 You can use the function @code{alloca} to get an effect much like
1365 variable-length arrays. The function @code{alloca} is available in
1366 many other C implementations (but not in all). On the other hand,
1367 variable-length arrays are more elegant.
1369 There are other differences between these two methods. Space allocated
1370 with @code{alloca} exists until the containing @emph{function} returns.
1371 The space for a variable-length array is deallocated as soon as the array
1372 name's scope ends. (If you use both variable-length arrays and
1373 @code{alloca} in the same function, deallocation of a variable-length array
1374 will also deallocate anything more recently allocated with @code{alloca}.)
1376 You can also use variable-length arrays as arguments to functions:
1380 tester (int len, char data[len][len])
1386 The length of an array is computed once when the storage is allocated
1387 and is remembered for the scope of the array in case you access it with
1390 If you want to pass the array first and the length afterward, you can
1391 use a forward declaration in the parameter list---another GNU extension.
1395 tester (int len; char data[len][len], int len)
1401 @cindex parameter forward declaration
1402 The @samp{int len} before the semicolon is a @dfn{parameter forward
1403 declaration}, and it serves the purpose of making the name @code{len}
1404 known when the declaration of @code{data} is parsed.
1406 You can write any number of such parameter forward declarations in the
1407 parameter list. They can be separated by commas or semicolons, but the
1408 last one must end with a semicolon, which is followed by the ``real''
1409 parameter declarations. Each forward declaration must match a ``real''
1410 declaration in parameter name and data type. ISO C99 does not support
1411 parameter forward declarations.
1413 @node Variadic Macros
1414 @section Macros with a Variable Number of Arguments.
1415 @cindex variable number of arguments
1416 @cindex macro with variable arguments
1417 @cindex rest argument (in macro)
1418 @cindex variadic macros
1420 In the ISO C standard of 1999, a macro can be declared to accept a
1421 variable number of arguments much as a function can. The syntax for
1422 defining the macro is similar to that of a function. Here is an
1426 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1429 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1430 such a macro, it represents the zero or more tokens until the closing
1431 parenthesis that ends the invocation, including any commas. This set of
1432 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1433 wherever it appears. See the CPP manual for more information.
1435 GCC has long supported variadic macros, and used a different syntax that
1436 allowed you to give a name to the variable arguments just like any other
1437 argument. Here is an example:
1440 #define debug(format, args...) fprintf (stderr, format, args)
1443 This is in all ways equivalent to the ISO C example above, but arguably
1444 more readable and descriptive.
1446 GNU CPP has two further variadic macro extensions, and permits them to
1447 be used with either of the above forms of macro definition.
1449 In standard C, you are not allowed to leave the variable argument out
1450 entirely; but you are allowed to pass an empty argument. For example,
1451 this invocation is invalid in ISO C, because there is no comma after
1458 GNU CPP permits you to completely omit the variable arguments in this
1459 way. In the above examples, the compiler would complain, though since
1460 the expansion of the macro still has the extra comma after the format
1463 To help solve this problem, CPP behaves specially for variable arguments
1464 used with the token paste operator, @samp{##}. If instead you write
1467 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1470 and if the variable arguments are omitted or empty, the @samp{##}
1471 operator causes the preprocessor to remove the comma before it. If you
1472 do provide some variable arguments in your macro invocation, GNU CPP
1473 does not complain about the paste operation and instead places the
1474 variable arguments after the comma. Just like any other pasted macro
1475 argument, these arguments are not macro expanded.
1477 @node Escaped Newlines
1478 @section Slightly Looser Rules for Escaped Newlines
1479 @cindex escaped newlines
1480 @cindex newlines (escaped)
1482 Recently, the preprocessor has relaxed its treatment of escaped
1483 newlines. Previously, the newline had to immediately follow a
1484 backslash. The current implementation allows whitespace in the form
1485 of spaces, horizontal and vertical tabs, and form feeds between the
1486 backslash and the subsequent newline. The preprocessor issues a
1487 warning, but treats it as a valid escaped newline and combines the two
1488 lines to form a single logical line. This works within comments and
1489 tokens, as well as between tokens. Comments are @emph{not} treated as
1490 whitespace for the purposes of this relaxation, since they have not
1491 yet been replaced with spaces.
1494 @section Non-Lvalue Arrays May Have Subscripts
1495 @cindex subscripting
1496 @cindex arrays, non-lvalue
1498 @cindex subscripting and function values
1499 In ISO C99, arrays that are not lvalues still decay to pointers, and
1500 may be subscripted, although they may not be modified or used after
1501 the next sequence point and the unary @samp{&} operator may not be
1502 applied to them. As an extension, GCC allows such arrays to be
1503 subscripted in C89 mode, though otherwise they do not decay to
1504 pointers outside C99 mode. For example,
1505 this is valid in GNU C though not valid in C89:
1509 struct foo @{int a[4];@};
1515 return f().a[index];
1521 @section Arithmetic on @code{void}- and Function-Pointers
1522 @cindex void pointers, arithmetic
1523 @cindex void, size of pointer to
1524 @cindex function pointers, arithmetic
1525 @cindex function, size of pointer to
1527 In GNU C, addition and subtraction operations are supported on pointers to
1528 @code{void} and on pointers to functions. This is done by treating the
1529 size of a @code{void} or of a function as 1.
1531 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1532 and on function types, and returns 1.
1534 @opindex Wpointer-arith
1535 The option @option{-Wpointer-arith} requests a warning if these extensions
1539 @section Non-Constant Initializers
1540 @cindex initializers, non-constant
1541 @cindex non-constant initializers
1543 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1544 automatic variable are not required to be constant expressions in GNU C@.
1545 Here is an example of an initializer with run-time varying elements:
1548 foo (float f, float g)
1550 float beat_freqs[2] = @{ f-g, f+g @};
1555 @node Compound Literals
1556 @section Compound Literals
1557 @cindex constructor expressions
1558 @cindex initializations in expressions
1559 @cindex structures, constructor expression
1560 @cindex expressions, constructor
1561 @cindex compound literals
1562 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1564 ISO C99 supports compound literals. A compound literal looks like
1565 a cast containing an initializer. Its value is an object of the
1566 type specified in the cast, containing the elements specified in
1567 the initializer; it is an lvalue. As an extension, GCC supports
1568 compound literals in C89 mode and in C++.
1570 Usually, the specified type is a structure. Assume that
1571 @code{struct foo} and @code{structure} are declared as shown:
1574 struct foo @{int a; char b[2];@} structure;
1578 Here is an example of constructing a @code{struct foo} with a compound literal:
1581 structure = ((struct foo) @{x + y, 'a', 0@});
1585 This is equivalent to writing the following:
1589 struct foo temp = @{x + y, 'a', 0@};
1594 You can also construct an array. If all the elements of the compound literal
1595 are (made up of) simple constant expressions, suitable for use in
1596 initializers of objects of static storage duration, then the compound
1597 literal can be coerced to a pointer to its first element and used in
1598 such an initializer, as shown here:
1601 char **foo = (char *[]) @{ "x", "y", "z" @};
1604 Compound literals for scalar types and union types are is
1605 also allowed, but then the compound literal is equivalent
1608 As a GNU extension, GCC allows initialization of objects with static storage
1609 duration by compound literals (which is not possible in ISO C99, because
1610 the initializer is not a constant).
1611 It is handled as if the object was initialized only with the bracket
1612 enclosed list if compound literal's and object types match.
1613 The initializer list of the compound literal must be constant.
1614 If the object being initialized has array type of unknown size, the size is
1615 determined by compound literal size.
1618 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1619 static int y[] = (int []) @{1, 2, 3@};
1620 static int z[] = (int [3]) @{1@};
1624 The above lines are equivalent to the following:
1626 static struct foo x = @{1, 'a', 'b'@};
1627 static int y[] = @{1, 2, 3@};
1628 static int z[] = @{1, 0, 0@};
1631 @node Designated Inits
1632 @section Designated Initializers
1633 @cindex initializers with labeled elements
1634 @cindex labeled elements in initializers
1635 @cindex case labels in initializers
1636 @cindex designated initializers
1638 Standard C89 requires the elements of an initializer to appear in a fixed
1639 order, the same as the order of the elements in the array or structure
1642 In ISO C99 you can give the elements in any order, specifying the array
1643 indices or structure field names they apply to, and GNU C allows this as
1644 an extension in C89 mode as well. This extension is not
1645 implemented in GNU C++.
1647 To specify an array index, write
1648 @samp{[@var{index}] =} before the element value. For example,
1651 int a[6] = @{ [4] = 29, [2] = 15 @};
1658 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1662 The index values must be constant expressions, even if the array being
1663 initialized is automatic.
1665 An alternative syntax for this which has been obsolete since GCC 2.5 but
1666 GCC still accepts is to write @samp{[@var{index}]} before the element
1667 value, with no @samp{=}.
1669 To initialize a range of elements to the same value, write
1670 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1671 extension. For example,
1674 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1678 If the value in it has side-effects, the side-effects will happen only once,
1679 not for each initialized field by the range initializer.
1682 Note that the length of the array is the highest value specified
1685 In a structure initializer, specify the name of a field to initialize
1686 with @samp{.@var{fieldname} =} before the element value. For example,
1687 given the following structure,
1690 struct point @{ int x, y; @};
1694 the following initialization
1697 struct point p = @{ .y = yvalue, .x = xvalue @};
1704 struct point p = @{ xvalue, yvalue @};
1707 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1708 @samp{@var{fieldname}:}, as shown here:
1711 struct point p = @{ y: yvalue, x: xvalue @};
1715 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1716 @dfn{designator}. You can also use a designator (or the obsolete colon
1717 syntax) when initializing a union, to specify which element of the union
1718 should be used. For example,
1721 union foo @{ int i; double d; @};
1723 union foo f = @{ .d = 4 @};
1727 will convert 4 to a @code{double} to store it in the union using
1728 the second element. By contrast, casting 4 to type @code{union foo}
1729 would store it into the union as the integer @code{i}, since it is
1730 an integer. (@xref{Cast to Union}.)
1732 You can combine this technique of naming elements with ordinary C
1733 initialization of successive elements. Each initializer element that
1734 does not have a designator applies to the next consecutive element of the
1735 array or structure. For example,
1738 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1745 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1748 Labeling the elements of an array initializer is especially useful
1749 when the indices are characters or belong to an @code{enum} type.
1754 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1755 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1758 @cindex designator lists
1759 You can also write a series of @samp{.@var{fieldname}} and
1760 @samp{[@var{index}]} designators before an @samp{=} to specify a
1761 nested subobject to initialize; the list is taken relative to the
1762 subobject corresponding to the closest surrounding brace pair. For
1763 example, with the @samp{struct point} declaration above:
1766 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1770 If the same field is initialized multiple times, it will have value from
1771 the last initialization. If any such overridden initialization has
1772 side-effect, it is unspecified whether the side-effect happens or not.
1773 Currently, GCC will discard them and issue a warning.
1776 @section Case Ranges
1778 @cindex ranges in case statements
1780 You can specify a range of consecutive values in a single @code{case} label,
1784 case @var{low} ... @var{high}:
1788 This has the same effect as the proper number of individual @code{case}
1789 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1791 This feature is especially useful for ranges of ASCII character codes:
1797 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1798 it may be parsed wrong when you use it with integer values. For example,
1813 @section Cast to a Union Type
1814 @cindex cast to a union
1815 @cindex union, casting to a
1817 A cast to union type is similar to other casts, except that the type
1818 specified is a union type. You can specify the type either with
1819 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1820 a constructor though, not a cast, and hence does not yield an lvalue like
1821 normal casts. (@xref{Compound Literals}.)
1823 The types that may be cast to the union type are those of the members
1824 of the union. Thus, given the following union and variables:
1827 union foo @{ int i; double d; @};
1833 both @code{x} and @code{y} can be cast to type @code{union foo}.
1835 Using the cast as the right-hand side of an assignment to a variable of
1836 union type is equivalent to storing in a member of the union:
1841 u = (union foo) x @equiv{} u.i = x
1842 u = (union foo) y @equiv{} u.d = y
1845 You can also use the union cast as a function argument:
1848 void hack (union foo);
1850 hack ((union foo) x);
1853 @node Mixed Declarations
1854 @section Mixed Declarations and Code
1855 @cindex mixed declarations and code
1856 @cindex declarations, mixed with code
1857 @cindex code, mixed with declarations
1859 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1860 within compound statements. As an extension, GCC also allows this in
1861 C89 mode. For example, you could do:
1870 Each identifier is visible from where it is declared until the end of
1871 the enclosing block.
1873 @node Function Attributes
1874 @section Declaring Attributes of Functions
1875 @cindex function attributes
1876 @cindex declaring attributes of functions
1877 @cindex functions that never return
1878 @cindex functions that have no side effects
1879 @cindex functions in arbitrary sections
1880 @cindex functions that behave like malloc
1881 @cindex @code{volatile} applied to function
1882 @cindex @code{const} applied to function
1883 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1884 @cindex functions with non-null pointer arguments
1885 @cindex functions that are passed arguments in registers on the 386
1886 @cindex functions that pop the argument stack on the 386
1887 @cindex functions that do not pop the argument stack on the 386
1889 In GNU C, you declare certain things about functions called in your program
1890 which help the compiler optimize function calls and check your code more
1893 The keyword @code{__attribute__} allows you to specify special
1894 attributes when making a declaration. This keyword is followed by an
1895 attribute specification inside double parentheses. The following
1896 attributes are currently defined for functions on all targets:
1897 @code{noreturn}, @code{noinline}, @code{always_inline},
1898 @code{pure}, @code{const}, @code{nothrow},
1899 @code{format}, @code{format_arg}, @code{no_instrument_function},
1900 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1901 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
1902 @code{alias}, @code{warn_unused_result} and @code{nonnull}. Several other
1903 attributes are defined for functions on particular target systems. Other
1904 attributes, including @code{section} are supported for variables declarations
1905 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1907 You may also specify attributes with @samp{__} preceding and following
1908 each keyword. This allows you to use them in header files without
1909 being concerned about a possible macro of the same name. For example,
1910 you may use @code{__noreturn__} instead of @code{noreturn}.
1912 @xref{Attribute Syntax}, for details of the exact syntax for using
1916 @c Keep this table alphabetized by attribute name. Treat _ as space.
1918 @item alias ("@var{target}")
1919 @cindex @code{alias} attribute
1920 The @code{alias} attribute causes the declaration to be emitted as an
1921 alias for another symbol, which must be specified. For instance,
1924 void __f () @{ /* @r{Do something.} */; @}
1925 void f () __attribute__ ((weak, alias ("__f")));
1928 declares @samp{f} to be a weak alias for @samp{__f}. In C++, the
1929 mangled name for the target must be used.
1931 Not all target machines support this attribute.
1934 @cindex @code{always_inline} function attribute
1935 Generally, functions are not inlined unless optimization is specified.
1936 For functions declared inline, this attribute inlines the function even
1937 if no optimization level was specified.
1940 @cindex functions that do pop the argument stack on the 386
1942 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1943 assume that the calling function will pop off the stack space used to
1944 pass arguments. This is
1945 useful to override the effects of the @option{-mrtd} switch.
1948 @cindex @code{const} function attribute
1949 Many functions do not examine any values except their arguments, and
1950 have no effects except the return value. Basically this is just slightly
1951 more strict class than the @code{pure} attribute above, since function is not
1952 allowed to read global memory.
1954 @cindex pointer arguments
1955 Note that a function that has pointer arguments and examines the data
1956 pointed to must @emph{not} be declared @code{const}. Likewise, a
1957 function that calls a non-@code{const} function usually must not be
1958 @code{const}. It does not make sense for a @code{const} function to
1961 The attribute @code{const} is not implemented in GCC versions earlier
1962 than 2.5. An alternative way to declare that a function has no side
1963 effects, which works in the current version and in some older versions,
1967 typedef int intfn ();
1969 extern const intfn square;
1972 This approach does not work in GNU C++ from 2.6.0 on, since the language
1973 specifies that the @samp{const} must be attached to the return value.
1977 @cindex @code{constructor} function attribute
1978 @cindex @code{destructor} function attribute
1979 The @code{constructor} attribute causes the function to be called
1980 automatically before execution enters @code{main ()}. Similarly, the
1981 @code{destructor} attribute causes the function to be called
1982 automatically after @code{main ()} has completed or @code{exit ()} has
1983 been called. Functions with these attributes are useful for
1984 initializing data that will be used implicitly during the execution of
1987 These attributes are not currently implemented for Objective-C@.
1990 @cindex @code{deprecated} attribute.
1991 The @code{deprecated} attribute results in a warning if the function
1992 is used anywhere in the source file. This is useful when identifying
1993 functions that are expected to be removed in a future version of a
1994 program. The warning also includes the location of the declaration
1995 of the deprecated function, to enable users to easily find further
1996 information about why the function is deprecated, or what they should
1997 do instead. Note that the warnings only occurs for uses:
2000 int old_fn () __attribute__ ((deprecated));
2002 int (*fn_ptr)() = old_fn;
2005 results in a warning on line 3 but not line 2.
2007 The @code{deprecated} attribute can also be used for variables and
2008 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2011 @cindex @code{__declspec(dllexport)}
2012 On Microsoft Windows targets and Symbian targets the @code{dllexport}
2013 attribute causes the compiler to provide a global pointer to a pointer
2014 in a dll, so that it can be referenced with the @code{dllimport}
2015 attribute. The pointer name is formed by combining @code{_imp__} and
2016 the function or variable name.
2018 Currently, the @code{dllexport}attribute is ignored for inlined
2019 functions, but export can be forced by using the
2020 @option{-fkeep-inline-functions} flag. The attribute is also ignored for
2023 When applied to C++ classes. the attribute marks defined non-inlined
2024 member functions and static data members as exports. Static consts
2025 initialized in-class are not marked unless they are also defined
2028 On cygwin, mingw, arm-pe and sh-symbianelf targets,
2029 @code{__declspec(dllexport)} is recognized as a synonym for
2030 @code{__attribute__ ((dllexport))} for compatibility with other
2031 Microsoft Windows and Symbian compilers.
2033 For Microsoft Windows targets there are alternative methods for
2034 including the symbol in the dll's export table such as using a
2035 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
2036 the @option{--export-all} linker flag.
2039 @cindex @code{__declspec(dllimport)}
2040 On Microsoft Windows and Symbian targets, the @code{dllimport}
2041 attribute causes the compiler to reference a function or variable via
2042 a global pointer to a pointer that is set up by the Microsoft Windows
2043 dll library. The pointer name is formed by combining @code{_imp__} and
2044 the function or variable name. The attribute implies @code{extern}
2047 Currently, the attribute is ignored for inlined functions. If the
2048 attribute is applied to a symbol @emph{definition}, an error is reported.
2049 If a symbol previously declared @code{dllimport} is later defined, the
2050 attribute is ignored in subsequent references, and a warning is emitted.
2051 The attribute is also overridden by a subsequent declaration as
2054 When applied to C++ classes, the attribute marks non-inlined
2055 member functions and static data members as imports. However, the
2056 attribute is ignored for virtual methods to allow creation of vtables
2059 For Symbian targets the @code{dllimport} attribute also has another
2060 affect - it can cause the vtable and run-time type information for a
2061 class to be exported. This happens when the class has a dllimport'ed
2062 constructor or a non-inline, non-pure virtual function and, for either
2063 of those two conditions, the class also has a inline constructor or
2064 destructor and has a key function that is defined in the current
2067 On cygwin, mingw, arm-pe sh-symbianelf targets,
2068 @code{__declspec(dllimport)} is recognized as a synonym for
2069 @code{__attribute__ ((dllimport))} for compatibility with other
2070 Microsoft Windows and Symbian compilers.
2072 For Microsoft Windows based targets the use of the @code{dllimport}
2073 attribute on functions is not necessary, but provides a small
2074 performance benefit by eliminating a thunk in the dll. The use of the
2075 @code{dllimport} attribute on imported variables was required on older
2076 versions of GNU ld, but can now be avoided by passing the
2077 @option{--enable-auto-import} switch to ld. As with functions, using
2078 the attribute for a variable eliminates a thunk in the dll.
2080 One drawback to using this attribute is that a pointer to a function or
2081 variable marked as dllimport cannot be used as a constant address. The
2082 attribute can be disabled for functions by setting the
2083 @option{-mnop-fun-dllimport} flag.
2086 @cindex eight bit data on the H8/300, H8/300H, and H8S
2087 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2088 variable should be placed into the eight bit data section.
2089 The compiler will generate more efficient code for certain operations
2090 on data in the eight bit data area. Note the eight bit data area is limited to
2093 You must use GAS and GLD from GNU binutils version 2.7 or later for
2094 this attribute to work correctly.
2097 @cindex functions which handle memory bank switching
2098 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
2099 use a calling convention that takes care of switching memory banks when
2100 entering and leaving a function. This calling convention is also the
2101 default when using the @option{-mlong-calls} option.
2103 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
2104 to call and return from a function.
2106 On 68HC11 the compiler will generate a sequence of instructions
2107 to invoke a board-specific routine to switch the memory bank and call the
2108 real function. The board-specific routine simulates a @code{call}.
2109 At the end of a function, it will jump to a board-specific routine
2110 instead of using @code{rts}. The board-specific return routine simulates
2114 @cindex functions that pop the argument stack on the 386
2115 On the Intel 386, the @code{fastcall} attribute causes the compiler to
2116 pass the first two arguments in the registers ECX and EDX. Subsequent
2117 arguments are passed on the stack. The called function will pop the
2118 arguments off the stack. If the number of arguments is variable all
2119 arguments are pushed on the stack.
2121 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2122 @cindex @code{format} function attribute
2124 The @code{format} attribute specifies that a function takes @code{printf},
2125 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
2126 should be type-checked against a format string. For example, the
2131 my_printf (void *my_object, const char *my_format, ...)
2132 __attribute__ ((format (printf, 2, 3)));
2136 causes the compiler to check the arguments in calls to @code{my_printf}
2137 for consistency with the @code{printf} style format string argument
2140 The parameter @var{archetype} determines how the format string is
2141 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
2142 or @code{strfmon}. (You can also use @code{__printf__},
2143 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
2144 parameter @var{string-index} specifies which argument is the format
2145 string argument (starting from 1), while @var{first-to-check} is the
2146 number of the first argument to check against the format string. For
2147 functions where the arguments are not available to be checked (such as
2148 @code{vprintf}), specify the third parameter as zero. In this case the
2149 compiler only checks the format string for consistency. For
2150 @code{strftime} formats, the third parameter is required to be zero.
2151 Since non-static C++ methods have an implicit @code{this} argument, the
2152 arguments of such methods should be counted from two, not one, when
2153 giving values for @var{string-index} and @var{first-to-check}.
2155 In the example above, the format string (@code{my_format}) is the second
2156 argument of the function @code{my_print}, and the arguments to check
2157 start with the third argument, so the correct parameters for the format
2158 attribute are 2 and 3.
2160 @opindex ffreestanding
2161 The @code{format} attribute allows you to identify your own functions
2162 which take format strings as arguments, so that GCC can check the
2163 calls to these functions for errors. The compiler always (unless
2164 @option{-ffreestanding} is used) checks formats
2165 for the standard library functions @code{printf}, @code{fprintf},
2166 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2167 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2168 warnings are requested (using @option{-Wformat}), so there is no need to
2169 modify the header file @file{stdio.h}. In C99 mode, the functions
2170 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2171 @code{vsscanf} are also checked. Except in strictly conforming C
2172 standard modes, the X/Open function @code{strfmon} is also checked as
2173 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2174 @xref{C Dialect Options,,Options Controlling C Dialect}.
2176 @item format_arg (@var{string-index})
2177 @cindex @code{format_arg} function attribute
2178 @opindex Wformat-nonliteral
2179 The @code{format_arg} attribute specifies that a function takes a format
2180 string for a @code{printf}, @code{scanf}, @code{strftime} or
2181 @code{strfmon} style function and modifies it (for example, to translate
2182 it into another language), so the result can be passed to a
2183 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2184 function (with the remaining arguments to the format function the same
2185 as they would have been for the unmodified string). For example, the
2190 my_dgettext (char *my_domain, const char *my_format)
2191 __attribute__ ((format_arg (2)));
2195 causes the compiler to check the arguments in calls to a @code{printf},
2196 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2197 format string argument is a call to the @code{my_dgettext} function, for
2198 consistency with the format string argument @code{my_format}. If the
2199 @code{format_arg} attribute had not been specified, all the compiler
2200 could tell in such calls to format functions would be that the format
2201 string argument is not constant; this would generate a warning when
2202 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2203 without the attribute.
2205 The parameter @var{string-index} specifies which argument is the format
2206 string argument (starting from one). Since non-static C++ methods have
2207 an implicit @code{this} argument, the arguments of such methods should
2208 be counted from two.
2210 The @code{format-arg} attribute allows you to identify your own
2211 functions which modify format strings, so that GCC can check the
2212 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2213 type function whose operands are a call to one of your own function.
2214 The compiler always treats @code{gettext}, @code{dgettext}, and
2215 @code{dcgettext} in this manner except when strict ISO C support is
2216 requested by @option{-ansi} or an appropriate @option{-std} option, or
2217 @option{-ffreestanding} is used. @xref{C Dialect Options,,Options
2218 Controlling C Dialect}.
2220 @item function_vector
2221 @cindex calling functions through the function vector on the H8/300 processors
2222 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2223 function should be called through the function vector. Calling a
2224 function through the function vector will reduce code size, however;
2225 the function vector has a limited size (maximum 128 entries on the H8/300
2226 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2228 You must use GAS and GLD from GNU binutils version 2.7 or later for
2229 this attribute to work correctly.
2232 @cindex interrupt handler functions
2233 Use this attribute on the ARM, AVR, C4x, M32R/D and Xstormy16 ports to indicate
2234 that the specified function is an interrupt handler. The compiler will
2235 generate function entry and exit sequences suitable for use in an
2236 interrupt handler when this attribute is present.
2238 Note, interrupt handlers for the m68k, H8/300, H8/300H, H8S, and SH processors
2239 can be specified via the @code{interrupt_handler} attribute.
2241 Note, on the AVR, interrupts will be enabled inside the function.
2243 Note, for the ARM, you can specify the kind of interrupt to be handled by
2244 adding an optional parameter to the interrupt attribute like this:
2247 void f () __attribute__ ((interrupt ("IRQ")));
2250 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2252 @item interrupt_handler
2253 @cindex interrupt handler functions on the m68k, H8/300 and SH processors
2254 Use this attribute on the m68k, H8/300, H8/300H, H8S, and SH to indicate that
2255 the specified function is an interrupt handler. The compiler will generate
2256 function entry and exit sequences suitable for use in an interrupt
2257 handler when this attribute is present.
2259 @item long_call/short_call
2260 @cindex indirect calls on ARM
2261 This attribute specifies how a particular function is called on
2262 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2263 command line switch and @code{#pragma long_calls} settings. The
2264 @code{long_call} attribute causes the compiler to always call the
2265 function by first loading its address into a register and then using the
2266 contents of that register. The @code{short_call} attribute always places
2267 the offset to the function from the call site into the @samp{BL}
2268 instruction directly.
2270 @item longcall/shortcall
2271 @cindex functions called via pointer on the RS/6000 and PowerPC
2272 On the RS/6000 and PowerPC, the @code{longcall} attribute causes the
2273 compiler to always call this function via a pointer, just as it would if
2274 the @option{-mlongcall} option had been specified. The @code{shortcall}
2275 attribute causes the compiler not to do this. These attributes override
2276 both the @option{-mlongcall} switch and the @code{#pragma longcall}
2279 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2280 calls are necessary.
2283 @cindex @code{malloc} attribute
2284 The @code{malloc} attribute is used to tell the compiler that a function
2285 may be treated as if any non-@code{NULL} pointer it returns cannot
2286 alias any other pointer valid when the function returns.
2287 This will often improve optimization.
2288 Standard functions with this property include @code{malloc} and
2289 @code{calloc}. @code{realloc}-like functions have this property as
2290 long as the old pointer is never referred to (including comparing it
2291 to the new pointer) after the function returns a non-@code{NULL}
2294 @item model (@var{model-name})
2295 @cindex function addressability on the M32R/D
2296 @cindex variable addressability on the IA-64
2298 On the M32R/D, use this attribute to set the addressability of an
2299 object, and of the code generated for a function. The identifier
2300 @var{model-name} is one of @code{small}, @code{medium}, or
2301 @code{large}, representing each of the code models.
2303 Small model objects live in the lower 16MB of memory (so that their
2304 addresses can be loaded with the @code{ld24} instruction), and are
2305 callable with the @code{bl} instruction.
2307 Medium model objects may live anywhere in the 32-bit address space (the
2308 compiler will generate @code{seth/add3} instructions to load their addresses),
2309 and are callable with the @code{bl} instruction.
2311 Large model objects may live anywhere in the 32-bit address space (the
2312 compiler will generate @code{seth/add3} instructions to load their addresses),
2313 and may not be reachable with the @code{bl} instruction (the compiler will
2314 generate the much slower @code{seth/add3/jl} instruction sequence).
2316 On IA-64, use this attribute to set the addressability of an object.
2317 At present, the only supported identifier for @var{model-name} is
2318 @code{small}, indicating addressability via ``small'' (22-bit)
2319 addresses (so that their addresses can be loaded with the @code{addl}
2320 instruction). Caveat: such addressing is by definition not position
2321 independent and hence this attribute must not be used for objects
2322 defined by shared libraries.
2325 @cindex function without a prologue/epilogue code
2326 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
2327 specified function does not need prologue/epilogue sequences generated by
2328 the compiler. It is up to the programmer to provide these sequences.
2331 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2332 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2333 use the normal calling convention based on @code{jsr} and @code{rts}.
2334 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2337 @item no_instrument_function
2338 @cindex @code{no_instrument_function} function attribute
2339 @opindex finstrument-functions
2340 If @option{-finstrument-functions} is given, profiling function calls will
2341 be generated at entry and exit of most user-compiled functions.
2342 Functions with this attribute will not be so instrumented.
2345 @cindex @code{noinline} function attribute
2346 This function attribute prevents a function from being considered for
2349 @item nonnull (@var{arg-index}, @dots{})
2350 @cindex @code{nonnull} function attribute
2351 The @code{nonnull} attribute specifies that some function parameters should
2352 be non-null pointers. For instance, the declaration:
2356 my_memcpy (void *dest, const void *src, size_t len)
2357 __attribute__((nonnull (1, 2)));
2361 causes the compiler to check that, in calls to @code{my_memcpy},
2362 arguments @var{dest} and @var{src} are non-null. If the compiler
2363 determines that a null pointer is passed in an argument slot marked
2364 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2365 is issued. The compiler may also choose to make optimizations based
2366 on the knowledge that certain function arguments will not be null.
2368 If no argument index list is given to the @code{nonnull} attribute,
2369 all pointer arguments are marked as non-null. To illustrate, the
2370 following declaration is equivalent to the previous example:
2374 my_memcpy (void *dest, const void *src, size_t len)
2375 __attribute__((nonnull));
2379 @cindex @code{noreturn} function attribute
2380 A few standard library functions, such as @code{abort} and @code{exit},
2381 cannot return. GCC knows this automatically. Some programs define
2382 their own functions that never return. You can declare them
2383 @code{noreturn} to tell the compiler this fact. For example,
2387 void fatal () __attribute__ ((noreturn));
2390 fatal (/* @r{@dots{}} */)
2392 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2398 The @code{noreturn} keyword tells the compiler to assume that
2399 @code{fatal} cannot return. It can then optimize without regard to what
2400 would happen if @code{fatal} ever did return. This makes slightly
2401 better code. More importantly, it helps avoid spurious warnings of
2402 uninitialized variables.
2404 The @code{noreturn} keyword does not affect the exceptional path when that
2405 applies: a @code{noreturn}-marked function may still return to the caller
2406 by throwing an exception.
2408 Do not assume that registers saved by the calling function are
2409 restored before calling the @code{noreturn} function.
2411 It does not make sense for a @code{noreturn} function to have a return
2412 type other than @code{void}.
2414 The attribute @code{noreturn} is not implemented in GCC versions
2415 earlier than 2.5. An alternative way to declare that a function does
2416 not return, which works in the current version and in some older
2417 versions, is as follows:
2420 typedef void voidfn ();
2422 volatile voidfn fatal;
2426 @cindex @code{nothrow} function attribute
2427 The @code{nothrow} attribute is used to inform the compiler that a
2428 function cannot throw an exception. For example, most functions in
2429 the standard C library can be guaranteed not to throw an exception
2430 with the notable exceptions of @code{qsort} and @code{bsearch} that
2431 take function pointer arguments. The @code{nothrow} attribute is not
2432 implemented in GCC versions earlier than 3.2.
2435 @cindex @code{pure} function attribute
2436 Many functions have no effects except the return value and their
2437 return value depends only on the parameters and/or global variables.
2438 Such a function can be subject
2439 to common subexpression elimination and loop optimization just as an
2440 arithmetic operator would be. These functions should be declared
2441 with the attribute @code{pure}. For example,
2444 int square (int) __attribute__ ((pure));
2448 says that the hypothetical function @code{square} is safe to call
2449 fewer times than the program says.
2451 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2452 Interesting non-pure functions are functions with infinite loops or those
2453 depending on volatile memory or other system resource, that may change between
2454 two consecutive calls (such as @code{feof} in a multithreading environment).
2456 The attribute @code{pure} is not implemented in GCC versions earlier
2459 @item regparm (@var{number})
2460 @cindex @code{regparm} attribute
2461 @cindex functions that are passed arguments in registers on the 386
2462 On the Intel 386, the @code{regparm} attribute causes the compiler to
2463 pass up to @var{number} integer arguments in registers EAX,
2464 EDX, and ECX instead of on the stack. Functions that take a
2465 variable number of arguments will continue to be passed all of their
2466 arguments on the stack.
2468 Beware that on some ELF systems this attribute is unsuitable for
2469 global functions in shared libraries with lazy binding (which is the
2470 default). Lazy binding will send the first call via resolving code in
2471 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2472 per the standard calling conventions. Solaris 8 is affected by this.
2473 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2474 safe since the loaders there save all registers. (Lazy binding can be
2475 disabled with the linker or the loader if desired, to avoid the
2479 @cindex save all registers on the H8/300, H8/300H, and H8S
2480 Use this attribute on the H8/300, H8/300H, and H8S to indicate that
2481 all registers except the stack pointer should be saved in the prologue
2482 regardless of whether they are used or not.
2484 @item section ("@var{section-name}")
2485 @cindex @code{section} function attribute
2486 Normally, the compiler places the code it generates in the @code{text} section.
2487 Sometimes, however, you need additional sections, or you need certain
2488 particular functions to appear in special sections. The @code{section}
2489 attribute specifies that a function lives in a particular section.
2490 For example, the declaration:
2493 extern void foobar (void) __attribute__ ((section ("bar")));
2497 puts the function @code{foobar} in the @code{bar} section.
2499 Some file formats do not support arbitrary sections so the @code{section}
2500 attribute is not available on all platforms.
2501 If you need to map the entire contents of a module to a particular
2502 section, consider using the facilities of the linker instead.
2505 See long_call/short_call.
2508 See longcall/shortcall.
2511 @cindex signal handler functions on the AVR processors
2512 Use this attribute on the AVR to indicate that the specified
2513 function is a signal handler. The compiler will generate function
2514 entry and exit sequences suitable for use in a signal handler when this
2515 attribute is present. Interrupts will be disabled inside the function.
2518 Use this attribute on the SH to indicate an @code{interrupt_handler}
2519 function should switch to an alternate stack. It expects a string
2520 argument that names a global variable holding the address of the
2525 void f () __attribute__ ((interrupt_handler,
2526 sp_switch ("alt_stack")));
2530 @cindex functions that pop the argument stack on the 386
2531 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2532 assume that the called function will pop off the stack space used to
2533 pass arguments, unless it takes a variable number of arguments.
2536 @cindex tiny data section on the H8/300H and H8S
2537 Use this attribute on the H8/300H and H8S to indicate that the specified
2538 variable should be placed into the tiny data section.
2539 The compiler will generate more efficient code for loads and stores
2540 on data in the tiny data section. Note the tiny data area is limited to
2541 slightly under 32kbytes of data.
2544 Use this attribute on the SH for an @code{interrupt_handler} to return using
2545 @code{trapa} instead of @code{rte}. This attribute expects an integer
2546 argument specifying the trap number to be used.
2549 @cindex @code{unused} attribute.
2550 This attribute, attached to a function, means that the function is meant
2551 to be possibly unused. GCC will not produce a warning for this
2555 @cindex @code{used} attribute.
2556 This attribute, attached to a function, means that code must be emitted
2557 for the function even if it appears that the function is not referenced.
2558 This is useful, for example, when the function is referenced only in
2561 @item visibility ("@var{visibility_type}")
2562 @cindex @code{visibility} attribute
2563 The @code{visibility} attribute on ELF targets causes the declaration
2564 to be emitted with default, hidden, protected or internal visibility.
2567 void __attribute__ ((visibility ("protected")))
2568 f () @{ /* @r{Do something.} */; @}
2569 int i __attribute__ ((visibility ("hidden")));
2572 See the ELF gABI for complete details, but the short story is:
2575 @c keep this list of visibilies in alphabetical order.
2578 Default visibility is the normal case for ELF. This value is
2579 available for the visibility attribute to override other options
2580 that may change the assumed visibility of symbols.
2583 Hidden visibility indicates that the symbol will not be placed into
2584 the dynamic symbol table, so no other @dfn{module} (executable or
2585 shared library) can reference it directly.
2588 Internal visibility is like hidden visibility, but with additional
2589 processor specific semantics. Unless otherwise specified by the psABI,
2590 GCC defines internal visibility to mean that the function is @emph{never}
2591 called from another module. Note that hidden symbols, while they cannot
2592 be referenced directly by other modules, can be referenced indirectly via
2593 function pointers. By indicating that a symbol cannot be called from
2594 outside the module, GCC may for instance omit the load of a PIC register
2595 since it is known that the calling function loaded the correct value.
2598 Protected visibility indicates that the symbol will be placed in the
2599 dynamic symbol table, but that references within the defining module
2600 will bind to the local symbol. That is, the symbol cannot be overridden
2605 Not all ELF targets support this attribute.
2607 @item warn_unused_result
2608 @cindex @code{warn_unused_result} attribute
2609 The @code{warn_unused_result} attribute causes a warning to be emitted
2610 if a caller of the function with this attribute does not use its
2611 return value. This is useful for functions where not checking
2612 the result is either a security problem or always a bug, such as
2616 int fn () __attribute__ ((warn_unused_result));
2619 if (fn () < 0) return -1;
2625 results in warning on line 5.
2628 @cindex @code{weak} attribute
2629 The @code{weak} attribute causes the declaration to be emitted as a weak
2630 symbol rather than a global. This is primarily useful in defining
2631 library functions which can be overridden in user code, though it can
2632 also be used with non-function declarations. Weak symbols are supported
2633 for ELF targets, and also for a.out targets when using the GNU assembler
2638 You can specify multiple attributes in a declaration by separating them
2639 by commas within the double parentheses or by immediately following an
2640 attribute declaration with another attribute declaration.
2642 @cindex @code{#pragma}, reason for not using
2643 @cindex pragma, reason for not using
2644 Some people object to the @code{__attribute__} feature, suggesting that
2645 ISO C's @code{#pragma} should be used instead. At the time
2646 @code{__attribute__} was designed, there were two reasons for not doing
2651 It is impossible to generate @code{#pragma} commands from a macro.
2654 There is no telling what the same @code{#pragma} might mean in another
2658 These two reasons applied to almost any application that might have been
2659 proposed for @code{#pragma}. It was basically a mistake to use
2660 @code{#pragma} for @emph{anything}.
2662 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2663 to be generated from macros. In addition, a @code{#pragma GCC}
2664 namespace is now in use for GCC-specific pragmas. However, it has been
2665 found convenient to use @code{__attribute__} to achieve a natural
2666 attachment of attributes to their corresponding declarations, whereas
2667 @code{#pragma GCC} is of use for constructs that do not naturally form
2668 part of the grammar. @xref{Other Directives,,Miscellaneous
2669 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2671 @node Attribute Syntax
2672 @section Attribute Syntax
2673 @cindex attribute syntax
2675 This section describes the syntax with which @code{__attribute__} may be
2676 used, and the constructs to which attribute specifiers bind, for the C
2677 language. Some details may vary for C++ and Objective-C@. Because of
2678 infelicities in the grammar for attributes, some forms described here
2679 may not be successfully parsed in all cases.
2681 There are some problems with the semantics of attributes in C++. For
2682 example, there are no manglings for attributes, although they may affect
2683 code generation, so problems may arise when attributed types are used in
2684 conjunction with templates or overloading. Similarly, @code{typeid}
2685 does not distinguish between types with different attributes. Support
2686 for attributes in C++ may be restricted in future to attributes on
2687 declarations only, but not on nested declarators.
2689 @xref{Function Attributes}, for details of the semantics of attributes
2690 applying to functions. @xref{Variable Attributes}, for details of the
2691 semantics of attributes applying to variables. @xref{Type Attributes},
2692 for details of the semantics of attributes applying to structure, union
2693 and enumerated types.
2695 An @dfn{attribute specifier} is of the form
2696 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2697 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2698 each attribute is one of the following:
2702 Empty. Empty attributes are ignored.
2705 A word (which may be an identifier such as @code{unused}, or a reserved
2706 word such as @code{const}).
2709 A word, followed by, in parentheses, parameters for the attribute.
2710 These parameters take one of the following forms:
2714 An identifier. For example, @code{mode} attributes use this form.
2717 An identifier followed by a comma and a non-empty comma-separated list
2718 of expressions. For example, @code{format} attributes use this form.
2721 A possibly empty comma-separated list of expressions. For example,
2722 @code{format_arg} attributes use this form with the list being a single
2723 integer constant expression, and @code{alias} attributes use this form
2724 with the list being a single string constant.
2728 An @dfn{attribute specifier list} is a sequence of one or more attribute
2729 specifiers, not separated by any other tokens.
2731 In GNU C, an attribute specifier list may appear after the colon following a
2732 label, other than a @code{case} or @code{default} label. The only
2733 attribute it makes sense to use after a label is @code{unused}. This
2734 feature is intended for code generated by programs which contains labels
2735 that may be unused but which is compiled with @option{-Wall}. It would
2736 not normally be appropriate to use in it human-written code, though it
2737 could be useful in cases where the code that jumps to the label is
2738 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2739 such placement of attribute lists, as it is permissible for a
2740 declaration, which could begin with an attribute list, to be labelled in
2741 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2742 does not arise there.
2744 An attribute specifier list may appear as part of a @code{struct},
2745 @code{union} or @code{enum} specifier. It may go either immediately
2746 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2747 the closing brace. It is ignored if the content of the structure, union
2748 or enumerated type is not defined in the specifier in which the
2749 attribute specifier list is used---that is, in usages such as
2750 @code{struct __attribute__((foo)) bar} with no following opening brace.
2751 Where attribute specifiers follow the closing brace, they are considered
2752 to relate to the structure, union or enumerated type defined, not to any
2753 enclosing declaration the type specifier appears in, and the type
2754 defined is not complete until after the attribute specifiers.
2755 @c Otherwise, there would be the following problems: a shift/reduce
2756 @c conflict between attributes binding the struct/union/enum and
2757 @c binding to the list of specifiers/qualifiers; and "aligned"
2758 @c attributes could use sizeof for the structure, but the size could be
2759 @c changed later by "packed" attributes.
2761 Otherwise, an attribute specifier appears as part of a declaration,
2762 counting declarations of unnamed parameters and type names, and relates
2763 to that declaration (which may be nested in another declaration, for
2764 example in the case of a parameter declaration), or to a particular declarator
2765 within a declaration. Where an
2766 attribute specifier is applied to a parameter declared as a function or
2767 an array, it should apply to the function or array rather than the
2768 pointer to which the parameter is implicitly converted, but this is not
2769 yet correctly implemented.
2771 Any list of specifiers and qualifiers at the start of a declaration may
2772 contain attribute specifiers, whether or not such a list may in that
2773 context contain storage class specifiers. (Some attributes, however,
2774 are essentially in the nature of storage class specifiers, and only make
2775 sense where storage class specifiers may be used; for example,
2776 @code{section}.) There is one necessary limitation to this syntax: the
2777 first old-style parameter declaration in a function definition cannot
2778 begin with an attribute specifier, because such an attribute applies to
2779 the function instead by syntax described below (which, however, is not
2780 yet implemented in this case). In some other cases, attribute
2781 specifiers are permitted by this grammar but not yet supported by the
2782 compiler. All attribute specifiers in this place relate to the
2783 declaration as a whole. In the obsolescent usage where a type of
2784 @code{int} is implied by the absence of type specifiers, such a list of
2785 specifiers and qualifiers may be an attribute specifier list with no
2786 other specifiers or qualifiers.
2788 An attribute specifier list may appear immediately before a declarator
2789 (other than the first) in a comma-separated list of declarators in a
2790 declaration of more than one identifier using a single list of
2791 specifiers and qualifiers. Such attribute specifiers apply
2792 only to the identifier before whose declarator they appear. For
2796 __attribute__((noreturn)) void d0 (void),
2797 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2802 the @code{noreturn} attribute applies to all the functions
2803 declared; the @code{format} attribute only applies to @code{d1}.
2805 An attribute specifier list may appear immediately before the comma,
2806 @code{=} or semicolon terminating the declaration of an identifier other
2807 than a function definition. At present, such attribute specifiers apply
2808 to the declared object or function, but in future they may attach to the
2809 outermost adjacent declarator. In simple cases there is no difference,
2810 but, for example, in
2813 void (****f)(void) __attribute__((noreturn));
2817 at present the @code{noreturn} attribute applies to @code{f}, which
2818 causes a warning since @code{f} is not a function, but in future it may
2819 apply to the function @code{****f}. The precise semantics of what
2820 attributes in such cases will apply to are not yet specified. Where an
2821 assembler name for an object or function is specified (@pxref{Asm
2822 Labels}), at present the attribute must follow the @code{asm}
2823 specification; in future, attributes before the @code{asm} specification
2824 may apply to the adjacent declarator, and those after it to the declared
2827 An attribute specifier list may, in future, be permitted to appear after
2828 the declarator in a function definition (before any old-style parameter
2829 declarations or the function body).
2831 Attribute specifiers may be mixed with type qualifiers appearing inside
2832 the @code{[]} of a parameter array declarator, in the C99 construct by
2833 which such qualifiers are applied to the pointer to which the array is
2834 implicitly converted. Such attribute specifiers apply to the pointer,
2835 not to the array, but at present this is not implemented and they are
2838 An attribute specifier list may appear at the start of a nested
2839 declarator. At present, there are some limitations in this usage: the
2840 attributes correctly apply to the declarator, but for most individual
2841 attributes the semantics this implies are not implemented.
2842 When attribute specifiers follow the @code{*} of a pointer
2843 declarator, they may be mixed with any type qualifiers present.
2844 The following describes the formal semantics of this syntax. It will make the
2845 most sense if you are familiar with the formal specification of
2846 declarators in the ISO C standard.
2848 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2849 D1}, where @code{T} contains declaration specifiers that specify a type
2850 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2851 contains an identifier @var{ident}. The type specified for @var{ident}
2852 for derived declarators whose type does not include an attribute
2853 specifier is as in the ISO C standard.
2855 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2856 and the declaration @code{T D} specifies the type
2857 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2858 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2859 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2861 If @code{D1} has the form @code{*
2862 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2863 declaration @code{T D} specifies the type
2864 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2865 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2866 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2872 void (__attribute__((noreturn)) ****f) (void);
2876 specifies the type ``pointer to pointer to pointer to pointer to
2877 non-returning function returning @code{void}''. As another example,
2880 char *__attribute__((aligned(8))) *f;
2884 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2885 Note again that this does not work with most attributes; for example,
2886 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2887 is not yet supported.
2889 For compatibility with existing code written for compiler versions that
2890 did not implement attributes on nested declarators, some laxity is
2891 allowed in the placing of attributes. If an attribute that only applies
2892 to types is applied to a declaration, it will be treated as applying to
2893 the type of that declaration. If an attribute that only applies to
2894 declarations is applied to the type of a declaration, it will be treated
2895 as applying to that declaration; and, for compatibility with code
2896 placing the attributes immediately before the identifier declared, such
2897 an attribute applied to a function return type will be treated as
2898 applying to the function type, and such an attribute applied to an array
2899 element type will be treated as applying to the array type. If an
2900 attribute that only applies to function types is applied to a
2901 pointer-to-function type, it will be treated as applying to the pointer
2902 target type; if such an attribute is applied to a function return type
2903 that is not a pointer-to-function type, it will be treated as applying
2904 to the function type.
2906 @node Function Prototypes
2907 @section Prototypes and Old-Style Function Definitions
2908 @cindex function prototype declarations
2909 @cindex old-style function definitions
2910 @cindex promotion of formal parameters
2912 GNU C extends ISO C to allow a function prototype to override a later
2913 old-style non-prototype definition. Consider the following example:
2916 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2923 /* @r{Prototype function declaration.} */
2924 int isroot P((uid_t));
2926 /* @r{Old-style function definition.} */
2928 isroot (x) /* ??? lossage here ??? */
2935 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2936 not allow this example, because subword arguments in old-style
2937 non-prototype definitions are promoted. Therefore in this example the
2938 function definition's argument is really an @code{int}, which does not
2939 match the prototype argument type of @code{short}.
2941 This restriction of ISO C makes it hard to write code that is portable
2942 to traditional C compilers, because the programmer does not know
2943 whether the @code{uid_t} type is @code{short}, @code{int}, or
2944 @code{long}. Therefore, in cases like these GNU C allows a prototype
2945 to override a later old-style definition. More precisely, in GNU C, a
2946 function prototype argument type overrides the argument type specified
2947 by a later old-style definition if the former type is the same as the
2948 latter type before promotion. Thus in GNU C the above example is
2949 equivalent to the following:
2962 GNU C++ does not support old-style function definitions, so this
2963 extension is irrelevant.
2966 @section C++ Style Comments
2968 @cindex C++ comments
2969 @cindex comments, C++ style
2971 In GNU C, you may use C++ style comments, which start with @samp{//} and
2972 continue until the end of the line. Many other C implementations allow
2973 such comments, and they are included in the 1999 C standard. However,
2974 C++ style comments are not recognized if you specify an @option{-std}
2975 option specifying a version of ISO C before C99, or @option{-ansi}
2976 (equivalent to @option{-std=c89}).
2979 @section Dollar Signs in Identifier Names
2981 @cindex dollar signs in identifier names
2982 @cindex identifier names, dollar signs in
2984 In GNU C, you may normally use dollar signs in identifier names.
2985 This is because many traditional C implementations allow such identifiers.
2986 However, dollar signs in identifiers are not supported on a few target
2987 machines, typically because the target assembler does not allow them.
2989 @node Character Escapes
2990 @section The Character @key{ESC} in Constants
2992 You can use the sequence @samp{\e} in a string or character constant to
2993 stand for the ASCII character @key{ESC}.
2996 @section Inquiring on Alignment of Types or Variables
2998 @cindex type alignment
2999 @cindex variable alignment
3001 The keyword @code{__alignof__} allows you to inquire about how an object
3002 is aligned, or the minimum alignment usually required by a type. Its
3003 syntax is just like @code{sizeof}.
3005 For example, if the target machine requires a @code{double} value to be
3006 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
3007 This is true on many RISC machines. On more traditional machine
3008 designs, @code{__alignof__ (double)} is 4 or even 2.
3010 Some machines never actually require alignment; they allow reference to any
3011 data type even at an odd address. For these machines, @code{__alignof__}
3012 reports the @emph{recommended} alignment of a type.
3014 If the operand of @code{__alignof__} is an lvalue rather than a type,
3015 its value is the required alignment for its type, taking into account
3016 any minimum alignment specified with GCC's @code{__attribute__}
3017 extension (@pxref{Variable Attributes}). For example, after this
3021 struct foo @{ int x; char y; @} foo1;
3025 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
3026 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
3028 It is an error to ask for the alignment of an incomplete type.
3030 @node Variable Attributes
3031 @section Specifying Attributes of Variables
3032 @cindex attribute of variables
3033 @cindex variable attributes
3035 The keyword @code{__attribute__} allows you to specify special
3036 attributes of variables or structure fields. This keyword is followed
3037 by an attribute specification inside double parentheses. Some
3038 attributes are currently defined generically for variables.
3039 Other attributes are defined for variables on particular target
3040 systems. Other attributes are available for functions
3041 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
3042 Other front ends might define more attributes
3043 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
3045 You may also specify attributes with @samp{__} preceding and following
3046 each keyword. This allows you to use them in header files without
3047 being concerned about a possible macro of the same name. For example,
3048 you may use @code{__aligned__} instead of @code{aligned}.
3050 @xref{Attribute Syntax}, for details of the exact syntax for using
3054 @cindex @code{aligned} attribute
3055 @item aligned (@var{alignment})
3056 This attribute specifies a minimum alignment for the variable or
3057 structure field, measured in bytes. For example, the declaration:
3060 int x __attribute__ ((aligned (16))) = 0;
3064 causes the compiler to allocate the global variable @code{x} on a
3065 16-byte boundary. On a 68040, this could be used in conjunction with
3066 an @code{asm} expression to access the @code{move16} instruction which
3067 requires 16-byte aligned operands.
3069 You can also specify the alignment of structure fields. For example, to
3070 create a double-word aligned @code{int} pair, you could write:
3073 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
3077 This is an alternative to creating a union with a @code{double} member
3078 that forces the union to be double-word aligned.
3080 As in the preceding examples, you can explicitly specify the alignment
3081 (in bytes) that you wish the compiler to use for a given variable or
3082 structure field. Alternatively, you can leave out the alignment factor
3083 and just ask the compiler to align a variable or field to the maximum
3084 useful alignment for the target machine you are compiling for. For
3085 example, you could write:
3088 short array[3] __attribute__ ((aligned));
3091 Whenever you leave out the alignment factor in an @code{aligned} attribute
3092 specification, the compiler automatically sets the alignment for the declared
3093 variable or field to the largest alignment which is ever used for any data
3094 type on the target machine you are compiling for. Doing this can often make
3095 copy operations more efficient, because the compiler can use whatever
3096 instructions copy the biggest chunks of memory when performing copies to
3097 or from the variables or fields that you have aligned this way.
3099 The @code{aligned} attribute can only increase the alignment; but you
3100 can decrease it by specifying @code{packed} as well. See below.
3102 Note that the effectiveness of @code{aligned} attributes may be limited
3103 by inherent limitations in your linker. On many systems, the linker is
3104 only able to arrange for variables to be aligned up to a certain maximum
3105 alignment. (For some linkers, the maximum supported alignment may
3106 be very very small.) If your linker is only able to align variables
3107 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3108 in an @code{__attribute__} will still only provide you with 8 byte
3109 alignment. See your linker documentation for further information.
3111 @item cleanup (@var{cleanup_function})
3112 @cindex @code{cleanup} attribute
3113 The @code{cleanup} attribute runs a function when the variable goes
3114 out of scope. This attribute can only be applied to auto function
3115 scope variables; it may not be applied to parameters or variables
3116 with static storage duration. The function must take one parameter,
3117 a pointer to a type compatible with the variable. The return value
3118 of the function (if any) is ignored.
3120 If @option{-fexceptions} is enabled, then @var{cleanup_function}
3121 will be run during the stack unwinding that happens during the
3122 processing of the exception. Note that the @code{cleanup} attribute
3123 does not allow the exception to be caught, only to perform an action.
3124 It is undefined what happens if @var{cleanup_function} does not
3129 @cindex @code{common} attribute
3130 @cindex @code{nocommon} attribute
3133 The @code{common} attribute requests GCC to place a variable in
3134 ``common'' storage. The @code{nocommon} attribute requests the
3135 opposite -- to allocate space for it directly.
3137 These attributes override the default chosen by the
3138 @option{-fno-common} and @option{-fcommon} flags respectively.
3141 @cindex @code{deprecated} attribute
3142 The @code{deprecated} attribute results in a warning if the variable
3143 is used anywhere in the source file. This is useful when identifying
3144 variables that are expected to be removed in a future version of a
3145 program. The warning also includes the location of the declaration
3146 of the deprecated variable, to enable users to easily find further
3147 information about why the variable is deprecated, or what they should
3148 do instead. Note that the warning only occurs for uses:
3151 extern int old_var __attribute__ ((deprecated));
3153 int new_fn () @{ return old_var; @}
3156 results in a warning on line 3 but not line 2.
3158 The @code{deprecated} attribute can also be used for functions and
3159 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3161 @item mode (@var{mode})
3162 @cindex @code{mode} attribute
3163 This attribute specifies the data type for the declaration---whichever
3164 type corresponds to the mode @var{mode}. This in effect lets you
3165 request an integer or floating point type according to its width.
3167 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3168 indicate the mode corresponding to a one-byte integer, @samp{word} or
3169 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3170 or @samp{__pointer__} for the mode used to represent pointers.
3173 @cindex @code{packed} attribute
3174 The @code{packed} attribute specifies that a variable or structure field
3175 should have the smallest possible alignment---one byte for a variable,
3176 and one bit for a field, unless you specify a larger value with the
3177 @code{aligned} attribute.
3179 Here is a structure in which the field @code{x} is packed, so that it
3180 immediately follows @code{a}:
3186 int x[2] __attribute__ ((packed));
3190 @item section ("@var{section-name}")
3191 @cindex @code{section} variable attribute
3192 Normally, the compiler places the objects it generates in sections like
3193 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3194 or you need certain particular variables to appear in special sections,
3195 for example to map to special hardware. The @code{section}
3196 attribute specifies that a variable (or function) lives in a particular
3197 section. For example, this small program uses several specific section names:
3200 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3201 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3202 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3203 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3207 /* Initialize stack pointer */
3208 init_sp (stack + sizeof (stack));
3210 /* Initialize initialized data */
3211 memcpy (&init_data, &data, &edata - &data);
3213 /* Turn on the serial ports */
3220 Use the @code{section} attribute with an @emph{initialized} definition
3221 of a @emph{global} variable, as shown in the example. GCC issues
3222 a warning and otherwise ignores the @code{section} attribute in
3223 uninitialized variable declarations.
3225 You may only use the @code{section} attribute with a fully initialized
3226 global definition because of the way linkers work. The linker requires
3227 each object be defined once, with the exception that uninitialized
3228 variables tentatively go in the @code{common} (or @code{bss}) section
3229 and can be multiply ``defined''. You can force a variable to be
3230 initialized with the @option{-fno-common} flag or the @code{nocommon}
3233 Some file formats do not support arbitrary sections so the @code{section}
3234 attribute is not available on all platforms.
3235 If you need to map the entire contents of a module to a particular
3236 section, consider using the facilities of the linker instead.
3239 @cindex @code{shared} variable attribute
3240 On Microsoft Windows, in addition to putting variable definitions in a named
3241 section, the section can also be shared among all running copies of an
3242 executable or DLL@. For example, this small program defines shared data
3243 by putting it in a named section @code{shared} and marking the section
3247 int foo __attribute__((section ("shared"), shared)) = 0;
3252 /* Read and write foo. All running
3253 copies see the same value. */
3259 You may only use the @code{shared} attribute along with @code{section}
3260 attribute with a fully initialized global definition because of the way
3261 linkers work. See @code{section} attribute for more information.
3263 The @code{shared} attribute is only available on Microsoft Windows@.
3265 @item tls_model ("@var{tls_model}")
3266 @cindex @code{tls_model} attribute
3267 The @code{tls_model} attribute sets thread-local storage model
3268 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3269 overriding @code{-ftls-model=} command line switch on a per-variable
3271 The @var{tls_model} argument should be one of @code{global-dynamic},
3272 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3274 Not all targets support this attribute.
3276 @item transparent_union
3277 This attribute, attached to a function parameter which is a union, means
3278 that the corresponding argument may have the type of any union member,
3279 but the argument is passed as if its type were that of the first union
3280 member. For more details see @xref{Type Attributes}. You can also use
3281 this attribute on a @code{typedef} for a union data type; then it
3282 applies to all function parameters with that type.
3285 This attribute, attached to a variable, means that the variable is meant
3286 to be possibly unused. GCC will not produce a warning for this
3289 @item vector_size (@var{bytes})
3290 This attribute specifies the vector size for the variable, measured in
3291 bytes. For example, the declaration:
3294 int foo __attribute__ ((vector_size (16)));
3298 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3299 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3300 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3302 This attribute is only applicable to integral and float scalars,
3303 although arrays, pointers, and function return values are allowed in
3304 conjunction with this construct.
3306 Aggregates with this attribute are invalid, even if they are of the same
3307 size as a corresponding scalar. For example, the declaration:
3310 struct S @{ int a; @};
3311 struct S __attribute__ ((vector_size (16))) foo;
3315 is invalid even if the size of the structure is the same as the size of
3319 The @code{weak} attribute is described in @xref{Function Attributes}.
3322 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3325 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3329 @subsection M32R/D Variable Attributes
3331 One attribute is currently defined for the M32R/D.
3334 @item model (@var{model-name})
3335 @cindex variable addressability on the M32R/D
3336 Use this attribute on the M32R/D to set the addressability of an object.
3337 The identifier @var{model-name} is one of @code{small}, @code{medium},
3338 or @code{large}, representing each of the code models.
3340 Small model objects live in the lower 16MB of memory (so that their
3341 addresses can be loaded with the @code{ld24} instruction).
3343 Medium and large model objects may live anywhere in the 32-bit address space
3344 (the compiler will generate @code{seth/add3} instructions to load their
3348 @subsection i386 Variable Attributes
3350 Two attributes are currently defined for i386 configurations:
3351 @code{ms_struct} and @code{gcc_struct}
3356 @cindex @code{ms_struct} attribute
3357 @cindex @code{gcc_struct} attribute
3359 If @code{packed} is used on a structure, or if bit-fields are used
3360 it may be that the Microsoft ABI packs them differently
3361 than GCC would normally pack them. Particularly when moving packed
3362 data between functions compiled with GCC and the native Microsoft compiler
3363 (either via function call or as data in a file), it may be necessary to access
3366 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3367 compilers to match the native Microsoft compiler.
3370 @node Type Attributes
3371 @section Specifying Attributes of Types
3372 @cindex attribute of types
3373 @cindex type attributes
3375 The keyword @code{__attribute__} allows you to specify special
3376 attributes of @code{struct} and @code{union} types when you define such
3377 types. This keyword is followed by an attribute specification inside
3378 double parentheses. Six attributes are currently defined for types:
3379 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3380 @code{deprecated} and @code{may_alias}. Other attributes are defined for
3381 functions (@pxref{Function Attributes}) and for variables
3382 (@pxref{Variable Attributes}).
3384 You may also specify any one of these attributes with @samp{__}
3385 preceding and following its keyword. This allows you to use these
3386 attributes in header files without being concerned about a possible
3387 macro of the same name. For example, you may use @code{__aligned__}
3388 instead of @code{aligned}.
3390 You may specify the @code{aligned} and @code{transparent_union}
3391 attributes either in a @code{typedef} declaration or just past the
3392 closing curly brace of a complete enum, struct or union type
3393 @emph{definition} and the @code{packed} attribute only past the closing
3394 brace of a definition.
3396 You may also specify attributes between the enum, struct or union
3397 tag and the name of the type rather than after the closing brace.
3399 @xref{Attribute Syntax}, for details of the exact syntax for using
3403 @cindex @code{aligned} attribute
3404 @item aligned (@var{alignment})
3405 This attribute specifies a minimum alignment (in bytes) for variables
3406 of the specified type. For example, the declarations:
3409 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3410 typedef int more_aligned_int __attribute__ ((aligned (8)));
3414 force the compiler to insure (as far as it can) that each variable whose
3415 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3416 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3417 variables of type @code{struct S} aligned to 8-byte boundaries allows
3418 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3419 store) instructions when copying one variable of type @code{struct S} to
3420 another, thus improving run-time efficiency.
3422 Note that the alignment of any given @code{struct} or @code{union} type
3423 is required by the ISO C standard to be at least a perfect multiple of
3424 the lowest common multiple of the alignments of all of the members of
3425 the @code{struct} or @code{union} in question. This means that you @emph{can}
3426 effectively adjust the alignment of a @code{struct} or @code{union}
3427 type by attaching an @code{aligned} attribute to any one of the members
3428 of such a type, but the notation illustrated in the example above is a
3429 more obvious, intuitive, and readable way to request the compiler to
3430 adjust the alignment of an entire @code{struct} or @code{union} type.
3432 As in the preceding example, you can explicitly specify the alignment
3433 (in bytes) that you wish the compiler to use for a given @code{struct}
3434 or @code{union} type. Alternatively, you can leave out the alignment factor
3435 and just ask the compiler to align a type to the maximum
3436 useful alignment for the target machine you are compiling for. For
3437 example, you could write:
3440 struct S @{ short f[3]; @} __attribute__ ((aligned));
3443 Whenever you leave out the alignment factor in an @code{aligned}
3444 attribute specification, the compiler automatically sets the alignment
3445 for the type to the largest alignment which is ever used for any data
3446 type on the target machine you are compiling for. Doing this can often
3447 make copy operations more efficient, because the compiler can use
3448 whatever instructions copy the biggest chunks of memory when performing
3449 copies to or from the variables which have types that you have aligned
3452 In the example above, if the size of each @code{short} is 2 bytes, then
3453 the size of the entire @code{struct S} type is 6 bytes. The smallest
3454 power of two which is greater than or equal to that is 8, so the
3455 compiler sets the alignment for the entire @code{struct S} type to 8
3458 Note that although you can ask the compiler to select a time-efficient
3459 alignment for a given type and then declare only individual stand-alone
3460 objects of that type, the compiler's ability to select a time-efficient
3461 alignment is primarily useful only when you plan to create arrays of
3462 variables having the relevant (efficiently aligned) type. If you
3463 declare or use arrays of variables of an efficiently-aligned type, then
3464 it is likely that your program will also be doing pointer arithmetic (or
3465 subscripting, which amounts to the same thing) on pointers to the
3466 relevant type, and the code that the compiler generates for these
3467 pointer arithmetic operations will often be more efficient for
3468 efficiently-aligned types than for other types.
3470 The @code{aligned} attribute can only increase the alignment; but you
3471 can decrease it by specifying @code{packed} as well. See below.
3473 Note that the effectiveness of @code{aligned} attributes may be limited
3474 by inherent limitations in your linker. On many systems, the linker is
3475 only able to arrange for variables to be aligned up to a certain maximum
3476 alignment. (For some linkers, the maximum supported alignment may
3477 be very very small.) If your linker is only able to align variables
3478 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3479 in an @code{__attribute__} will still only provide you with 8 byte
3480 alignment. See your linker documentation for further information.
3483 This attribute, attached to @code{struct} or @code{union} type
3484 definition, specifies that each member of the structure or union is
3485 placed to minimize the memory required. When attached to an @code{enum}
3486 definition, it indicates that the smallest integral type should be used.
3488 @opindex fshort-enums
3489 Specifying this attribute for @code{struct} and @code{union} types is
3490 equivalent to specifying the @code{packed} attribute on each of the
3491 structure or union members. Specifying the @option{-fshort-enums}
3492 flag on the line is equivalent to specifying the @code{packed}
3493 attribute on all @code{enum} definitions.
3495 In the following example @code{struct my_packed_struct}'s members are
3496 packed closely together, but the internal layout of its @code{s} member
3497 is not packed -- to do that, @code{struct my_unpacked_struct} would need to
3501 struct my_unpacked_struct
3507 struct my_packed_struct __attribute__ ((__packed__))
3511 struct my_unpacked_struct s;
3515 You may only specify this attribute on the definition of a @code{enum},
3516 @code{struct} or @code{union}, not on a @code{typedef} which does not
3517 also define the enumerated type, structure or union.
3519 @item transparent_union
3520 This attribute, attached to a @code{union} type definition, indicates
3521 that any function parameter having that union type causes calls to that
3522 function to be treated in a special way.
3524 First, the argument corresponding to a transparent union type can be of
3525 any type in the union; no cast is required. Also, if the union contains
3526 a pointer type, the corresponding argument can be a null pointer
3527 constant or a void pointer expression; and if the union contains a void
3528 pointer type, the corresponding argument can be any pointer expression.
3529 If the union member type is a pointer, qualifiers like @code{const} on
3530 the referenced type must be respected, just as with normal pointer
3533 Second, the argument is passed to the function using the calling
3534 conventions of the first member of the transparent union, not the calling
3535 conventions of the union itself. All members of the union must have the
3536 same machine representation; this is necessary for this argument passing
3539 Transparent unions are designed for library functions that have multiple
3540 interfaces for compatibility reasons. For example, suppose the
3541 @code{wait} function must accept either a value of type @code{int *} to
3542 comply with Posix, or a value of type @code{union wait *} to comply with
3543 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3544 @code{wait} would accept both kinds of arguments, but it would also
3545 accept any other pointer type and this would make argument type checking
3546 less useful. Instead, @code{<sys/wait.h>} might define the interface
3554 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3556 pid_t wait (wait_status_ptr_t);
3559 This interface allows either @code{int *} or @code{union wait *}
3560 arguments to be passed, using the @code{int *} calling convention.
3561 The program can call @code{wait} with arguments of either type:
3564 int w1 () @{ int w; return wait (&w); @}
3565 int w2 () @{ union wait w; return wait (&w); @}
3568 With this interface, @code{wait}'s implementation might look like this:
3571 pid_t wait (wait_status_ptr_t p)
3573 return waitpid (-1, p.__ip, 0);
3578 When attached to a type (including a @code{union} or a @code{struct}),
3579 this attribute means that variables of that type are meant to appear
3580 possibly unused. GCC will not produce a warning for any variables of
3581 that type, even if the variable appears to do nothing. This is often
3582 the case with lock or thread classes, which are usually defined and then
3583 not referenced, but contain constructors and destructors that have
3584 nontrivial bookkeeping functions.
3587 The @code{deprecated} attribute results in a warning if the type
3588 is used anywhere in the source file. This is useful when identifying
3589 types that are expected to be removed in a future version of a program.
3590 If possible, the warning also includes the location of the declaration
3591 of the deprecated type, to enable users to easily find further
3592 information about why the type is deprecated, or what they should do
3593 instead. Note that the warnings only occur for uses and then only
3594 if the type is being applied to an identifier that itself is not being
3595 declared as deprecated.
3598 typedef int T1 __attribute__ ((deprecated));
3602 typedef T1 T3 __attribute__ ((deprecated));
3603 T3 z __attribute__ ((deprecated));
3606 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3607 warning is issued for line 4 because T2 is not explicitly
3608 deprecated. Line 5 has no warning because T3 is explicitly
3609 deprecated. Similarly for line 6.
3611 The @code{deprecated} attribute can also be used for functions and
3612 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3615 Accesses to objects with types with this attribute are not subjected to
3616 type-based alias analysis, but are instead assumed to be able to alias
3617 any other type of objects, just like the @code{char} type. See
3618 @option{-fstrict-aliasing} for more information on aliasing issues.
3623 typedef short __attribute__((__may_alias__)) short_a;
3629 short_a *b = (short_a *) &a;
3633 if (a == 0x12345678)
3640 If you replaced @code{short_a} with @code{short} in the variable
3641 declaration, the above program would abort when compiled with
3642 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3643 above in recent GCC versions.
3645 @subsection i386 Type Attributes
3647 Two attributes are currently defined for i386 configurations:
3648 @code{ms_struct} and @code{gcc_struct}
3652 @cindex @code{ms_struct}
3653 @cindex @code{gcc_struct}
3655 If @code{packed} is used on a structure, or if bit-fields are used
3656 it may be that the Microsoft ABI packs them differently
3657 than GCC would normally pack them. Particularly when moving packed
3658 data between functions compiled with GCC and the native Microsoft compiler
3659 (either via function call or as data in a file), it may be necessary to access
3662 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3663 compilers to match the native Microsoft compiler.
3666 To specify multiple attributes, separate them by commas within the
3667 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3671 @section An Inline Function is As Fast As a Macro
3672 @cindex inline functions
3673 @cindex integrating function code
3675 @cindex macros, inline alternative
3677 By declaring a function @code{inline}, you can direct GCC to
3678 integrate that function's code into the code for its callers. This
3679 makes execution faster by eliminating the function-call overhead; in
3680 addition, if any of the actual argument values are constant, their known
3681 values may permit simplifications at compile time so that not all of the
3682 inline function's code needs to be included. The effect on code size is
3683 less predictable; object code may be larger or smaller with function
3684 inlining, depending on the particular case. Inlining of functions is an
3685 optimization and it really ``works'' only in optimizing compilation. If
3686 you don't use @option{-O}, no function is really inline.
3688 Inline functions are included in the ISO C99 standard, but there are
3689 currently substantial differences between what GCC implements and what
3690 the ISO C99 standard requires.
3692 To declare a function inline, use the @code{inline} keyword in its
3693 declaration, like this:
3703 (If you are writing a header file to be included in ISO C programs, write
3704 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3705 You can also make all ``simple enough'' functions inline with the option
3706 @option{-finline-functions}.
3709 Note that certain usages in a function definition can make it unsuitable
3710 for inline substitution. Among these usages are: use of varargs, use of
3711 alloca, use of variable sized data types (@pxref{Variable Length}),
3712 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3713 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3714 will warn when a function marked @code{inline} could not be substituted,
3715 and will give the reason for the failure.
3717 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3718 does not affect the linkage of the function.
3720 @cindex automatic @code{inline} for C++ member fns
3721 @cindex @code{inline} automatic for C++ member fns
3722 @cindex member fns, automatically @code{inline}
3723 @cindex C++ member fns, automatically @code{inline}
3724 @opindex fno-default-inline
3725 GCC automatically inlines member functions defined within the class
3726 body of C++ programs even if they are not explicitly declared
3727 @code{inline}. (You can override this with @option{-fno-default-inline};
3728 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3730 @cindex inline functions, omission of
3731 @opindex fkeep-inline-functions
3732 When a function is both inline and @code{static}, if all calls to the
3733 function are integrated into the caller, and the function's address is
3734 never used, then the function's own assembler code is never referenced.
3735 In this case, GCC does not actually output assembler code for the
3736 function, unless you specify the option @option{-fkeep-inline-functions}.
3737 Some calls cannot be integrated for various reasons (in particular,
3738 calls that precede the function's definition cannot be integrated, and
3739 neither can recursive calls within the definition). If there is a
3740 nonintegrated call, then the function is compiled to assembler code as
3741 usual. The function must also be compiled as usual if the program
3742 refers to its address, because that can't be inlined.
3744 @cindex non-static inline function
3745 When an inline function is not @code{static}, then the compiler must assume
3746 that there may be calls from other source files; since a global symbol can
3747 be defined only once in any program, the function must not be defined in
3748 the other source files, so the calls therein cannot be integrated.
3749 Therefore, a non-@code{static} inline function is always compiled on its
3750 own in the usual fashion.
3752 If you specify both @code{inline} and @code{extern} in the function
3753 definition, then the definition is used only for inlining. In no case
3754 is the function compiled on its own, not even if you refer to its
3755 address explicitly. Such an address becomes an external reference, as
3756 if you had only declared the function, and had not defined it.
3758 This combination of @code{inline} and @code{extern} has almost the
3759 effect of a macro. The way to use it is to put a function definition in
3760 a header file with these keywords, and put another copy of the
3761 definition (lacking @code{inline} and @code{extern}) in a library file.
3762 The definition in the header file will cause most calls to the function
3763 to be inlined. If any uses of the function remain, they will refer to
3764 the single copy in the library.
3766 Since GCC eventually will implement ISO C99 semantics for
3767 inline functions, it is best to use @code{static inline} only
3768 to guarantee compatibility. (The
3769 existing semantics will remain available when @option{-std=gnu89} is
3770 specified, but eventually the default will be @option{-std=gnu99} and
3771 that will implement the C99 semantics, though it does not do so yet.)
3773 GCC does not inline any functions when not optimizing unless you specify
3774 the @samp{always_inline} attribute for the function, like this:
3778 inline void foo (const char) __attribute__((always_inline));
3782 @section Assembler Instructions with C Expression Operands
3783 @cindex extended @code{asm}
3784 @cindex @code{asm} expressions
3785 @cindex assembler instructions
3788 In an assembler instruction using @code{asm}, you can specify the
3789 operands of the instruction using C expressions. This means you need not
3790 guess which registers or memory locations will contain the data you want
3793 You must specify an assembler instruction template much like what
3794 appears in a machine description, plus an operand constraint string for
3797 For example, here is how to use the 68881's @code{fsinx} instruction:
3800 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3804 Here @code{angle} is the C expression for the input operand while
3805 @code{result} is that of the output operand. Each has @samp{"f"} as its
3806 operand constraint, saying that a floating point register is required.
3807 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3808 output operands' constraints must use @samp{=}. The constraints use the
3809 same language used in the machine description (@pxref{Constraints}).
3811 Each operand is described by an operand-constraint string followed by
3812 the C expression in parentheses. A colon separates the assembler
3813 template from the first output operand and another separates the last
3814 output operand from the first input, if any. Commas separate the
3815 operands within each group. The total number of operands is currently
3816 limited to 30; this limitation may be lifted in some future version of
3819 If there are no output operands but there are input operands, you must
3820 place two consecutive colons surrounding the place where the output
3823 As of GCC version 3.1, it is also possible to specify input and output
3824 operands using symbolic names which can be referenced within the
3825 assembler code. These names are specified inside square brackets
3826 preceding the constraint string, and can be referenced inside the
3827 assembler code using @code{%[@var{name}]} instead of a percentage sign
3828 followed by the operand number. Using named operands the above example
3832 asm ("fsinx %[angle],%[output]"
3833 : [output] "=f" (result)
3834 : [angle] "f" (angle));
3838 Note that the symbolic operand names have no relation whatsoever to
3839 other C identifiers. You may use any name you like, even those of
3840 existing C symbols, but you must ensure that no two operands within the same
3841 assembler construct use the same symbolic name.
3843 Output operand expressions must be lvalues; the compiler can check this.
3844 The input operands need not be lvalues. The compiler cannot check
3845 whether the operands have data types that are reasonable for the
3846 instruction being executed. It does not parse the assembler instruction
3847 template and does not know what it means or even whether it is valid
3848 assembler input. The extended @code{asm} feature is most often used for
3849 machine instructions the compiler itself does not know exist. If
3850 the output expression cannot be directly addressed (for example, it is a
3851 bit-field), your constraint must allow a register. In that case, GCC
3852 will use the register as the output of the @code{asm}, and then store
3853 that register into the output.
3855 The ordinary output operands must be write-only; GCC will assume that
3856 the values in these operands before the instruction are dead and need
3857 not be generated. Extended asm supports input-output or read-write
3858 operands. Use the constraint character @samp{+} to indicate such an
3859 operand and list it with the output operands. You should only use
3860 read-write operands when the constraints for the operand (or the
3861 operand in which only some of the bits are to be changed) allow a
3864 You may, as an alternative, logically split its function into two
3865 separate operands, one input operand and one write-only output
3866 operand. The connection between them is expressed by constraints
3867 which say they need to be in the same location when the instruction
3868 executes. You can use the same C expression for both operands, or
3869 different expressions. For example, here we write the (fictitious)
3870 @samp{combine} instruction with @code{bar} as its read-only source
3871 operand and @code{foo} as its read-write destination:
3874 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3878 The constraint @samp{"0"} for operand 1 says that it must occupy the
3879 same location as operand 0. A number in constraint is allowed only in
3880 an input operand and it must refer to an output operand.
3882 Only a number in the constraint can guarantee that one operand will be in
3883 the same place as another. The mere fact that @code{foo} is the value
3884 of both operands is not enough to guarantee that they will be in the
3885 same place in the generated assembler code. The following would not
3889 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3892 Various optimizations or reloading could cause operands 0 and 1 to be in
3893 different registers; GCC knows no reason not to do so. For example, the
3894 compiler might find a copy of the value of @code{foo} in one register and
3895 use it for operand 1, but generate the output operand 0 in a different
3896 register (copying it afterward to @code{foo}'s own address). Of course,
3897 since the register for operand 1 is not even mentioned in the assembler
3898 code, the result will not work, but GCC can't tell that.
3900 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3901 the operand number for a matching constraint. For example:
3904 asm ("cmoveq %1,%2,%[result]"
3905 : [result] "=r"(result)
3906 : "r" (test), "r"(new), "[result]"(old));
3909 Some instructions clobber specific hard registers. To describe this,
3910 write a third colon after the input operands, followed by the names of
3911 the clobbered hard registers (given as strings). Here is a realistic
3912 example for the VAX:
3915 asm volatile ("movc3 %0,%1,%2"
3917 : "g" (from), "g" (to), "g" (count)
3918 : "r0", "r1", "r2", "r3", "r4", "r5");
3921 You may not write a clobber description in a way that overlaps with an
3922 input or output operand. For example, you may not have an operand
3923 describing a register class with one member if you mention that register
3924 in the clobber list. Variables declared to live in specific registers
3925 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
3926 have no part mentioned in the clobber description.
3927 There is no way for you to specify that an input
3928 operand is modified without also specifying it as an output
3929 operand. Note that if all the output operands you specify are for this
3930 purpose (and hence unused), you will then also need to specify
3931 @code{volatile} for the @code{asm} construct, as described below, to
3932 prevent GCC from deleting the @code{asm} statement as unused.
3934 If you refer to a particular hardware register from the assembler code,
3935 you will probably have to list the register after the third colon to
3936 tell the compiler the register's value is modified. In some assemblers,
3937 the register names begin with @samp{%}; to produce one @samp{%} in the
3938 assembler code, you must write @samp{%%} in the input.
3940 If your assembler instruction can alter the condition code register, add
3941 @samp{cc} to the list of clobbered registers. GCC on some machines
3942 represents the condition codes as a specific hardware register;
3943 @samp{cc} serves to name this register. On other machines, the
3944 condition code is handled differently, and specifying @samp{cc} has no
3945 effect. But it is valid no matter what the machine.
3947 If your assembler instructions access memory in an unpredictable
3948 fashion, add @samp{memory} to the list of clobbered registers. This
3949 will cause GCC to not keep memory values cached in registers across the
3950 assembler instruction and not optimize stores or loads to that memory.
3951 You will also want to add the @code{volatile} keyword if the memory
3952 affected is not listed in the inputs or outputs of the @code{asm}, as
3953 the @samp{memory} clobber does not count as a side-effect of the
3954 @code{asm}. If you know how large the accessed memory is, you can add
3955 it as input or output but if this is not known, you should add
3956 @samp{memory}. As an example, if you access ten bytes of a string, you
3957 can use a memory input like:
3960 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
3963 Note that in the following example the memory input is necessary,
3964 otherwise GCC might optimize the store to @code{x} away:
3971 asm ("magic stuff accessing an 'int' pointed to by '%1'"
3972 "=&d" (r) : "a" (y), "m" (*y));
3977 You can put multiple assembler instructions together in a single
3978 @code{asm} template, separated by the characters normally used in assembly
3979 code for the system. A combination that works in most places is a newline
3980 to break the line, plus a tab character to move to the instruction field
3981 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
3982 assembler allows semicolons as a line-breaking character. Note that some
3983 assembler dialects use semicolons to start a comment.
3984 The input operands are guaranteed not to use any of the clobbered
3985 registers, and neither will the output operands' addresses, so you can
3986 read and write the clobbered registers as many times as you like. Here
3987 is an example of multiple instructions in a template; it assumes the
3988 subroutine @code{_foo} accepts arguments in registers 9 and 10:
3991 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
3993 : "g" (from), "g" (to)
3997 Unless an output operand has the @samp{&} constraint modifier, GCC
3998 may allocate it in the same register as an unrelated input operand, on
3999 the assumption the inputs are consumed before the outputs are produced.
4000 This assumption may be false if the assembler code actually consists of
4001 more than one instruction. In such a case, use @samp{&} for each output
4002 operand that may not overlap an input. @xref{Modifiers}.
4004 If you want to test the condition code produced by an assembler
4005 instruction, you must include a branch and a label in the @code{asm}
4006 construct, as follows:
4009 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
4015 This assumes your assembler supports local labels, as the GNU assembler
4016 and most Unix assemblers do.
4018 Speaking of labels, jumps from one @code{asm} to another are not
4019 supported. The compiler's optimizers do not know about these jumps, and
4020 therefore they cannot take account of them when deciding how to
4023 @cindex macros containing @code{asm}
4024 Usually the most convenient way to use these @code{asm} instructions is to
4025 encapsulate them in macros that look like functions. For example,
4029 (@{ double __value, __arg = (x); \
4030 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
4035 Here the variable @code{__arg} is used to make sure that the instruction
4036 operates on a proper @code{double} value, and to accept only those
4037 arguments @code{x} which can convert automatically to a @code{double}.
4039 Another way to make sure the instruction operates on the correct data
4040 type is to use a cast in the @code{asm}. This is different from using a
4041 variable @code{__arg} in that it converts more different types. For
4042 example, if the desired type were @code{int}, casting the argument to
4043 @code{int} would accept a pointer with no complaint, while assigning the
4044 argument to an @code{int} variable named @code{__arg} would warn about
4045 using a pointer unless the caller explicitly casts it.
4047 If an @code{asm} has output operands, GCC assumes for optimization
4048 purposes the instruction has no side effects except to change the output
4049 operands. This does not mean instructions with a side effect cannot be
4050 used, but you must be careful, because the compiler may eliminate them
4051 if the output operands aren't used, or move them out of loops, or
4052 replace two with one if they constitute a common subexpression. Also,
4053 if your instruction does have a side effect on a variable that otherwise
4054 appears not to change, the old value of the variable may be reused later
4055 if it happens to be found in a register.
4057 You can prevent an @code{asm} instruction from being deleted, moved
4058 significantly, or combined, by writing the keyword @code{volatile} after
4059 the @code{asm}. For example:
4062 #define get_and_set_priority(new) \
4064 asm volatile ("get_and_set_priority %0, %1" \
4065 : "=g" (__old) : "g" (new)); \
4070 If you write an @code{asm} instruction with no outputs, GCC will know
4071 the instruction has side-effects and will not delete the instruction or
4072 move it outside of loops.
4074 The @code{volatile} keyword indicates that the instruction has
4075 important side-effects. GCC will not delete a volatile @code{asm} if
4076 it is reachable. (The instruction can still be deleted if GCC can
4077 prove that control-flow will never reach the location of the
4078 instruction.) In addition, GCC will not reschedule instructions
4079 across a volatile @code{asm} instruction. For example:
4082 *(volatile int *)addr = foo;
4083 asm volatile ("eieio" : : );
4087 Assume @code{addr} contains the address of a memory mapped device
4088 register. The PowerPC @code{eieio} instruction (Enforce In-order
4089 Execution of I/O) tells the CPU to make sure that the store to that
4090 device register happens before it issues any other I/O@.
4092 Note that even a volatile @code{asm} instruction can be moved in ways
4093 that appear insignificant to the compiler, such as across jump
4094 instructions. You can't expect a sequence of volatile @code{asm}
4095 instructions to remain perfectly consecutive. If you want consecutive
4096 output, use a single @code{asm}. Also, GCC will perform some
4097 optimizations across a volatile @code{asm} instruction; GCC does not
4098 ``forget everything'' when it encounters a volatile @code{asm}
4099 instruction the way some other compilers do.
4101 An @code{asm} instruction without any operands or clobbers (an ``old
4102 style'' @code{asm}) will be treated identically to a volatile
4103 @code{asm} instruction.
4105 It is a natural idea to look for a way to give access to the condition
4106 code left by the assembler instruction. However, when we attempted to
4107 implement this, we found no way to make it work reliably. The problem
4108 is that output operands might need reloading, which would result in
4109 additional following ``store'' instructions. On most machines, these
4110 instructions would alter the condition code before there was time to
4111 test it. This problem doesn't arise for ordinary ``test'' and
4112 ``compare'' instructions because they don't have any output operands.
4114 For reasons similar to those described above, it is not possible to give
4115 an assembler instruction access to the condition code left by previous
4118 If you are writing a header file that should be includable in ISO C
4119 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
4122 @subsection Size of an @code{asm}
4124 Some targets require that GCC track the size of each instruction used in
4125 order to generate correct code. Because the final length of an
4126 @code{asm} is only known by the assembler, GCC must make an estimate as
4127 to how big it will be. The estimate is formed by counting the number of
4128 statements in the pattern of the @code{asm} and multiplying that by the
4129 length of the longest instruction on that processor. Statements in the
4130 @code{asm} are identified by newline characters and whatever statement
4131 separator characters are supported by the assembler; on most processors
4132 this is the `@code{;}' character.
4134 Normally, GCC's estimate is perfectly adequate to ensure that correct
4135 code is generated, but it is possible to confuse the compiler if you use
4136 pseudo instructions or assembler macros that expand into multiple real
4137 instructions or if you use assembler directives that expand to more
4138 space in the object file than would be needed for a single instruction.
4139 If this happens then the assembler will produce a diagnostic saying that
4140 a label is unreachable.
4142 @subsection i386 floating point asm operands
4144 There are several rules on the usage of stack-like regs in
4145 asm_operands insns. These rules apply only to the operands that are
4150 Given a set of input regs that die in an asm_operands, it is
4151 necessary to know which are implicitly popped by the asm, and
4152 which must be explicitly popped by gcc.
4154 An input reg that is implicitly popped by the asm must be
4155 explicitly clobbered, unless it is constrained to match an
4159 For any input reg that is implicitly popped by an asm, it is
4160 necessary to know how to adjust the stack to compensate for the pop.
4161 If any non-popped input is closer to the top of the reg-stack than
4162 the implicitly popped reg, it would not be possible to know what the
4163 stack looked like---it's not clear how the rest of the stack ``slides
4166 All implicitly popped input regs must be closer to the top of
4167 the reg-stack than any input that is not implicitly popped.
4169 It is possible that if an input dies in an insn, reload might
4170 use the input reg for an output reload. Consider this example:
4173 asm ("foo" : "=t" (a) : "f" (b));
4176 This asm says that input B is not popped by the asm, and that
4177 the asm pushes a result onto the reg-stack, i.e., the stack is one
4178 deeper after the asm than it was before. But, it is possible that
4179 reload will think that it can use the same reg for both the input and
4180 the output, if input B dies in this insn.
4182 If any input operand uses the @code{f} constraint, all output reg
4183 constraints must use the @code{&} earlyclobber.
4185 The asm above would be written as
4188 asm ("foo" : "=&t" (a) : "f" (b));
4192 Some operands need to be in particular places on the stack. All
4193 output operands fall in this category---there is no other way to
4194 know which regs the outputs appear in unless the user indicates
4195 this in the constraints.
4197 Output operands must specifically indicate which reg an output
4198 appears in after an asm. @code{=f} is not allowed: the operand
4199 constraints must select a class with a single reg.
4202 Output operands may not be ``inserted'' between existing stack regs.
4203 Since no 387 opcode uses a read/write operand, all output operands
4204 are dead before the asm_operands, and are pushed by the asm_operands.
4205 It makes no sense to push anywhere but the top of the reg-stack.
4207 Output operands must start at the top of the reg-stack: output
4208 operands may not ``skip'' a reg.
4211 Some asm statements may need extra stack space for internal
4212 calculations. This can be guaranteed by clobbering stack registers
4213 unrelated to the inputs and outputs.
4217 Here are a couple of reasonable asms to want to write. This asm
4218 takes one input, which is internally popped, and produces two outputs.
4221 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4224 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4225 and replaces them with one output. The user must code the @code{st(1)}
4226 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4229 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4235 @section Controlling Names Used in Assembler Code
4236 @cindex assembler names for identifiers
4237 @cindex names used in assembler code
4238 @cindex identifiers, names in assembler code
4240 You can specify the name to be used in the assembler code for a C
4241 function or variable by writing the @code{asm} (or @code{__asm__})
4242 keyword after the declarator as follows:
4245 int foo asm ("myfoo") = 2;
4249 This specifies that the name to be used for the variable @code{foo} in
4250 the assembler code should be @samp{myfoo} rather than the usual
4253 On systems where an underscore is normally prepended to the name of a C
4254 function or variable, this feature allows you to define names for the
4255 linker that do not start with an underscore.
4257 It does not make sense to use this feature with a non-static local
4258 variable since such variables do not have assembler names. If you are
4259 trying to put the variable in a particular register, see @ref{Explicit
4260 Reg Vars}. GCC presently accepts such code with a warning, but will
4261 probably be changed to issue an error, rather than a warning, in the
4264 You cannot use @code{asm} in this way in a function @emph{definition}; but
4265 you can get the same effect by writing a declaration for the function
4266 before its definition and putting @code{asm} there, like this:
4269 extern func () asm ("FUNC");
4276 It is up to you to make sure that the assembler names you choose do not
4277 conflict with any other assembler symbols. Also, you must not use a
4278 register name; that would produce completely invalid assembler code. GCC
4279 does not as yet have the ability to store static variables in registers.
4280 Perhaps that will be added.
4282 @node Explicit Reg Vars
4283 @section Variables in Specified Registers
4284 @cindex explicit register variables
4285 @cindex variables in specified registers
4286 @cindex specified registers
4287 @cindex registers, global allocation
4289 GNU C allows you to put a few global variables into specified hardware
4290 registers. You can also specify the register in which an ordinary
4291 register variable should be allocated.
4295 Global register variables reserve registers throughout the program.
4296 This may be useful in programs such as programming language
4297 interpreters which have a couple of global variables that are accessed
4301 Local register variables in specific registers do not reserve the
4302 registers. The compiler's data flow analysis is capable of determining
4303 where the specified registers contain live values, and where they are
4304 available for other uses. Stores into local register variables may be deleted
4305 when they appear to be dead according to dataflow analysis. References
4306 to local register variables may be deleted or moved or simplified.
4308 These local variables are sometimes convenient for use with the extended
4309 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4310 output of the assembler instruction directly into a particular register.
4311 (This will work provided the register you specify fits the constraints
4312 specified for that operand in the @code{asm}.)
4320 @node Global Reg Vars
4321 @subsection Defining Global Register Variables
4322 @cindex global register variables
4323 @cindex registers, global variables in
4325 You can define a global register variable in GNU C like this:
4328 register int *foo asm ("a5");
4332 Here @code{a5} is the name of the register which should be used. Choose a
4333 register which is normally saved and restored by function calls on your
4334 machine, so that library routines will not clobber it.
4336 Naturally the register name is cpu-dependent, so you would need to
4337 conditionalize your program according to cpu type. The register
4338 @code{a5} would be a good choice on a 68000 for a variable of pointer
4339 type. On machines with register windows, be sure to choose a ``global''
4340 register that is not affected magically by the function call mechanism.
4342 In addition, operating systems on one type of cpu may differ in how they
4343 name the registers; then you would need additional conditionals. For
4344 example, some 68000 operating systems call this register @code{%a5}.
4346 Eventually there may be a way of asking the compiler to choose a register
4347 automatically, but first we need to figure out how it should choose and
4348 how to enable you to guide the choice. No solution is evident.
4350 Defining a global register variable in a certain register reserves that
4351 register entirely for this use, at least within the current compilation.
4352 The register will not be allocated for any other purpose in the functions
4353 in the current compilation. The register will not be saved and restored by
4354 these functions. Stores into this register are never deleted even if they
4355 would appear to be dead, but references may be deleted or moved or
4358 It is not safe to access the global register variables from signal
4359 handlers, or from more than one thread of control, because the system
4360 library routines may temporarily use the register for other things (unless
4361 you recompile them specially for the task at hand).
4363 @cindex @code{qsort}, and global register variables
4364 It is not safe for one function that uses a global register variable to
4365 call another such function @code{foo} by way of a third function
4366 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4367 different source file in which the variable wasn't declared). This is
4368 because @code{lose} might save the register and put some other value there.
4369 For example, you can't expect a global register variable to be available in
4370 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4371 might have put something else in that register. (If you are prepared to
4372 recompile @code{qsort} with the same global register variable, you can
4373 solve this problem.)
4375 If you want to recompile @code{qsort} or other source files which do not
4376 actually use your global register variable, so that they will not use that
4377 register for any other purpose, then it suffices to specify the compiler
4378 option @option{-ffixed-@var{reg}}. You need not actually add a global
4379 register declaration to their source code.
4381 A function which can alter the value of a global register variable cannot
4382 safely be called from a function compiled without this variable, because it
4383 could clobber the value the caller expects to find there on return.
4384 Therefore, the function which is the entry point into the part of the
4385 program that uses the global register variable must explicitly save and
4386 restore the value which belongs to its caller.
4388 @cindex register variable after @code{longjmp}
4389 @cindex global register after @code{longjmp}
4390 @cindex value after @code{longjmp}
4393 On most machines, @code{longjmp} will restore to each global register
4394 variable the value it had at the time of the @code{setjmp}. On some
4395 machines, however, @code{longjmp} will not change the value of global
4396 register variables. To be portable, the function that called @code{setjmp}
4397 should make other arrangements to save the values of the global register
4398 variables, and to restore them in a @code{longjmp}. This way, the same
4399 thing will happen regardless of what @code{longjmp} does.
4401 All global register variable declarations must precede all function
4402 definitions. If such a declaration could appear after function
4403 definitions, the declaration would be too late to prevent the register from
4404 being used for other purposes in the preceding functions.
4406 Global register variables may not have initial values, because an
4407 executable file has no means to supply initial contents for a register.
4409 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4410 registers, but certain library functions, such as @code{getwd}, as well
4411 as the subroutines for division and remainder, modify g3 and g4. g1 and
4412 g2 are local temporaries.
4414 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4415 Of course, it will not do to use more than a few of those.
4417 @node Local Reg Vars
4418 @subsection Specifying Registers for Local Variables
4419 @cindex local variables, specifying registers
4420 @cindex specifying registers for local variables
4421 @cindex registers for local variables
4423 You can define a local register variable with a specified register
4427 register int *foo asm ("a5");
4431 Here @code{a5} is the name of the register which should be used. Note
4432 that this is the same syntax used for defining global register
4433 variables, but for a local variable it would appear within a function.
4435 Naturally the register name is cpu-dependent, but this is not a
4436 problem, since specific registers are most often useful with explicit
4437 assembler instructions (@pxref{Extended Asm}). Both of these things
4438 generally require that you conditionalize your program according to
4441 In addition, operating systems on one type of cpu may differ in how they
4442 name the registers; then you would need additional conditionals. For
4443 example, some 68000 operating systems call this register @code{%a5}.
4445 Defining such a register variable does not reserve the register; it
4446 remains available for other uses in places where flow control determines
4447 the variable's value is not live.
4449 This option does not guarantee that GCC will generate code that has
4450 this variable in the register you specify at all times. You may not
4451 code an explicit reference to this register in an @code{asm} statement
4452 and assume it will always refer to this variable.
4454 Stores into local register variables may be deleted when they appear to be dead
4455 according to dataflow analysis. References to local register variables may
4456 be deleted or moved or simplified.
4458 @node Alternate Keywords
4459 @section Alternate Keywords
4460 @cindex alternate keywords
4461 @cindex keywords, alternate
4463 @option{-ansi} and the various @option{-std} options disable certain
4464 keywords. This causes trouble when you want to use GNU C extensions, or
4465 a general-purpose header file that should be usable by all programs,
4466 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4467 @code{inline} are not available in programs compiled with
4468 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4469 program compiled with @option{-std=c99}). The ISO C99 keyword
4470 @code{restrict} is only available when @option{-std=gnu99} (which will
4471 eventually be the default) or @option{-std=c99} (or the equivalent
4472 @option{-std=iso9899:1999}) is used.
4474 The way to solve these problems is to put @samp{__} at the beginning and
4475 end of each problematical keyword. For example, use @code{__asm__}
4476 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4478 Other C compilers won't accept these alternative keywords; if you want to
4479 compile with another compiler, you can define the alternate keywords as
4480 macros to replace them with the customary keywords. It looks like this:
4488 @findex __extension__
4490 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4492 prevent such warnings within one expression by writing
4493 @code{__extension__} before the expression. @code{__extension__} has no
4494 effect aside from this.
4496 @node Incomplete Enums
4497 @section Incomplete @code{enum} Types
4499 You can define an @code{enum} tag without specifying its possible values.
4500 This results in an incomplete type, much like what you get if you write
4501 @code{struct foo} without describing the elements. A later declaration
4502 which does specify the possible values completes the type.
4504 You can't allocate variables or storage using the type while it is
4505 incomplete. However, you can work with pointers to that type.
4507 This extension may not be very useful, but it makes the handling of
4508 @code{enum} more consistent with the way @code{struct} and @code{union}
4511 This extension is not supported by GNU C++.
4513 @node Function Names
4514 @section Function Names as Strings
4515 @cindex @code{__func__} identifier
4516 @cindex @code{__FUNCTION__} identifier
4517 @cindex @code{__PRETTY_FUNCTION__} identifier
4519 GCC provides three magic variables which hold the name of the current
4520 function, as a string. The first of these is @code{__func__}, which
4521 is part of the C99 standard:
4524 The identifier @code{__func__} is implicitly declared by the translator
4525 as if, immediately following the opening brace of each function
4526 definition, the declaration
4529 static const char __func__[] = "function-name";
4532 appeared, where function-name is the name of the lexically-enclosing
4533 function. This name is the unadorned name of the function.
4536 @code{__FUNCTION__} is another name for @code{__func__}. Older
4537 versions of GCC recognize only this name. However, it is not
4538 standardized. For maximum portability, we recommend you use
4539 @code{__func__}, but provide a fallback definition with the
4543 #if __STDC_VERSION__ < 199901L
4545 # define __func__ __FUNCTION__
4547 # define __func__ "<unknown>"
4552 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4553 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4554 the type signature of the function as well as its bare name. For
4555 example, this program:
4559 extern int printf (char *, ...);
4566 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4567 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4585 __PRETTY_FUNCTION__ = void a::sub(int)
4588 These identifiers are not preprocessor macros. In GCC 3.3 and
4589 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4590 were treated as string literals; they could be used to initialize
4591 @code{char} arrays, and they could be concatenated with other string
4592 literals. GCC 3.4 and later treat them as variables, like
4593 @code{__func__}. In C++, @code{__FUNCTION__} and
4594 @code{__PRETTY_FUNCTION__} have always been variables.
4596 @node Return Address
4597 @section Getting the Return or Frame Address of a Function
4599 These functions may be used to get information about the callers of a
4602 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4603 This function returns the return address of the current function, or of
4604 one of its callers. The @var{level} argument is number of frames to
4605 scan up the call stack. A value of @code{0} yields the return address
4606 of the current function, a value of @code{1} yields the return address
4607 of the caller of the current function, and so forth. When inlining
4608 the expected behavior is that the function will return the address of
4609 the function that will be returned to. To work around this behavior use
4610 the @code{noinline} function attribute.
4612 The @var{level} argument must be a constant integer.
4614 On some machines it may be impossible to determine the return address of
4615 any function other than the current one; in such cases, or when the top
4616 of the stack has been reached, this function will return @code{0} or a
4617 random value. In addition, @code{__builtin_frame_address} may be used
4618 to determine if the top of the stack has been reached.
4620 This function should only be used with a nonzero argument for debugging
4624 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4625 This function is similar to @code{__builtin_return_address}, but it
4626 returns the address of the function frame rather than the return address
4627 of the function. Calling @code{__builtin_frame_address} with a value of
4628 @code{0} yields the frame address of the current function, a value of
4629 @code{1} yields the frame address of the caller of the current function,
4632 The frame is the area on the stack which holds local variables and saved
4633 registers. The frame address is normally the address of the first word
4634 pushed on to the stack by the function. However, the exact definition
4635 depends upon the processor and the calling convention. If the processor
4636 has a dedicated frame pointer register, and the function has a frame,
4637 then @code{__builtin_frame_address} will return the value of the frame
4640 On some machines it may be impossible to determine the frame address of
4641 any function other than the current one; in such cases, or when the top
4642 of the stack has been reached, this function will return @code{0} if
4643 the first frame pointer is properly initialized by the startup code.
4645 This function should only be used with a nonzero argument for debugging
4649 @node Vector Extensions
4650 @section Using vector instructions through built-in functions
4652 On some targets, the instruction set contains SIMD vector instructions that
4653 operate on multiple values contained in one large register at the same time.
4654 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4657 The first step in using these extensions is to provide the necessary data
4658 types. This should be done using an appropriate @code{typedef}:
4661 typedef int v4si __attribute__ ((vector_size (16)));
4664 The @code{int} type specifies the base type, while the attribute specifies
4665 the vector size for the variable, measured in bytes. For example, the
4666 declaration above causes the compiler to set the mode for the @code{v4si}
4667 type to be 16 bytes wide and divided into @code{int} sized units. For
4668 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4669 corresponding mode of @code{foo} will be @acronym{V4SI}.
4671 The @code{vector_size} attribute is only applicable to integral and
4672 float scalars, although arrays, pointers, and function return values
4673 are allowed in conjunction with this construct.
4675 All the basic integer types can be used as base types, both as signed
4676 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4677 @code{long long}. In addition, @code{float} and @code{double} can be
4678 used to build floating-point vector types.
4680 Specifying a combination that is not valid for the current architecture
4681 will cause GCC to synthesize the instructions using a narrower mode.
4682 For example, if you specify a variable of type @code{V4SI} and your
4683 architecture does not allow for this specific SIMD type, GCC will
4684 produce code that uses 4 @code{SIs}.
4686 The types defined in this manner can be used with a subset of normal C
4687 operations. Currently, GCC will allow using the following operators
4688 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4690 The operations behave like C++ @code{valarrays}. Addition is defined as
4691 the addition of the corresponding elements of the operands. For
4692 example, in the code below, each of the 4 elements in @var{a} will be
4693 added to the corresponding 4 elements in @var{b} and the resulting
4694 vector will be stored in @var{c}.
4697 typedef int v4si __attribute__ ((vector_size (16)));
4704 Subtraction, multiplication, division, and the logical operations
4705 operate in a similar manner. Likewise, the result of using the unary
4706 minus or complement operators on a vector type is a vector whose
4707 elements are the negative or complemented values of the corresponding
4708 elements in the operand.
4710 You can declare variables and use them in function calls and returns, as
4711 well as in assignments and some casts. You can specify a vector type as
4712 a return type for a function. Vector types can also be used as function
4713 arguments. It is possible to cast from one vector type to another,
4714 provided they are of the same size (in fact, you can also cast vectors
4715 to and from other datatypes of the same size).
4717 You cannot operate between vectors of different lengths or different
4718 signedness without a cast.
4720 A port that supports hardware vector operations, usually provides a set
4721 of built-in functions that can be used to operate on vectors. For
4722 example, a function to add two vectors and multiply the result by a
4723 third could look like this:
4726 v4si f (v4si a, v4si b, v4si c)
4728 v4si tmp = __builtin_addv4si (a, b);
4729 return __builtin_mulv4si (tmp, c);
4736 @findex __builtin_offsetof
4738 GCC implements for both C and C++ a syntactic extension to implement
4739 the @code{offsetof} macro.
4743 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
4745 offsetof_member_designator:
4747 | offsetof_member_designator "." @code{identifier}
4748 | offsetof_member_designator "[" @code{expr} "]"
4751 This extension is sufficient such that
4754 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
4757 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
4758 may be dependent. In either case, @var{member} may consist of a single
4759 identifier, or a sequence of member accesses and array references.
4761 @node Other Builtins
4762 @section Other built-in functions provided by GCC
4763 @cindex built-in functions
4764 @findex __builtin_isgreater
4765 @findex __builtin_isgreaterequal
4766 @findex __builtin_isless
4767 @findex __builtin_islessequal
4768 @findex __builtin_islessgreater
4769 @findex __builtin_isunordered
4924 @findex fprintf_unlocked
4926 @findex fputs_unlocked
5036 @findex printf_unlocked
5065 @findex significandf
5066 @findex significandl
5133 GCC provides a large number of built-in functions other than the ones
5134 mentioned above. Some of these are for internal use in the processing
5135 of exceptions or variable-length argument lists and will not be
5136 documented here because they may change from time to time; we do not
5137 recommend general use of these functions.
5139 The remaining functions are provided for optimization purposes.
5141 @opindex fno-builtin
5142 GCC includes built-in versions of many of the functions in the standard
5143 C library. The versions prefixed with @code{__builtin_} will always be
5144 treated as having the same meaning as the C library function even if you
5145 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
5146 Many of these functions are only optimized in certain cases; if they are
5147 not optimized in a particular case, a call to the library function will
5152 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
5153 @option{-std=c99}), the functions
5154 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
5155 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
5156 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
5157 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
5158 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
5159 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
5160 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
5161 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
5162 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
5163 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
5164 @code{significandf}, @code{significandl}, @code{significand},
5165 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
5166 @code{strdup}, @code{strfmon}, @code{toascii}, @code{y0f}, @code{y0l},
5167 @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
5169 may be handled as built-in functions.
5170 All these functions have corresponding versions
5171 prefixed with @code{__builtin_}, which may be used even in strict C89
5174 The ISO C99 functions
5175 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
5176 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
5177 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
5178 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
5179 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
5180 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
5181 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
5182 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
5183 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
5184 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
5185 @code{cimagl}, @code{cimag}, @code{conjf}, @code{conjl}, @code{conj},
5186 @code{copysignf}, @code{copysignl}, @code{copysign}, @code{cpowf},
5187 @code{cpowl}, @code{cpow}, @code{cprojf}, @code{cprojl}, @code{cproj},
5188 @code{crealf}, @code{creall}, @code{creal}, @code{csinf}, @code{csinhf},
5189 @code{csinhl}, @code{csinh}, @code{csinl}, @code{csin}, @code{csqrtf},
5190 @code{csqrtl}, @code{csqrt}, @code{ctanf}, @code{ctanhf}, @code{ctanhl},
5191 @code{ctanh}, @code{ctanl}, @code{ctan}, @code{erfcf}, @code{erfcl},
5192 @code{erfc}, @code{erff}, @code{erfl}, @code{erf}, @code{exp2f},
5193 @code{exp2l}, @code{exp2}, @code{expm1f}, @code{expm1l}, @code{expm1},
5194 @code{fdimf}, @code{fdiml}, @code{fdim}, @code{fmaf}, @code{fmal},
5195 @code{fmaxf}, @code{fmaxl}, @code{fmax}, @code{fma}, @code{fminf},
5196 @code{fminl}, @code{fmin}, @code{hypotf}, @code{hypotl}, @code{hypot},
5197 @code{ilogbf}, @code{ilogbl}, @code{ilogb}, @code{imaxabs},
5198 @code{isblank}, @code{iswblank}, @code{lgammaf}, @code{lgammal},
5199 @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
5200 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
5201 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
5202 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
5203 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
5204 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
5205 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
5206 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
5207 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
5208 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
5209 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
5210 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
5211 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
5212 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
5213 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
5214 are handled as built-in functions
5215 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5217 There are also built-in versions of the ISO C99 functions
5218 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5219 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5220 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5221 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5222 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5223 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5224 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5225 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5226 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5227 that are recognized in any mode since ISO C90 reserves these names for
5228 the purpose to which ISO C99 puts them. All these functions have
5229 corresponding versions prefixed with @code{__builtin_}.
5231 The ISO C94 functions
5232 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
5233 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
5234 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
5236 are handled as built-in functions
5237 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5239 The ISO C90 functions
5240 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5241 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5242 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5243 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
5244 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
5245 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
5246 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
5247 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
5248 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
5249 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
5250 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
5251 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
5252 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
5253 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
5254 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
5255 @code{vprintf} and @code{vsprintf}
5256 are all recognized as built-in functions unless
5257 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5258 is specified for an individual function). All of these functions have
5259 corresponding versions prefixed with @code{__builtin_}.
5261 GCC provides built-in versions of the ISO C99 floating point comparison
5262 macros that avoid raising exceptions for unordered operands. They have
5263 the same names as the standard macros ( @code{isgreater},
5264 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5265 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5266 prefixed. We intend for a library implementor to be able to simply
5267 @code{#define} each standard macro to its built-in equivalent.
5269 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
5271 You can use the built-in function @code{__builtin_types_compatible_p} to
5272 determine whether two types are the same.
5274 This built-in function returns 1 if the unqualified versions of the
5275 types @var{type1} and @var{type2} (which are types, not expressions) are
5276 compatible, 0 otherwise. The result of this built-in function can be
5277 used in integer constant expressions.
5279 This built-in function ignores top level qualifiers (e.g., @code{const},
5280 @code{volatile}). For example, @code{int} is equivalent to @code{const
5283 The type @code{int[]} and @code{int[5]} are compatible. On the other
5284 hand, @code{int} and @code{char *} are not compatible, even if the size
5285 of their types, on the particular architecture are the same. Also, the
5286 amount of pointer indirection is taken into account when determining
5287 similarity. Consequently, @code{short *} is not similar to
5288 @code{short **}. Furthermore, two types that are typedefed are
5289 considered compatible if their underlying types are compatible.
5291 An @code{enum} type is not considered to be compatible with another
5292 @code{enum} type even if both are compatible with the same integer
5293 type; this is what the C standard specifies.
5294 For example, @code{enum @{foo, bar@}} is not similar to
5295 @code{enum @{hot, dog@}}.
5297 You would typically use this function in code whose execution varies
5298 depending on the arguments' types. For example:
5304 if (__builtin_types_compatible_p (typeof (x), long double)) \
5305 tmp = foo_long_double (tmp); \
5306 else if (__builtin_types_compatible_p (typeof (x), double)) \
5307 tmp = foo_double (tmp); \
5308 else if (__builtin_types_compatible_p (typeof (x), float)) \
5309 tmp = foo_float (tmp); \
5316 @emph{Note:} This construct is only available for C.
5320 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
5322 You can use the built-in function @code{__builtin_choose_expr} to
5323 evaluate code depending on the value of a constant expression. This
5324 built-in function returns @var{exp1} if @var{const_exp}, which is a
5325 constant expression that must be able to be determined at compile time,
5326 is nonzero. Otherwise it returns 0.
5328 This built-in function is analogous to the @samp{? :} operator in C,
5329 except that the expression returned has its type unaltered by promotion
5330 rules. Also, the built-in function does not evaluate the expression
5331 that was not chosen. For example, if @var{const_exp} evaluates to true,
5332 @var{exp2} is not evaluated even if it has side-effects.
5334 This built-in function can return an lvalue if the chosen argument is an
5337 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5338 type. Similarly, if @var{exp2} is returned, its return type is the same
5345 __builtin_choose_expr ( \
5346 __builtin_types_compatible_p (typeof (x), double), \
5348 __builtin_choose_expr ( \
5349 __builtin_types_compatible_p (typeof (x), float), \
5351 /* @r{The void expression results in a compile-time error} \
5352 @r{when assigning the result to something.} */ \
5356 @emph{Note:} This construct is only available for C. Furthermore, the
5357 unused expression (@var{exp1} or @var{exp2} depending on the value of
5358 @var{const_exp}) may still generate syntax errors. This may change in
5363 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5364 You can use the built-in function @code{__builtin_constant_p} to
5365 determine if a value is known to be constant at compile-time and hence
5366 that GCC can perform constant-folding on expressions involving that
5367 value. The argument of the function is the value to test. The function
5368 returns the integer 1 if the argument is known to be a compile-time
5369 constant and 0 if it is not known to be a compile-time constant. A
5370 return of 0 does not indicate that the value is @emph{not} a constant,
5371 but merely that GCC cannot prove it is a constant with the specified
5372 value of the @option{-O} option.
5374 You would typically use this function in an embedded application where
5375 memory was a critical resource. If you have some complex calculation,
5376 you may want it to be folded if it involves constants, but need to call
5377 a function if it does not. For example:
5380 #define Scale_Value(X) \
5381 (__builtin_constant_p (X) \
5382 ? ((X) * SCALE + OFFSET) : Scale (X))
5385 You may use this built-in function in either a macro or an inline
5386 function. However, if you use it in an inlined function and pass an
5387 argument of the function as the argument to the built-in, GCC will
5388 never return 1 when you call the inline function with a string constant
5389 or compound literal (@pxref{Compound Literals}) and will not return 1
5390 when you pass a constant numeric value to the inline function unless you
5391 specify the @option{-O} option.
5393 You may also use @code{__builtin_constant_p} in initializers for static
5394 data. For instance, you can write
5397 static const int table[] = @{
5398 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5404 This is an acceptable initializer even if @var{EXPRESSION} is not a
5405 constant expression. GCC must be more conservative about evaluating the
5406 built-in in this case, because it has no opportunity to perform
5409 Previous versions of GCC did not accept this built-in in data
5410 initializers. The earliest version where it is completely safe is
5414 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5415 @opindex fprofile-arcs
5416 You may use @code{__builtin_expect} to provide the compiler with
5417 branch prediction information. In general, you should prefer to
5418 use actual profile feedback for this (@option{-fprofile-arcs}), as
5419 programmers are notoriously bad at predicting how their programs
5420 actually perform. However, there are applications in which this
5421 data is hard to collect.
5423 The return value is the value of @var{exp}, which should be an
5424 integral expression. The value of @var{c} must be a compile-time
5425 constant. The semantics of the built-in are that it is expected
5426 that @var{exp} == @var{c}. For example:
5429 if (__builtin_expect (x, 0))
5434 would indicate that we do not expect to call @code{foo}, since
5435 we expect @code{x} to be zero. Since you are limited to integral
5436 expressions for @var{exp}, you should use constructions such as
5439 if (__builtin_expect (ptr != NULL, 1))
5444 when testing pointer or floating-point values.
5447 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5448 This function is used to minimize cache-miss latency by moving data into
5449 a cache before it is accessed.
5450 You can insert calls to @code{__builtin_prefetch} into code for which
5451 you know addresses of data in memory that is likely to be accessed soon.
5452 If the target supports them, data prefetch instructions will be generated.
5453 If the prefetch is done early enough before the access then the data will
5454 be in the cache by the time it is accessed.
5456 The value of @var{addr} is the address of the memory to prefetch.
5457 There are two optional arguments, @var{rw} and @var{locality}.
5458 The value of @var{rw} is a compile-time constant one or zero; one
5459 means that the prefetch is preparing for a write to the memory address
5460 and zero, the default, means that the prefetch is preparing for a read.
5461 The value @var{locality} must be a compile-time constant integer between
5462 zero and three. A value of zero means that the data has no temporal
5463 locality, so it need not be left in the cache after the access. A value
5464 of three means that the data has a high degree of temporal locality and
5465 should be left in all levels of cache possible. Values of one and two
5466 mean, respectively, a low or moderate degree of temporal locality. The
5470 for (i = 0; i < n; i++)
5473 __builtin_prefetch (&a[i+j], 1, 1);
5474 __builtin_prefetch (&b[i+j], 0, 1);
5479 Data prefetch does not generate faults if @var{addr} is invalid, but
5480 the address expression itself must be valid. For example, a prefetch
5481 of @code{p->next} will not fault if @code{p->next} is not a valid
5482 address, but evaluation will fault if @code{p} is not a valid address.
5484 If the target does not support data prefetch, the address expression
5485 is evaluated if it includes side effects but no other code is generated
5486 and GCC does not issue a warning.
5489 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5490 Returns a positive infinity, if supported by the floating-point format,
5491 else @code{DBL_MAX}. This function is suitable for implementing the
5492 ISO C macro @code{HUGE_VAL}.
5495 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5496 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5499 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5500 Similar to @code{__builtin_huge_val}, except the return
5501 type is @code{long double}.
5504 @deftypefn {Built-in Function} double __builtin_inf (void)
5505 Similar to @code{__builtin_huge_val}, except a warning is generated
5506 if the target floating-point format does not support infinities.
5507 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5510 @deftypefn {Built-in Function} float __builtin_inff (void)
5511 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5514 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5515 Similar to @code{__builtin_inf}, except the return
5516 type is @code{long double}.
5519 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5520 This is an implementation of the ISO C99 function @code{nan}.
5522 Since ISO C99 defines this function in terms of @code{strtod}, which we
5523 do not implement, a description of the parsing is in order. The string
5524 is parsed as by @code{strtol}; that is, the base is recognized by
5525 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5526 in the significand such that the least significant bit of the number
5527 is at the least significant bit of the significand. The number is
5528 truncated to fit the significand field provided. The significand is
5529 forced to be a quiet NaN.
5531 This function, if given a string literal, is evaluated early enough
5532 that it is considered a compile-time constant.
5535 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
5536 Similar to @code{__builtin_nan}, except the return type is @code{float}.
5539 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
5540 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
5543 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
5544 Similar to @code{__builtin_nan}, except the significand is forced
5545 to be a signaling NaN. The @code{nans} function is proposed by
5546 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
5549 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
5550 Similar to @code{__builtin_nans}, except the return type is @code{float}.
5553 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
5554 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
5557 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
5558 Returns one plus the index of the least significant 1-bit of @var{x}, or
5559 if @var{x} is zero, returns zero.
5562 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
5563 Returns the number of leading 0-bits in @var{x}, starting at the most
5564 significant bit position. If @var{x} is 0, the result is undefined.
5567 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
5568 Returns the number of trailing 0-bits in @var{x}, starting at the least
5569 significant bit position. If @var{x} is 0, the result is undefined.
5572 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
5573 Returns the number of 1-bits in @var{x}.
5576 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
5577 Returns the parity of @var{x}, i.@:e. the number of 1-bits in @var{x}
5581 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
5582 Similar to @code{__builtin_ffs}, except the argument type is
5583 @code{unsigned long}.
5586 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
5587 Similar to @code{__builtin_clz}, except the argument type is
5588 @code{unsigned long}.
5591 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
5592 Similar to @code{__builtin_ctz}, except the argument type is
5593 @code{unsigned long}.
5596 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
5597 Similar to @code{__builtin_popcount}, except the argument type is
5598 @code{unsigned long}.
5601 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
5602 Similar to @code{__builtin_parity}, except the argument type is
5603 @code{unsigned long}.
5606 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
5607 Similar to @code{__builtin_ffs}, except the argument type is
5608 @code{unsigned long long}.
5611 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
5612 Similar to @code{__builtin_clz}, except the argument type is
5613 @code{unsigned long long}.
5616 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
5617 Similar to @code{__builtin_ctz}, except the argument type is
5618 @code{unsigned long long}.
5621 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
5622 Similar to @code{__builtin_popcount}, except the argument type is
5623 @code{unsigned long long}.
5626 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
5627 Similar to @code{__builtin_parity}, except the argument type is
5628 @code{unsigned long long}.
5632 @node Target Builtins
5633 @section Built-in Functions Specific to Particular Target Machines
5635 On some target machines, GCC supports many built-in functions specific
5636 to those machines. Generally these generate calls to specific machine
5637 instructions, but allow the compiler to schedule those calls.
5640 * Alpha Built-in Functions::
5641 * ARM Built-in Functions::
5642 * X86 Built-in Functions::
5643 * PowerPC AltiVec Built-in Functions::
5646 @node Alpha Built-in Functions
5647 @subsection Alpha Built-in Functions
5649 These built-in functions are available for the Alpha family of
5650 processors, depending on the command-line switches used.
5652 The following built-in functions are always available. They
5653 all generate the machine instruction that is part of the name.
5656 long __builtin_alpha_implver (void)
5657 long __builtin_alpha_rpcc (void)
5658 long __builtin_alpha_amask (long)
5659 long __builtin_alpha_cmpbge (long, long)
5660 long __builtin_alpha_extbl (long, long)
5661 long __builtin_alpha_extwl (long, long)
5662 long __builtin_alpha_extll (long, long)
5663 long __builtin_alpha_extql (long, long)
5664 long __builtin_alpha_extwh (long, long)
5665 long __builtin_alpha_extlh (long, long)
5666 long __builtin_alpha_extqh (long, long)
5667 long __builtin_alpha_insbl (long, long)
5668 long __builtin_alpha_inswl (long, long)
5669 long __builtin_alpha_insll (long, long)
5670 long __builtin_alpha_insql (long, long)
5671 long __builtin_alpha_inswh (long, long)
5672 long __builtin_alpha_inslh (long, long)
5673 long __builtin_alpha_insqh (long, long)
5674 long __builtin_alpha_mskbl (long, long)
5675 long __builtin_alpha_mskwl (long, long)
5676 long __builtin_alpha_mskll (long, long)
5677 long __builtin_alpha_mskql (long, long)
5678 long __builtin_alpha_mskwh (long, long)
5679 long __builtin_alpha_msklh (long, long)
5680 long __builtin_alpha_mskqh (long, long)
5681 long __builtin_alpha_umulh (long, long)
5682 long __builtin_alpha_zap (long, long)
5683 long __builtin_alpha_zapnot (long, long)
5686 The following built-in functions are always with @option{-mmax}
5687 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
5688 later. They all generate the machine instruction that is part
5692 long __builtin_alpha_pklb (long)
5693 long __builtin_alpha_pkwb (long)
5694 long __builtin_alpha_unpkbl (long)
5695 long __builtin_alpha_unpkbw (long)
5696 long __builtin_alpha_minub8 (long, long)
5697 long __builtin_alpha_minsb8 (long, long)
5698 long __builtin_alpha_minuw4 (long, long)
5699 long __builtin_alpha_minsw4 (long, long)
5700 long __builtin_alpha_maxub8 (long, long)
5701 long __builtin_alpha_maxsb8 (long, long)
5702 long __builtin_alpha_maxuw4 (long, long)
5703 long __builtin_alpha_maxsw4 (long, long)
5704 long __builtin_alpha_perr (long, long)
5707 The following built-in functions are always with @option{-mcix}
5708 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
5709 later. They all generate the machine instruction that is part
5713 long __builtin_alpha_cttz (long)
5714 long __builtin_alpha_ctlz (long)
5715 long __builtin_alpha_ctpop (long)
5718 The following builtins are available on systems that use the OSF/1
5719 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
5720 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
5721 @code{rdval} and @code{wrval}.
5724 void *__builtin_thread_pointer (void)
5725 void __builtin_set_thread_pointer (void *)
5728 @node ARM Built-in Functions
5729 @subsection ARM Built-in Functions
5731 These built-in functions are available for the ARM family of
5732 processors, when the @option{-mcpu=iwmmxt} switch is used:
5735 typedef int v2si __attribute__ ((vector_size (8)));
5736 typedef short v4hi __attribute__ ((vector_size (8)));
5737 typedef char v8qi __attribute__ ((vector_size (8)));
5739 int __builtin_arm_getwcx (int)
5740 void __builtin_arm_setwcx (int, int)
5741 int __builtin_arm_textrmsb (v8qi, int)
5742 int __builtin_arm_textrmsh (v4hi, int)
5743 int __builtin_arm_textrmsw (v2si, int)
5744 int __builtin_arm_textrmub (v8qi, int)
5745 int __builtin_arm_textrmuh (v4hi, int)
5746 int __builtin_arm_textrmuw (v2si, int)
5747 v8qi __builtin_arm_tinsrb (v8qi, int)
5748 v4hi __builtin_arm_tinsrh (v4hi, int)
5749 v2si __builtin_arm_tinsrw (v2si, int)
5750 long long __builtin_arm_tmia (long long, int, int)
5751 long long __builtin_arm_tmiabb (long long, int, int)
5752 long long __builtin_arm_tmiabt (long long, int, int)
5753 long long __builtin_arm_tmiaph (long long, int, int)
5754 long long __builtin_arm_tmiatb (long long, int, int)
5755 long long __builtin_arm_tmiatt (long long, int, int)
5756 int __builtin_arm_tmovmskb (v8qi)
5757 int __builtin_arm_tmovmskh (v4hi)
5758 int __builtin_arm_tmovmskw (v2si)
5759 long long __builtin_arm_waccb (v8qi)
5760 long long __builtin_arm_wacch (v4hi)
5761 long long __builtin_arm_waccw (v2si)
5762 v8qi __builtin_arm_waddb (v8qi, v8qi)
5763 v8qi __builtin_arm_waddbss (v8qi, v8qi)
5764 v8qi __builtin_arm_waddbus (v8qi, v8qi)
5765 v4hi __builtin_arm_waddh (v4hi, v4hi)
5766 v4hi __builtin_arm_waddhss (v4hi, v4hi)
5767 v4hi __builtin_arm_waddhus (v4hi, v4hi)
5768 v2si __builtin_arm_waddw (v2si, v2si)
5769 v2si __builtin_arm_waddwss (v2si, v2si)
5770 v2si __builtin_arm_waddwus (v2si, v2si)
5771 v8qi __builtin_arm_walign (v8qi, v8qi, int)
5772 long long __builtin_arm_wand(long long, long long)
5773 long long __builtin_arm_wandn (long long, long long)
5774 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
5775 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
5776 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
5777 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
5778 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
5779 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
5780 v2si __builtin_arm_wcmpeqw (v2si, v2si)
5781 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
5782 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
5783 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
5784 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
5785 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
5786 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
5787 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
5788 long long __builtin_arm_wmacsz (v4hi, v4hi)
5789 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
5790 long long __builtin_arm_wmacuz (v4hi, v4hi)
5791 v4hi __builtin_arm_wmadds (v4hi, v4hi)
5792 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
5793 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
5794 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
5795 v2si __builtin_arm_wmaxsw (v2si, v2si)
5796 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
5797 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
5798 v2si __builtin_arm_wmaxuw (v2si, v2si)
5799 v8qi __builtin_arm_wminsb (v8qi, v8qi)
5800 v4hi __builtin_arm_wminsh (v4hi, v4hi)
5801 v2si __builtin_arm_wminsw (v2si, v2si)
5802 v8qi __builtin_arm_wminub (v8qi, v8qi)
5803 v4hi __builtin_arm_wminuh (v4hi, v4hi)
5804 v2si __builtin_arm_wminuw (v2si, v2si)
5805 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
5806 v4hi __builtin_arm_wmulul (v4hi, v4hi)
5807 v4hi __builtin_arm_wmulum (v4hi, v4hi)
5808 long long __builtin_arm_wor (long long, long long)
5809 v2si __builtin_arm_wpackdss (long long, long long)
5810 v2si __builtin_arm_wpackdus (long long, long long)
5811 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
5812 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
5813 v4hi __builtin_arm_wpackwss (v2si, v2si)
5814 v4hi __builtin_arm_wpackwus (v2si, v2si)
5815 long long __builtin_arm_wrord (long long, long long)
5816 long long __builtin_arm_wrordi (long long, int)
5817 v4hi __builtin_arm_wrorh (v4hi, long long)
5818 v4hi __builtin_arm_wrorhi (v4hi, int)
5819 v2si __builtin_arm_wrorw (v2si, long long)
5820 v2si __builtin_arm_wrorwi (v2si, int)
5821 v2si __builtin_arm_wsadb (v8qi, v8qi)
5822 v2si __builtin_arm_wsadbz (v8qi, v8qi)
5823 v2si __builtin_arm_wsadh (v4hi, v4hi)
5824 v2si __builtin_arm_wsadhz (v4hi, v4hi)
5825 v4hi __builtin_arm_wshufh (v4hi, int)
5826 long long __builtin_arm_wslld (long long, long long)
5827 long long __builtin_arm_wslldi (long long, int)
5828 v4hi __builtin_arm_wsllh (v4hi, long long)
5829 v4hi __builtin_arm_wsllhi (v4hi, int)
5830 v2si __builtin_arm_wsllw (v2si, long long)
5831 v2si __builtin_arm_wsllwi (v2si, int)
5832 long long __builtin_arm_wsrad (long long, long long)
5833 long long __builtin_arm_wsradi (long long, int)
5834 v4hi __builtin_arm_wsrah (v4hi, long long)
5835 v4hi __builtin_arm_wsrahi (v4hi, int)
5836 v2si __builtin_arm_wsraw (v2si, long long)
5837 v2si __builtin_arm_wsrawi (v2si, int)
5838 long long __builtin_arm_wsrld (long long, long long)
5839 long long __builtin_arm_wsrldi (long long, int)
5840 v4hi __builtin_arm_wsrlh (v4hi, long long)
5841 v4hi __builtin_arm_wsrlhi (v4hi, int)
5842 v2si __builtin_arm_wsrlw (v2si, long long)
5843 v2si __builtin_arm_wsrlwi (v2si, int)
5844 v8qi __builtin_arm_wsubb (v8qi, v8qi)
5845 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
5846 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
5847 v4hi __builtin_arm_wsubh (v4hi, v4hi)
5848 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
5849 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
5850 v2si __builtin_arm_wsubw (v2si, v2si)
5851 v2si __builtin_arm_wsubwss (v2si, v2si)
5852 v2si __builtin_arm_wsubwus (v2si, v2si)
5853 v4hi __builtin_arm_wunpckehsb (v8qi)
5854 v2si __builtin_arm_wunpckehsh (v4hi)
5855 long long __builtin_arm_wunpckehsw (v2si)
5856 v4hi __builtin_arm_wunpckehub (v8qi)
5857 v2si __builtin_arm_wunpckehuh (v4hi)
5858 long long __builtin_arm_wunpckehuw (v2si)
5859 v4hi __builtin_arm_wunpckelsb (v8qi)
5860 v2si __builtin_arm_wunpckelsh (v4hi)
5861 long long __builtin_arm_wunpckelsw (v2si)
5862 v4hi __builtin_arm_wunpckelub (v8qi)
5863 v2si __builtin_arm_wunpckeluh (v4hi)
5864 long long __builtin_arm_wunpckeluw (v2si)
5865 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
5866 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
5867 v2si __builtin_arm_wunpckihw (v2si, v2si)
5868 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
5869 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
5870 v2si __builtin_arm_wunpckilw (v2si, v2si)
5871 long long __builtin_arm_wxor (long long, long long)
5872 long long __builtin_arm_wzero ()
5875 @node X86 Built-in Functions
5876 @subsection X86 Built-in Functions
5878 These built-in functions are available for the i386 and x86-64 family
5879 of computers, depending on the command-line switches used.
5881 The following machine modes are available for use with MMX built-in functions
5882 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
5883 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
5884 vector of eight 8-bit integers. Some of the built-in functions operate on
5885 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
5887 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
5888 of two 32-bit floating point values.
5890 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
5891 floating point values. Some instructions use a vector of four 32-bit
5892 integers, these use @code{V4SI}. Finally, some instructions operate on an
5893 entire vector register, interpreting it as a 128-bit integer, these use mode
5896 The following built-in functions are made available by @option{-mmmx}.
5897 All of them generate the machine instruction that is part of the name.
5900 v8qi __builtin_ia32_paddb (v8qi, v8qi)
5901 v4hi __builtin_ia32_paddw (v4hi, v4hi)
5902 v2si __builtin_ia32_paddd (v2si, v2si)
5903 v8qi __builtin_ia32_psubb (v8qi, v8qi)
5904 v4hi __builtin_ia32_psubw (v4hi, v4hi)
5905 v2si __builtin_ia32_psubd (v2si, v2si)
5906 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
5907 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
5908 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
5909 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
5910 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
5911 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
5912 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
5913 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
5914 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
5915 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
5916 di __builtin_ia32_pand (di, di)
5917 di __builtin_ia32_pandn (di,di)
5918 di __builtin_ia32_por (di, di)
5919 di __builtin_ia32_pxor (di, di)
5920 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
5921 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
5922 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
5923 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
5924 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
5925 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
5926 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
5927 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
5928 v2si __builtin_ia32_punpckhdq (v2si, v2si)
5929 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
5930 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
5931 v2si __builtin_ia32_punpckldq (v2si, v2si)
5932 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
5933 v4hi __builtin_ia32_packssdw (v2si, v2si)
5934 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
5937 The following built-in functions are made available either with
5938 @option{-msse}, or with a combination of @option{-m3dnow} and
5939 @option{-march=athlon}. All of them generate the machine
5940 instruction that is part of the name.
5943 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
5944 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
5945 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
5946 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
5947 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
5948 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
5949 v8qi __builtin_ia32_pminub (v8qi, v8qi)
5950 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
5951 int __builtin_ia32_pextrw (v4hi, int)
5952 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
5953 int __builtin_ia32_pmovmskb (v8qi)
5954 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
5955 void __builtin_ia32_movntq (di *, di)
5956 void __builtin_ia32_sfence (void)
5959 The following built-in functions are available when @option{-msse} is used.
5960 All of them generate the machine instruction that is part of the name.
5963 int __builtin_ia32_comieq (v4sf, v4sf)
5964 int __builtin_ia32_comineq (v4sf, v4sf)
5965 int __builtin_ia32_comilt (v4sf, v4sf)
5966 int __builtin_ia32_comile (v4sf, v4sf)
5967 int __builtin_ia32_comigt (v4sf, v4sf)
5968 int __builtin_ia32_comige (v4sf, v4sf)
5969 int __builtin_ia32_ucomieq (v4sf, v4sf)
5970 int __builtin_ia32_ucomineq (v4sf, v4sf)
5971 int __builtin_ia32_ucomilt (v4sf, v4sf)
5972 int __builtin_ia32_ucomile (v4sf, v4sf)
5973 int __builtin_ia32_ucomigt (v4sf, v4sf)
5974 int __builtin_ia32_ucomige (v4sf, v4sf)
5975 v4sf __builtin_ia32_addps (v4sf, v4sf)
5976 v4sf __builtin_ia32_subps (v4sf, v4sf)
5977 v4sf __builtin_ia32_mulps (v4sf, v4sf)
5978 v4sf __builtin_ia32_divps (v4sf, v4sf)
5979 v4sf __builtin_ia32_addss (v4sf, v4sf)
5980 v4sf __builtin_ia32_subss (v4sf, v4sf)
5981 v4sf __builtin_ia32_mulss (v4sf, v4sf)
5982 v4sf __builtin_ia32_divss (v4sf, v4sf)
5983 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
5984 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
5985 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
5986 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
5987 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
5988 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
5989 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
5990 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
5991 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
5992 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
5993 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
5994 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
5995 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
5996 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
5997 v4si __builtin_ia32_cmpless (v4sf, v4sf)
5998 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
5999 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
6000 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
6001 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
6002 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
6003 v4sf __builtin_ia32_maxps (v4sf, v4sf)
6004 v4sf __builtin_ia32_maxss (v4sf, v4sf)
6005 v4sf __builtin_ia32_minps (v4sf, v4sf)
6006 v4sf __builtin_ia32_minss (v4sf, v4sf)
6007 v4sf __builtin_ia32_andps (v4sf, v4sf)
6008 v4sf __builtin_ia32_andnps (v4sf, v4sf)
6009 v4sf __builtin_ia32_orps (v4sf, v4sf)
6010 v4sf __builtin_ia32_xorps (v4sf, v4sf)
6011 v4sf __builtin_ia32_movss (v4sf, v4sf)
6012 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
6013 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
6014 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
6015 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
6016 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
6017 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
6018 v2si __builtin_ia32_cvtps2pi (v4sf)
6019 int __builtin_ia32_cvtss2si (v4sf)
6020 v2si __builtin_ia32_cvttps2pi (v4sf)
6021 int __builtin_ia32_cvttss2si (v4sf)
6022 v4sf __builtin_ia32_rcpps (v4sf)
6023 v4sf __builtin_ia32_rsqrtps (v4sf)
6024 v4sf __builtin_ia32_sqrtps (v4sf)
6025 v4sf __builtin_ia32_rcpss (v4sf)
6026 v4sf __builtin_ia32_rsqrtss (v4sf)
6027 v4sf __builtin_ia32_sqrtss (v4sf)
6028 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
6029 void __builtin_ia32_movntps (float *, v4sf)
6030 int __builtin_ia32_movmskps (v4sf)
6033 The following built-in functions are available when @option{-msse} is used.
6036 @item v4sf __builtin_ia32_loadaps (float *)
6037 Generates the @code{movaps} machine instruction as a load from memory.
6038 @item void __builtin_ia32_storeaps (float *, v4sf)
6039 Generates the @code{movaps} machine instruction as a store to memory.
6040 @item v4sf __builtin_ia32_loadups (float *)
6041 Generates the @code{movups} machine instruction as a load from memory.
6042 @item void __builtin_ia32_storeups (float *, v4sf)
6043 Generates the @code{movups} machine instruction as a store to memory.
6044 @item v4sf __builtin_ia32_loadsss (float *)
6045 Generates the @code{movss} machine instruction as a load from memory.
6046 @item void __builtin_ia32_storess (float *, v4sf)
6047 Generates the @code{movss} machine instruction as a store to memory.
6048 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
6049 Generates the @code{movhps} machine instruction as a load from memory.
6050 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
6051 Generates the @code{movlps} machine instruction as a load from memory
6052 @item void __builtin_ia32_storehps (v4sf, v2si *)
6053 Generates the @code{movhps} machine instruction as a store to memory.
6054 @item void __builtin_ia32_storelps (v4sf, v2si *)
6055 Generates the @code{movlps} machine instruction as a store to memory.
6058 The following built-in functions are available when @option{-msse3} is used.
6059 All of them generate the machine instruction that is part of the name.
6062 v2df __builtin_ia32_addsubpd (v2df, v2df)
6063 v2df __builtin_ia32_addsubps (v2df, v2df)
6064 v2df __builtin_ia32_haddpd (v2df, v2df)
6065 v2df __builtin_ia32_haddps (v2df, v2df)
6066 v2df __builtin_ia32_hsubpd (v2df, v2df)
6067 v2df __builtin_ia32_hsubps (v2df, v2df)
6068 v16qi __builtin_ia32_lddqu (char const *)
6069 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
6070 v2df __builtin_ia32_movddup (v2df)
6071 v4sf __builtin_ia32_movshdup (v4sf)
6072 v4sf __builtin_ia32_movsldup (v4sf)
6073 void __builtin_ia32_mwait (unsigned int, unsigned int)
6076 The following built-in functions are available when @option{-msse3} is used.
6079 @item v2df __builtin_ia32_loadddup (double const *)
6080 Generates the @code{movddup} machine instruction as a load from memory.
6083 The following built-in functions are available when @option{-m3dnow} is used.
6084 All of them generate the machine instruction that is part of the name.
6087 void __builtin_ia32_femms (void)
6088 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
6089 v2si __builtin_ia32_pf2id (v2sf)
6090 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
6091 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
6092 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
6093 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
6094 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
6095 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
6096 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
6097 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
6098 v2sf __builtin_ia32_pfrcp (v2sf)
6099 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
6100 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
6101 v2sf __builtin_ia32_pfrsqrt (v2sf)
6102 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
6103 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
6104 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
6105 v2sf __builtin_ia32_pi2fd (v2si)
6106 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
6109 The following built-in functions are available when both @option{-m3dnow}
6110 and @option{-march=athlon} are used. All of them generate the machine
6111 instruction that is part of the name.
6114 v2si __builtin_ia32_pf2iw (v2sf)
6115 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
6116 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
6117 v2sf __builtin_ia32_pi2fw (v2si)
6118 v2sf __builtin_ia32_pswapdsf (v2sf)
6119 v2si __builtin_ia32_pswapdsi (v2si)
6122 @node PowerPC AltiVec Built-in Functions
6123 @subsection PowerPC AltiVec Built-in Functions
6125 These built-in functions are available for the PowerPC family
6126 of computers, depending on the command-line switches used.
6128 The following machine modes are available for use with AltiVec built-in
6129 functions (@pxref{Vector Extensions}): @code{V4SI} for a vector of four
6130 32-bit integers, @code{V4SF} for a vector of four 32-bit floating point
6131 numbers, @code{V8HI} for a vector of eight 16-bit integers, and
6132 @code{V16QI} for a vector of sixteen 8-bit integers.
6134 The following functions are made available by including
6135 @code{<altivec.h>} and using @option{-maltivec} and
6136 @option{-mabi=altivec}. The functions implement the functionality
6137 described in Motorola's AltiVec Programming Interface Manual.
6139 There are a few differences from Motorola's documentation and GCC's
6140 implementation. Vector constants are done with curly braces (not
6141 parentheses). Vector initializers require no casts if the vector
6142 constant is of the same type as the variable it is initializing. The
6143 @code{vector bool} type is deprecated and will be discontinued in
6144 further revisions. Use @code{vector signed} instead. If @code{signed}
6145 or @code{unsigned} is omitted, the vector type will default to
6146 @code{signed}. Lastly, all overloaded functions are implemented with macros
6147 for the C implementation. So code the following example will not work:
6150 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
6153 Since vec_add is a macro, the vector constant in the above example will
6154 be treated as four different arguments. Wrap the entire argument in
6155 parentheses for this to work. The C++ implementation does not use
6158 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
6159 Internally, GCC uses built-in functions to achieve the functionality in
6160 the aforementioned header file, but they are not supported and are
6161 subject to change without notice.
6164 vector signed char vec_abs (vector signed char, vector signed char);
6165 vector signed short vec_abs (vector signed short, vector signed short);
6166 vector signed int vec_abs (vector signed int, vector signed int);
6167 vector signed float vec_abs (vector signed float, vector signed float);
6169 vector signed char vec_abss (vector signed char, vector signed char);
6170 vector signed short vec_abss (vector signed short, vector signed short);
6172 vector signed char vec_add (vector signed char, vector signed char);
6173 vector unsigned char vec_add (vector signed char, vector unsigned char);
6175 vector unsigned char vec_add (vector unsigned char, vector signed char);
6177 vector unsigned char vec_add (vector unsigned char,
6178 vector unsigned char);
6179 vector signed short vec_add (vector signed short, vector signed short);
6180 vector unsigned short vec_add (vector signed short,
6181 vector unsigned short);
6182 vector unsigned short vec_add (vector unsigned short,
6183 vector signed short);
6184 vector unsigned short vec_add (vector unsigned short,
6185 vector unsigned short);
6186 vector signed int vec_add (vector signed int, vector signed int);
6187 vector unsigned int vec_add (vector signed int, vector unsigned int);
6188 vector unsigned int vec_add (vector unsigned int, vector signed int);
6189 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
6190 vector float vec_add (vector float, vector float);
6192 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
6194 vector unsigned char vec_adds (vector signed char,
6195 vector unsigned char);
6196 vector unsigned char vec_adds (vector unsigned char,
6197 vector signed char);
6198 vector unsigned char vec_adds (vector unsigned char,
6199 vector unsigned char);
6200 vector signed char vec_adds (vector signed char, vector signed char);
6201 vector unsigned short vec_adds (vector signed short,
6202 vector unsigned short);
6203 vector unsigned short vec_adds (vector unsigned short,
6204 vector signed short);
6205 vector unsigned short vec_adds (vector unsigned short,
6206 vector unsigned short);
6207 vector signed short vec_adds (vector signed short, vector signed short);
6209 vector unsigned int vec_adds (vector signed int, vector unsigned int);
6210 vector unsigned int vec_adds (vector unsigned int, vector signed int);
6211 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
6213 vector signed int vec_adds (vector signed int, vector signed int);
6215 vector float vec_and (vector float, vector float);
6216 vector float vec_and (vector float, vector signed int);
6217 vector float vec_and (vector signed int, vector float);
6218 vector signed int vec_and (vector signed int, vector signed int);
6219 vector unsigned int vec_and (vector signed int, vector unsigned int);
6220 vector unsigned int vec_and (vector unsigned int, vector signed int);
6221 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
6222 vector signed short vec_and (vector signed short, vector signed short);
6223 vector unsigned short vec_and (vector signed short,
6224 vector unsigned short);
6225 vector unsigned short vec_and (vector unsigned short,
6226 vector signed short);
6227 vector unsigned short vec_and (vector unsigned short,
6228 vector unsigned short);
6229 vector signed char vec_and (vector signed char, vector signed char);
6230 vector unsigned char vec_and (vector signed char, vector unsigned char);
6232 vector unsigned char vec_and (vector unsigned char, vector signed char);
6234 vector unsigned char vec_and (vector unsigned char,
6235 vector unsigned char);
6237 vector float vec_andc (vector float, vector float);
6238 vector float vec_andc (vector float, vector signed int);
6239 vector float vec_andc (vector signed int, vector float);
6240 vector signed int vec_andc (vector signed int, vector signed int);
6241 vector unsigned int vec_andc (vector signed int, vector unsigned int);
6242 vector unsigned int vec_andc (vector unsigned int, vector signed int);
6243 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
6245 vector signed short vec_andc (vector signed short, vector signed short);
6247 vector unsigned short vec_andc (vector signed short,
6248 vector unsigned short);
6249 vector unsigned short vec_andc (vector unsigned short,
6250 vector signed short);
6251 vector unsigned short vec_andc (vector unsigned short,
6252 vector unsigned short);
6253 vector signed char vec_andc (vector signed char, vector signed char);
6254 vector unsigned char vec_andc (vector signed char,
6255 vector unsigned char);
6256 vector unsigned char vec_andc (vector unsigned char,
6257 vector signed char);
6258 vector unsigned char vec_andc (vector unsigned char,
6259 vector unsigned char);
6261 vector unsigned char vec_avg (vector unsigned char,
6262 vector unsigned char);
6263 vector signed char vec_avg (vector signed char, vector signed char);
6264 vector unsigned short vec_avg (vector unsigned short,
6265 vector unsigned short);
6266 vector signed short vec_avg (vector signed short, vector signed short);
6267 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
6268 vector signed int vec_avg (vector signed int, vector signed int);
6270 vector float vec_ceil (vector float);
6272 vector signed int vec_cmpb (vector float, vector float);
6274 vector signed char vec_cmpeq (vector signed char, vector signed char);
6275 vector signed char vec_cmpeq (vector unsigned char,
6276 vector unsigned char);
6277 vector signed short vec_cmpeq (vector signed short,
6278 vector signed short);
6279 vector signed short vec_cmpeq (vector unsigned short,
6280 vector unsigned short);
6281 vector signed int vec_cmpeq (vector signed int, vector signed int);
6282 vector signed int vec_cmpeq (vector unsigned int, vector unsigned int);
6283 vector signed int vec_cmpeq (vector float, vector float);
6285 vector signed int vec_cmpge (vector float, vector float);
6287 vector signed char vec_cmpgt (vector unsigned char,
6288 vector unsigned char);
6289 vector signed char vec_cmpgt (vector signed char, vector signed char);
6290 vector signed short vec_cmpgt (vector unsigned short,
6291 vector unsigned short);
6292 vector signed short vec_cmpgt (vector signed short,
6293 vector signed short);
6294 vector signed int vec_cmpgt (vector unsigned int, vector unsigned int);
6295 vector signed int vec_cmpgt (vector signed int, vector signed int);
6296 vector signed int vec_cmpgt (vector float, vector float);
6298 vector signed int vec_cmple (vector float, vector float);
6300 vector signed char vec_cmplt (vector unsigned char,
6301 vector unsigned char);
6302 vector signed char vec_cmplt (vector signed char, vector signed char);
6303 vector signed short vec_cmplt (vector unsigned short,
6304 vector unsigned short);
6305 vector signed short vec_cmplt (vector signed short,
6306 vector signed short);
6307 vector signed int vec_cmplt (vector unsigned int, vector unsigned int);
6308 vector signed int vec_cmplt (vector signed int, vector signed int);
6309 vector signed int vec_cmplt (vector float, vector float);
6311 vector float vec_ctf (vector unsigned int, const char);
6312 vector float vec_ctf (vector signed int, const char);
6314 vector signed int vec_cts (vector float, const char);
6316 vector unsigned int vec_ctu (vector float, const char);
6318 void vec_dss (const char);
6320 void vec_dssall (void);
6322 void vec_dst (void *, int, const char);
6324 void vec_dstst (void *, int, const char);
6326 void vec_dststt (void *, int, const char);
6328 void vec_dstt (void *, int, const char);
6330 vector float vec_expte (vector float, vector float);
6332 vector float vec_floor (vector float, vector float);
6334 vector float vec_ld (int, vector float *);
6335 vector float vec_ld (int, float *):
6336 vector signed int vec_ld (int, int *);
6337 vector signed int vec_ld (int, vector signed int *);
6338 vector unsigned int vec_ld (int, vector unsigned int *);
6339 vector unsigned int vec_ld (int, unsigned int *);
6340 vector signed short vec_ld (int, short *, vector signed short *);
6341 vector unsigned short vec_ld (int, unsigned short *,
6342 vector unsigned short *);
6343 vector signed char vec_ld (int, signed char *);
6344 vector signed char vec_ld (int, vector signed char *);
6345 vector unsigned char vec_ld (int, unsigned char *);
6346 vector unsigned char vec_ld (int, vector unsigned char *);
6348 vector signed char vec_lde (int, signed char *);
6349 vector unsigned char vec_lde (int, unsigned char *);
6350 vector signed short vec_lde (int, short *);
6351 vector unsigned short vec_lde (int, unsigned short *);
6352 vector float vec_lde (int, float *);
6353 vector signed int vec_lde (int, int *);
6354 vector unsigned int vec_lde (int, unsigned int *);
6356 void float vec_ldl (int, float *);
6357 void float vec_ldl (int, vector float *);
6358 void signed int vec_ldl (int, vector signed int *);
6359 void signed int vec_ldl (int, int *);
6360 void unsigned int vec_ldl (int, unsigned int *);
6361 void unsigned int vec_ldl (int, vector unsigned int *);
6362 void signed short vec_ldl (int, vector signed short *);
6363 void signed short vec_ldl (int, short *);
6364 void unsigned short vec_ldl (int, vector unsigned short *);
6365 void unsigned short vec_ldl (int, unsigned short *);
6366 void signed char vec_ldl (int, vector signed char *);
6367 void signed char vec_ldl (int, signed char *);
6368 void unsigned char vec_ldl (int, vector unsigned char *);
6369 void unsigned char vec_ldl (int, unsigned char *);
6371 vector float vec_loge (vector float);
6373 vector unsigned char vec_lvsl (int, void *, int *);
6375 vector unsigned char vec_lvsr (int, void *, int *);
6377 vector float vec_madd (vector float, vector float, vector float);
6379 vector signed short vec_madds (vector signed short, vector signed short,
6380 vector signed short);
6382 vector unsigned char vec_max (vector signed char, vector unsigned char);
6384 vector unsigned char vec_max (vector unsigned char, vector signed char);
6386 vector unsigned char vec_max (vector unsigned char,
6387 vector unsigned char);
6388 vector signed char vec_max (vector signed char, vector signed char);
6389 vector unsigned short vec_max (vector signed short,
6390 vector unsigned short);
6391 vector unsigned short vec_max (vector unsigned short,
6392 vector signed short);
6393 vector unsigned short vec_max (vector unsigned short,
6394 vector unsigned short);
6395 vector signed short vec_max (vector signed short, vector signed short);
6396 vector unsigned int vec_max (vector signed int, vector unsigned int);
6397 vector unsigned int vec_max (vector unsigned int, vector signed int);
6398 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
6399 vector signed int vec_max (vector signed int, vector signed int);
6400 vector float vec_max (vector float, vector float);
6402 vector signed char vec_mergeh (vector signed char, vector signed char);
6403 vector unsigned char vec_mergeh (vector unsigned char,
6404 vector unsigned char);
6405 vector signed short vec_mergeh (vector signed short,
6406 vector signed short);
6407 vector unsigned short vec_mergeh (vector unsigned short,
6408 vector unsigned short);
6409 vector float vec_mergeh (vector float, vector float);
6410 vector signed int vec_mergeh (vector signed int, vector signed int);
6411 vector unsigned int vec_mergeh (vector unsigned int,
6412 vector unsigned int);
6414 vector signed char vec_mergel (vector signed char, vector signed char);
6415 vector unsigned char vec_mergel (vector unsigned char,
6416 vector unsigned char);
6417 vector signed short vec_mergel (vector signed short,
6418 vector signed short);
6419 vector unsigned short vec_mergel (vector unsigned short,
6420 vector unsigned short);
6421 vector float vec_mergel (vector float, vector float);
6422 vector signed int vec_mergel (vector signed int, vector signed int);
6423 vector unsigned int vec_mergel (vector unsigned int,
6424 vector unsigned int);
6426 vector unsigned short vec_mfvscr (void);
6428 vector unsigned char vec_min (vector signed char, vector unsigned char);
6430 vector unsigned char vec_min (vector unsigned char, vector signed char);
6432 vector unsigned char vec_min (vector unsigned char,
6433 vector unsigned char);
6434 vector signed char vec_min (vector signed char, vector signed char);
6435 vector unsigned short vec_min (vector signed short,
6436 vector unsigned short);
6437 vector unsigned short vec_min (vector unsigned short,
6438 vector signed short);
6439 vector unsigned short vec_min (vector unsigned short,
6440 vector unsigned short);
6441 vector signed short vec_min (vector signed short, vector signed short);
6442 vector unsigned int vec_min (vector signed int, vector unsigned int);
6443 vector unsigned int vec_min (vector unsigned int, vector signed int);
6444 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
6445 vector signed int vec_min (vector signed int, vector signed int);
6446 vector float vec_min (vector float, vector float);
6448 vector signed short vec_mladd (vector signed short, vector signed short,
6449 vector signed short);
6450 vector signed short vec_mladd (vector signed short,
6451 vector unsigned short,
6452 vector unsigned short);
6453 vector signed short vec_mladd (vector unsigned short,
6454 vector signed short,
6455 vector signed short);
6456 vector unsigned short vec_mladd (vector unsigned short,
6457 vector unsigned short,
6458 vector unsigned short);
6460 vector signed short vec_mradds (vector signed short,
6461 vector signed short,
6462 vector signed short);
6464 vector unsigned int vec_msum (vector unsigned char,
6465 vector unsigned char,
6466 vector unsigned int);
6467 vector signed int vec_msum (vector signed char, vector unsigned char,
6469 vector unsigned int vec_msum (vector unsigned short,
6470 vector unsigned short,
6471 vector unsigned int);
6472 vector signed int vec_msum (vector signed short, vector signed short,
6475 vector unsigned int vec_msums (vector unsigned short,
6476 vector unsigned short,
6477 vector unsigned int);
6478 vector signed int vec_msums (vector signed short, vector signed short,
6481 void vec_mtvscr (vector signed int);
6482 void vec_mtvscr (vector unsigned int);
6483 void vec_mtvscr (vector signed short);
6484 void vec_mtvscr (vector unsigned short);
6485 void vec_mtvscr (vector signed char);
6486 void vec_mtvscr (vector unsigned char);
6488 vector unsigned short vec_mule (vector unsigned char,
6489 vector unsigned char);
6490 vector signed short vec_mule (vector signed char, vector signed char);
6491 vector unsigned int vec_mule (vector unsigned short,
6492 vector unsigned short);
6493 vector signed int vec_mule (vector signed short, vector signed short);
6495 vector unsigned short vec_mulo (vector unsigned char,
6496 vector unsigned char);
6497 vector signed short vec_mulo (vector signed char, vector signed char);
6498 vector unsigned int vec_mulo (vector unsigned short,
6499 vector unsigned short);
6500 vector signed int vec_mulo (vector signed short, vector signed short);
6502 vector float vec_nmsub (vector float, vector float, vector float);
6504 vector float vec_nor (vector float, vector float);
6505 vector signed int vec_nor (vector signed int, vector signed int);
6506 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
6507 vector signed short vec_nor (vector signed short, vector signed short);
6508 vector unsigned short vec_nor (vector unsigned short,
6509 vector unsigned short);
6510 vector signed char vec_nor (vector signed char, vector signed char);
6511 vector unsigned char vec_nor (vector unsigned char,
6512 vector unsigned char);
6514 vector float vec_or (vector float, vector float);
6515 vector float vec_or (vector float, vector signed int);
6516 vector float vec_or (vector signed int, vector float);
6517 vector signed int vec_or (vector signed int, vector signed int);
6518 vector unsigned int vec_or (vector signed int, vector unsigned int);
6519 vector unsigned int vec_or (vector unsigned int, vector signed int);
6520 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
6521 vector signed short vec_or (vector signed short, vector signed short);
6522 vector unsigned short vec_or (vector signed short,
6523 vector unsigned short);
6524 vector unsigned short vec_or (vector unsigned short,
6525 vector signed short);
6526 vector unsigned short vec_or (vector unsigned short,
6527 vector unsigned short);
6528 vector signed char vec_or (vector signed char, vector signed char);
6529 vector unsigned char vec_or (vector signed char, vector unsigned char);
6530 vector unsigned char vec_or (vector unsigned char, vector signed char);
6531 vector unsigned char vec_or (vector unsigned char,
6532 vector unsigned char);
6534 vector signed char vec_pack (vector signed short, vector signed short);
6535 vector unsigned char vec_pack (vector unsigned short,
6536 vector unsigned short);
6537 vector signed short vec_pack (vector signed int, vector signed int);
6538 vector unsigned short vec_pack (vector unsigned int,
6539 vector unsigned int);
6541 vector signed short vec_packpx (vector unsigned int,
6542 vector unsigned int);
6544 vector unsigned char vec_packs (vector unsigned short,
6545 vector unsigned short);
6546 vector signed char vec_packs (vector signed short, vector signed short);
6548 vector unsigned short vec_packs (vector unsigned int,
6549 vector unsigned int);
6550 vector signed short vec_packs (vector signed int, vector signed int);
6552 vector unsigned char vec_packsu (vector unsigned short,
6553 vector unsigned short);
6554 vector unsigned char vec_packsu (vector signed short,
6555 vector signed short);
6556 vector unsigned short vec_packsu (vector unsigned int,
6557 vector unsigned int);
6558 vector unsigned short vec_packsu (vector signed int, vector signed int);
6560 vector float vec_perm (vector float, vector float,
6561 vector unsigned char);
6562 vector signed int vec_perm (vector signed int, vector signed int,
6563 vector unsigned char);
6564 vector unsigned int vec_perm (vector unsigned int, vector unsigned int,
6565 vector unsigned char);
6566 vector signed short vec_perm (vector signed short, vector signed short,
6567 vector unsigned char);
6568 vector unsigned short vec_perm (vector unsigned short,
6569 vector unsigned short,
6570 vector unsigned char);
6571 vector signed char vec_perm (vector signed char, vector signed char,
6572 vector unsigned char);
6573 vector unsigned char vec_perm (vector unsigned char,
6574 vector unsigned char,
6575 vector unsigned char);
6577 vector float vec_re (vector float);
6579 vector signed char vec_rl (vector signed char, vector unsigned char);
6580 vector unsigned char vec_rl (vector unsigned char,
6581 vector unsigned char);
6582 vector signed short vec_rl (vector signed short, vector unsigned short);
6584 vector unsigned short vec_rl (vector unsigned short,
6585 vector unsigned short);
6586 vector signed int vec_rl (vector signed int, vector unsigned int);
6587 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
6589 vector float vec_round (vector float);
6591 vector float vec_rsqrte (vector float);
6593 vector float vec_sel (vector float, vector float, vector signed int);
6594 vector float vec_sel (vector float, vector float, vector unsigned int);
6595 vector signed int vec_sel (vector signed int, vector signed int,
6597 vector signed int vec_sel (vector signed int, vector signed int,
6598 vector unsigned int);
6599 vector unsigned int vec_sel (vector unsigned int, vector unsigned int,
6601 vector unsigned int vec_sel (vector unsigned int, vector unsigned int,
6602 vector unsigned int);
6603 vector signed short vec_sel (vector signed short, vector signed short,
6604 vector signed short);
6605 vector signed short vec_sel (vector signed short, vector signed short,
6606 vector unsigned short);
6607 vector unsigned short vec_sel (vector unsigned short,
6608 vector unsigned short,
6609 vector signed short);
6610 vector unsigned short vec_sel (vector unsigned short,
6611 vector unsigned short,
6612 vector unsigned short);
6613 vector signed char vec_sel (vector signed char, vector signed char,
6614 vector signed char);
6615 vector signed char vec_sel (vector signed char, vector signed char,
6616 vector unsigned char);
6617 vector unsigned char vec_sel (vector unsigned char,
6618 vector unsigned char,
6619 vector signed char);
6620 vector unsigned char vec_sel (vector unsigned char,
6621 vector unsigned char,
6622 vector unsigned char);
6624 vector signed char vec_sl (vector signed char, vector unsigned char);
6625 vector unsigned char vec_sl (vector unsigned char,
6626 vector unsigned char);
6627 vector signed short vec_sl (vector signed short, vector unsigned short);
6629 vector unsigned short vec_sl (vector unsigned short,
6630 vector unsigned short);
6631 vector signed int vec_sl (vector signed int, vector unsigned int);
6632 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
6634 vector float vec_sld (vector float, vector float, const char);
6635 vector signed int vec_sld (vector signed int, vector signed int,
6637 vector unsigned int vec_sld (vector unsigned int, vector unsigned int,
6639 vector signed short vec_sld (vector signed short, vector signed short,
6641 vector unsigned short vec_sld (vector unsigned short,
6642 vector unsigned short, const char);
6643 vector signed char vec_sld (vector signed char, vector signed char,
6645 vector unsigned char vec_sld (vector unsigned char,
6646 vector unsigned char,
6649 vector signed int vec_sll (vector signed int, vector unsigned int);
6650 vector signed int vec_sll (vector signed int, vector unsigned short);
6651 vector signed int vec_sll (vector signed int, vector unsigned char);
6652 vector unsigned int vec_sll (vector unsigned int, vector unsigned int);
6653 vector unsigned int vec_sll (vector unsigned int,
6654 vector unsigned short);
6655 vector unsigned int vec_sll (vector unsigned int, vector unsigned char);
6657 vector signed short vec_sll (vector signed short, vector unsigned int);
6658 vector signed short vec_sll (vector signed short,
6659 vector unsigned short);
6660 vector signed short vec_sll (vector signed short, vector unsigned char);
6662 vector unsigned short vec_sll (vector unsigned short,
6663 vector unsigned int);
6664 vector unsigned short vec_sll (vector unsigned short,
6665 vector unsigned short);
6666 vector unsigned short vec_sll (vector unsigned short,
6667 vector unsigned char);
6668 vector signed char vec_sll (vector signed char, vector unsigned int);
6669 vector signed char vec_sll (vector signed char, vector unsigned short);
6670 vector signed char vec_sll (vector signed char, vector unsigned char);
6671 vector unsigned char vec_sll (vector unsigned char,
6672 vector unsigned int);
6673 vector unsigned char vec_sll (vector unsigned char,
6674 vector unsigned short);
6675 vector unsigned char vec_sll (vector unsigned char,
6676 vector unsigned char);
6678 vector float vec_slo (vector float, vector signed char);
6679 vector float vec_slo (vector float, vector unsigned char);
6680 vector signed int vec_slo (vector signed int, vector signed char);
6681 vector signed int vec_slo (vector signed int, vector unsigned char);
6682 vector unsigned int vec_slo (vector unsigned int, vector signed char);
6683 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
6685 vector signed short vec_slo (vector signed short, vector signed char);
6686 vector signed short vec_slo (vector signed short, vector unsigned char);
6688 vector unsigned short vec_slo (vector unsigned short,
6689 vector signed char);
6690 vector unsigned short vec_slo (vector unsigned short,
6691 vector unsigned char);
6692 vector signed char vec_slo (vector signed char, vector signed char);
6693 vector signed char vec_slo (vector signed char, vector unsigned char);
6694 vector unsigned char vec_slo (vector unsigned char, vector signed char);
6696 vector unsigned char vec_slo (vector unsigned char,
6697 vector unsigned char);
6699 vector signed char vec_splat (vector signed char, const char);
6700 vector unsigned char vec_splat (vector unsigned char, const char);
6701 vector signed short vec_splat (vector signed short, const char);
6702 vector unsigned short vec_splat (vector unsigned short, const char);
6703 vector float vec_splat (vector float, const char);
6704 vector signed int vec_splat (vector signed int, const char);
6705 vector unsigned int vec_splat (vector unsigned int, const char);
6707 vector signed char vec_splat_s8 (const char);
6709 vector signed short vec_splat_s16 (const char);
6711 vector signed int vec_splat_s32 (const char);
6713 vector unsigned char vec_splat_u8 (const char);
6715 vector unsigned short vec_splat_u16 (const char);
6717 vector unsigned int vec_splat_u32 (const char);
6719 vector signed char vec_sr (vector signed char, vector unsigned char);
6720 vector unsigned char vec_sr (vector unsigned char,
6721 vector unsigned char);
6722 vector signed short vec_sr (vector signed short, vector unsigned short);
6724 vector unsigned short vec_sr (vector unsigned short,
6725 vector unsigned short);
6726 vector signed int vec_sr (vector signed int, vector unsigned int);
6727 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
6729 vector signed char vec_sra (vector signed char, vector unsigned char);
6730 vector unsigned char vec_sra (vector unsigned char,
6731 vector unsigned char);
6732 vector signed short vec_sra (vector signed short,
6733 vector unsigned short);
6734 vector unsigned short vec_sra (vector unsigned short,
6735 vector unsigned short);
6736 vector signed int vec_sra (vector signed int, vector unsigned int);
6737 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
6739 vector signed int vec_srl (vector signed int, vector unsigned int);
6740 vector signed int vec_srl (vector signed int, vector unsigned short);
6741 vector signed int vec_srl (vector signed int, vector unsigned char);
6742 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
6743 vector unsigned int vec_srl (vector unsigned int,
6744 vector unsigned short);
6745 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
6747 vector signed short vec_srl (vector signed short, vector unsigned int);
6748 vector signed short vec_srl (vector signed short,
6749 vector unsigned short);
6750 vector signed short vec_srl (vector signed short, vector unsigned char);
6752 vector unsigned short vec_srl (vector unsigned short,
6753 vector unsigned int);
6754 vector unsigned short vec_srl (vector unsigned short,
6755 vector unsigned short);
6756 vector unsigned short vec_srl (vector unsigned short,
6757 vector unsigned char);
6758 vector signed char vec_srl (vector signed char, vector unsigned int);
6759 vector signed char vec_srl (vector signed char, vector unsigned short);
6760 vector signed char vec_srl (vector signed char, vector unsigned char);
6761 vector unsigned char vec_srl (vector unsigned char,
6762 vector unsigned int);
6763 vector unsigned char vec_srl (vector unsigned char,
6764 vector unsigned short);
6765 vector unsigned char vec_srl (vector unsigned char,
6766 vector unsigned char);
6768 vector float vec_sro (vector float, vector signed char);
6769 vector float vec_sro (vector float, vector unsigned char);
6770 vector signed int vec_sro (vector signed int, vector signed char);
6771 vector signed int vec_sro (vector signed int, vector unsigned char);
6772 vector unsigned int vec_sro (vector unsigned int, vector signed char);
6773 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
6775 vector signed short vec_sro (vector signed short, vector signed char);
6776 vector signed short vec_sro (vector signed short, vector unsigned char);
6778 vector unsigned short vec_sro (vector unsigned short,
6779 vector signed char);
6780 vector unsigned short vec_sro (vector unsigned short,
6781 vector unsigned char);
6782 vector signed char vec_sro (vector signed char, vector signed char);
6783 vector signed char vec_sro (vector signed char, vector unsigned char);
6784 vector unsigned char vec_sro (vector unsigned char, vector signed char);
6786 vector unsigned char vec_sro (vector unsigned char,
6787 vector unsigned char);
6789 void vec_st (vector float, int, float *);
6790 void vec_st (vector float, int, vector float *);
6791 void vec_st (vector signed int, int, int *);
6792 void vec_st (vector signed int, int, unsigned int *);
6793 void vec_st (vector unsigned int, int, unsigned int *);
6794 void vec_st (vector unsigned int, int, vector unsigned int *);
6795 void vec_st (vector signed short, int, short *);
6796 void vec_st (vector signed short, int, vector unsigned short *);
6797 void vec_st (vector signed short, int, vector signed short *);
6798 void vec_st (vector unsigned short, int, unsigned short *);
6799 void vec_st (vector unsigned short, int, vector unsigned short *);
6800 void vec_st (vector signed char, int, signed char *);
6801 void vec_st (vector signed char, int, unsigned char *);
6802 void vec_st (vector signed char, int, vector signed char *);
6803 void vec_st (vector unsigned char, int, unsigned char *);
6804 void vec_st (vector unsigned char, int, vector unsigned char *);
6806 void vec_ste (vector signed char, int, unsigned char *);
6807 void vec_ste (vector signed char, int, signed char *);
6808 void vec_ste (vector unsigned char, int, unsigned char *);
6809 void vec_ste (vector signed short, int, short *);
6810 void vec_ste (vector signed short, int, unsigned short *);
6811 void vec_ste (vector unsigned short, int, void *);
6812 void vec_ste (vector signed int, int, unsigned int *);
6813 void vec_ste (vector signed int, int, int *);
6814 void vec_ste (vector unsigned int, int, unsigned int *);
6815 void vec_ste (vector float, int, float *);
6817 void vec_stl (vector float, int, vector float *);
6818 void vec_stl (vector float, int, float *);
6819 void vec_stl (vector signed int, int, vector signed int *);
6820 void vec_stl (vector signed int, int, int *);
6821 void vec_stl (vector signed int, int, unsigned int *);
6822 void vec_stl (vector unsigned int, int, vector unsigned int *);
6823 void vec_stl (vector unsigned int, int, unsigned int *);
6824 void vec_stl (vector signed short, int, short *);
6825 void vec_stl (vector signed short, int, unsigned short *);
6826 void vec_stl (vector signed short, int, vector signed short *);
6827 void vec_stl (vector unsigned short, int, unsigned short *);
6828 void vec_stl (vector unsigned short, int, vector signed short *);
6829 void vec_stl (vector signed char, int, signed char *);
6830 void vec_stl (vector signed char, int, unsigned char *);
6831 void vec_stl (vector signed char, int, vector signed char *);
6832 void vec_stl (vector unsigned char, int, unsigned char *);
6833 void vec_stl (vector unsigned char, int, vector unsigned char *);
6835 vector signed char vec_sub (vector signed char, vector signed char);
6836 vector unsigned char vec_sub (vector signed char, vector unsigned char);
6838 vector unsigned char vec_sub (vector unsigned char, vector signed char);
6840 vector unsigned char vec_sub (vector unsigned char,
6841 vector unsigned char);
6842 vector signed short vec_sub (vector signed short, vector signed short);
6843 vector unsigned short vec_sub (vector signed short,
6844 vector unsigned short);
6845 vector unsigned short vec_sub (vector unsigned short,
6846 vector signed short);
6847 vector unsigned short vec_sub (vector unsigned short,
6848 vector unsigned short);
6849 vector signed int vec_sub (vector signed int, vector signed int);
6850 vector unsigned int vec_sub (vector signed int, vector unsigned int);
6851 vector unsigned int vec_sub (vector unsigned int, vector signed int);
6852 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
6853 vector float vec_sub (vector float, vector float);
6855 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
6857 vector unsigned char vec_subs (vector signed char,
6858 vector unsigned char);
6859 vector unsigned char vec_subs (vector unsigned char,
6860 vector signed char);
6861 vector unsigned char vec_subs (vector unsigned char,
6862 vector unsigned char);
6863 vector signed char vec_subs (vector signed char, vector signed char);
6864 vector unsigned short vec_subs (vector signed short,
6865 vector unsigned short);
6866 vector unsigned short vec_subs (vector unsigned short,
6867 vector signed short);
6868 vector unsigned short vec_subs (vector unsigned short,
6869 vector unsigned short);
6870 vector signed short vec_subs (vector signed short, vector signed short);
6872 vector unsigned int vec_subs (vector signed int, vector unsigned int);
6873 vector unsigned int vec_subs (vector unsigned int, vector signed int);
6874 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
6876 vector signed int vec_subs (vector signed int, vector signed int);
6878 vector unsigned int vec_sum4s (vector unsigned char,
6879 vector unsigned int);
6880 vector signed int vec_sum4s (vector signed char, vector signed int);
6881 vector signed int vec_sum4s (vector signed short, vector signed int);
6883 vector signed int vec_sum2s (vector signed int, vector signed int);
6885 vector signed int vec_sums (vector signed int, vector signed int);
6887 vector float vec_trunc (vector float);
6889 vector signed short vec_unpackh (vector signed char);
6890 vector unsigned int vec_unpackh (vector signed short);
6891 vector signed int vec_unpackh (vector signed short);
6893 vector signed short vec_unpackl (vector signed char);
6894 vector unsigned int vec_unpackl (vector signed short);
6895 vector signed int vec_unpackl (vector signed short);
6897 vector float vec_xor (vector float, vector float);
6898 vector float vec_xor (vector float, vector signed int);
6899 vector float vec_xor (vector signed int, vector float);
6900 vector signed int vec_xor (vector signed int, vector signed int);
6901 vector unsigned int vec_xor (vector signed int, vector unsigned int);
6902 vector unsigned int vec_xor (vector unsigned int, vector signed int);
6903 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
6904 vector signed short vec_xor (vector signed short, vector signed short);
6905 vector unsigned short vec_xor (vector signed short,
6906 vector unsigned short);
6907 vector unsigned short vec_xor (vector unsigned short,
6908 vector signed short);
6909 vector unsigned short vec_xor (vector unsigned short,
6910 vector unsigned short);
6911 vector signed char vec_xor (vector signed char, vector signed char);
6912 vector unsigned char vec_xor (vector signed char, vector unsigned char);
6914 vector unsigned char vec_xor (vector unsigned char, vector signed char);
6916 vector unsigned char vec_xor (vector unsigned char,
6917 vector unsigned char);
6919 vector signed int vec_all_eq (vector signed char, vector unsigned char);
6921 vector signed int vec_all_eq (vector signed char, vector signed char);
6922 vector signed int vec_all_eq (vector unsigned char, vector signed char);
6924 vector signed int vec_all_eq (vector unsigned char,
6925 vector unsigned char);
6926 vector signed int vec_all_eq (vector signed short,
6927 vector unsigned short);
6928 vector signed int vec_all_eq (vector signed short, vector signed short);
6930 vector signed int vec_all_eq (vector unsigned short,
6931 vector signed short);
6932 vector signed int vec_all_eq (vector unsigned short,
6933 vector unsigned short);
6934 vector signed int vec_all_eq (vector signed int, vector unsigned int);
6935 vector signed int vec_all_eq (vector signed int, vector signed int);
6936 vector signed int vec_all_eq (vector unsigned int, vector signed int);
6937 vector signed int vec_all_eq (vector unsigned int, vector unsigned int);
6939 vector signed int vec_all_eq (vector float, vector float);
6941 vector signed int vec_all_ge (vector signed char, vector unsigned char);
6943 vector signed int vec_all_ge (vector unsigned char, vector signed char);
6945 vector signed int vec_all_ge (vector unsigned char,
6946 vector unsigned char);
6947 vector signed int vec_all_ge (vector signed char, vector signed char);
6948 vector signed int vec_all_ge (vector signed short,
6949 vector unsigned short);
6950 vector signed int vec_all_ge (vector unsigned short,
6951 vector signed short);
6952 vector signed int vec_all_ge (vector unsigned short,
6953 vector unsigned short);
6954 vector signed int vec_all_ge (vector signed short, vector signed short);
6956 vector signed int vec_all_ge (vector signed int, vector unsigned int);
6957 vector signed int vec_all_ge (vector unsigned int, vector signed int);
6958 vector signed int vec_all_ge (vector unsigned int, vector unsigned int);
6960 vector signed int vec_all_ge (vector signed int, vector signed int);
6961 vector signed int vec_all_ge (vector float, vector float);
6963 vector signed int vec_all_gt (vector signed char, vector unsigned char);
6965 vector signed int vec_all_gt (vector unsigned char, vector signed char);
6967 vector signed int vec_all_gt (vector unsigned char,
6968 vector unsigned char);
6969 vector signed int vec_all_gt (vector signed char, vector signed char);
6970 vector signed int vec_all_gt (vector signed short,
6971 vector unsigned short);
6972 vector signed int vec_all_gt (vector unsigned short,
6973 vector signed short);
6974 vector signed int vec_all_gt (vector unsigned short,
6975 vector unsigned short);
6976 vector signed int vec_all_gt (vector signed short, vector signed short);
6978 vector signed int vec_all_gt (vector signed int, vector unsigned int);
6979 vector signed int vec_all_gt (vector unsigned int, vector signed int);
6980 vector signed int vec_all_gt (vector unsigned int, vector unsigned int);
6982 vector signed int vec_all_gt (vector signed int, vector signed int);
6983 vector signed int vec_all_gt (vector float, vector float);
6985 vector signed int vec_all_in (vector float, vector float);
6987 vector signed int vec_all_le (vector signed char, vector unsigned char);
6989 vector signed int vec_all_le (vector unsigned char, vector signed char);
6991 vector signed int vec_all_le (vector unsigned char,
6992 vector unsigned char);
6993 vector signed int vec_all_le (vector signed char, vector signed char);
6994 vector signed int vec_all_le (vector signed short,
6995 vector unsigned short);
6996 vector signed int vec_all_le (vector unsigned short,
6997 vector signed short);
6998 vector signed int vec_all_le (vector unsigned short,
6999 vector unsigned short);
7000 vector signed int vec_all_le (vector signed short, vector signed short);
7002 vector signed int vec_all_le (vector signed int, vector unsigned int);
7003 vector signed int vec_all_le (vector unsigned int, vector signed int);
7004 vector signed int vec_all_le (vector unsigned int, vector unsigned int);
7006 vector signed int vec_all_le (vector signed int, vector signed int);
7007 vector signed int vec_all_le (vector float, vector float);
7009 vector signed int vec_all_lt (vector signed char, vector unsigned char);
7011 vector signed int vec_all_lt (vector unsigned char, vector signed char);
7013 vector signed int vec_all_lt (vector unsigned char,
7014 vector unsigned char);
7015 vector signed int vec_all_lt (vector signed char, vector signed char);
7016 vector signed int vec_all_lt (vector signed short,
7017 vector unsigned short);
7018 vector signed int vec_all_lt (vector unsigned short,
7019 vector signed short);
7020 vector signed int vec_all_lt (vector unsigned short,
7021 vector unsigned short);
7022 vector signed int vec_all_lt (vector signed short, vector signed short);
7024 vector signed int vec_all_lt (vector signed int, vector unsigned int);
7025 vector signed int vec_all_lt (vector unsigned int, vector signed int);
7026 vector signed int vec_all_lt (vector unsigned int, vector unsigned int);
7028 vector signed int vec_all_lt (vector signed int, vector signed int);
7029 vector signed int vec_all_lt (vector float, vector float);
7031 vector signed int vec_all_nan (vector float);
7033 vector signed int vec_all_ne (vector signed char, vector unsigned char);
7035 vector signed int vec_all_ne (vector signed char, vector signed char);
7036 vector signed int vec_all_ne (vector unsigned char, vector signed char);
7038 vector signed int vec_all_ne (vector unsigned char,
7039 vector unsigned char);
7040 vector signed int vec_all_ne (vector signed short,
7041 vector unsigned short);
7042 vector signed int vec_all_ne (vector signed short, vector signed short);
7044 vector signed int vec_all_ne (vector unsigned short,
7045 vector signed short);
7046 vector signed int vec_all_ne (vector unsigned short,
7047 vector unsigned short);
7048 vector signed int vec_all_ne (vector signed int, vector unsigned int);
7049 vector signed int vec_all_ne (vector signed int, vector signed int);
7050 vector signed int vec_all_ne (vector unsigned int, vector signed int);
7051 vector signed int vec_all_ne (vector unsigned int, vector unsigned int);
7053 vector signed int vec_all_ne (vector float, vector float);
7055 vector signed int vec_all_nge (vector float, vector float);
7057 vector signed int vec_all_ngt (vector float, vector float);
7059 vector signed int vec_all_nle (vector float, vector float);
7061 vector signed int vec_all_nlt (vector float, vector float);
7063 vector signed int vec_all_numeric (vector float);
7065 vector signed int vec_any_eq (vector signed char, vector unsigned char);
7067 vector signed int vec_any_eq (vector signed char, vector signed char);
7068 vector signed int vec_any_eq (vector unsigned char, vector signed char);
7070 vector signed int vec_any_eq (vector unsigned char,
7071 vector unsigned char);
7072 vector signed int vec_any_eq (vector signed short,
7073 vector unsigned short);
7074 vector signed int vec_any_eq (vector signed short, vector signed short);
7076 vector signed int vec_any_eq (vector unsigned short,
7077 vector signed short);
7078 vector signed int vec_any_eq (vector unsigned short,
7079 vector unsigned short);
7080 vector signed int vec_any_eq (vector signed int, vector unsigned int);
7081 vector signed int vec_any_eq (vector signed int, vector signed int);
7082 vector signed int vec_any_eq (vector unsigned int, vector signed int);
7083 vector signed int vec_any_eq (vector unsigned int, vector unsigned int);
7085 vector signed int vec_any_eq (vector float, vector float);
7087 vector signed int vec_any_ge (vector signed char, vector unsigned char);
7089 vector signed int vec_any_ge (vector unsigned char, vector signed char);
7091 vector signed int vec_any_ge (vector unsigned char,
7092 vector unsigned char);
7093 vector signed int vec_any_ge (vector signed char, vector signed char);
7094 vector signed int vec_any_ge (vector signed short,
7095 vector unsigned short);
7096 vector signed int vec_any_ge (vector unsigned short,
7097 vector signed short);
7098 vector signed int vec_any_ge (vector unsigned short,
7099 vector unsigned short);
7100 vector signed int vec_any_ge (vector signed short, vector signed short);
7102 vector signed int vec_any_ge (vector signed int, vector unsigned int);
7103 vector signed int vec_any_ge (vector unsigned int, vector signed int);
7104 vector signed int vec_any_ge (vector unsigned int, vector unsigned int);
7106 vector signed int vec_any_ge (vector signed int, vector signed int);
7107 vector signed int vec_any_ge (vector float, vector float);
7109 vector signed int vec_any_gt (vector signed char, vector unsigned char);
7111 vector signed int vec_any_gt (vector unsigned char, vector signed char);
7113 vector signed int vec_any_gt (vector unsigned char,
7114 vector unsigned char);
7115 vector signed int vec_any_gt (vector signed char, vector signed char);
7116 vector signed int vec_any_gt (vector signed short,
7117 vector unsigned short);
7118 vector signed int vec_any_gt (vector unsigned short,
7119 vector signed short);
7120 vector signed int vec_any_gt (vector unsigned short,
7121 vector unsigned short);
7122 vector signed int vec_any_gt (vector signed short, vector signed short);
7124 vector signed int vec_any_gt (vector signed int, vector unsigned int);
7125 vector signed int vec_any_gt (vector unsigned int, vector signed int);
7126 vector signed int vec_any_gt (vector unsigned int, vector unsigned int);
7128 vector signed int vec_any_gt (vector signed int, vector signed int);
7129 vector signed int vec_any_gt (vector float, vector float);
7131 vector signed int vec_any_le (vector signed char, vector unsigned char);
7133 vector signed int vec_any_le (vector unsigned char, vector signed char);
7135 vector signed int vec_any_le (vector unsigned char,
7136 vector unsigned char);
7137 vector signed int vec_any_le (vector signed char, vector signed char);
7138 vector signed int vec_any_le (vector signed short,
7139 vector unsigned short);
7140 vector signed int vec_any_le (vector unsigned short,
7141 vector signed short);
7142 vector signed int vec_any_le (vector unsigned short,
7143 vector unsigned short);
7144 vector signed int vec_any_le (vector signed short, vector signed short);
7146 vector signed int vec_any_le (vector signed int, vector unsigned int);
7147 vector signed int vec_any_le (vector unsigned int, vector signed int);
7148 vector signed int vec_any_le (vector unsigned int, vector unsigned int);
7150 vector signed int vec_any_le (vector signed int, vector signed int);
7151 vector signed int vec_any_le (vector float, vector float);
7153 vector signed int vec_any_lt (vector signed char, vector unsigned char);
7155 vector signed int vec_any_lt (vector unsigned char, vector signed char);
7157 vector signed int vec_any_lt (vector unsigned char,
7158 vector unsigned char);
7159 vector signed int vec_any_lt (vector signed char, vector signed char);
7160 vector signed int vec_any_lt (vector signed short,
7161 vector unsigned short);
7162 vector signed int vec_any_lt (vector unsigned short,
7163 vector signed short);
7164 vector signed int vec_any_lt (vector unsigned short,
7165 vector unsigned short);
7166 vector signed int vec_any_lt (vector signed short, vector signed short);
7168 vector signed int vec_any_lt (vector signed int, vector unsigned int);
7169 vector signed int vec_any_lt (vector unsigned int, vector signed int);
7170 vector signed int vec_any_lt (vector unsigned int, vector unsigned int);
7172 vector signed int vec_any_lt (vector signed int, vector signed int);
7173 vector signed int vec_any_lt (vector float, vector float);
7175 vector signed int vec_any_nan (vector float);
7177 vector signed int vec_any_ne (vector signed char, vector unsigned char);
7179 vector signed int vec_any_ne (vector signed char, vector signed char);
7180 vector signed int vec_any_ne (vector unsigned char, vector signed char);
7182 vector signed int vec_any_ne (vector unsigned char,
7183 vector unsigned char);
7184 vector signed int vec_any_ne (vector signed short,
7185 vector unsigned short);
7186 vector signed int vec_any_ne (vector signed short, vector signed short);
7188 vector signed int vec_any_ne (vector unsigned short,
7189 vector signed short);
7190 vector signed int vec_any_ne (vector unsigned short,
7191 vector unsigned short);
7192 vector signed int vec_any_ne (vector signed int, vector unsigned int);
7193 vector signed int vec_any_ne (vector signed int, vector signed int);
7194 vector signed int vec_any_ne (vector unsigned int, vector signed int);
7195 vector signed int vec_any_ne (vector unsigned int, vector unsigned int);
7197 vector signed int vec_any_ne (vector float, vector float);
7199 vector signed int vec_any_nge (vector float, vector float);
7201 vector signed int vec_any_ngt (vector float, vector float);
7203 vector signed int vec_any_nle (vector float, vector float);
7205 vector signed int vec_any_nlt (vector float, vector float);
7207 vector signed int vec_any_numeric (vector float);
7209 vector signed int vec_any_out (vector float, vector float);
7213 @section Pragmas Accepted by GCC
7217 GCC supports several types of pragmas, primarily in order to compile
7218 code originally written for other compilers. Note that in general
7219 we do not recommend the use of pragmas; @xref{Function Attributes},
7220 for further explanation.
7224 * RS/6000 and PowerPC Pragmas::
7226 * Symbol-Renaming Pragmas::
7230 @subsection ARM Pragmas
7232 The ARM target defines pragmas for controlling the default addition of
7233 @code{long_call} and @code{short_call} attributes to functions.
7234 @xref{Function Attributes}, for information about the effects of these
7239 @cindex pragma, long_calls
7240 Set all subsequent functions to have the @code{long_call} attribute.
7243 @cindex pragma, no_long_calls
7244 Set all subsequent functions to have the @code{short_call} attribute.
7246 @item long_calls_off
7247 @cindex pragma, long_calls_off
7248 Do not affect the @code{long_call} or @code{short_call} attributes of
7249 subsequent functions.
7252 @node RS/6000 and PowerPC Pragmas
7253 @subsection RS/6000 and PowerPC Pragmas
7255 The RS/6000 and PowerPC targets define one pragma for controlling
7256 whether or not the @code{longcall} attribute is added to function
7257 declarations by default. This pragma overrides the @option{-mlongcall}
7258 option, but not the @code{longcall} and @code{shortcall} attributes.
7259 @xref{RS/6000 and PowerPC Options}, for more information about when long
7260 calls are and are not necessary.
7264 @cindex pragma, longcall
7265 Apply the @code{longcall} attribute to all subsequent function
7269 Do not apply the @code{longcall} attribute to subsequent function
7273 @c Describe c4x pragmas here.
7274 @c Describe h8300 pragmas here.
7275 @c Describe sh pragmas here.
7276 @c Describe v850 pragmas here.
7278 @node Darwin Pragmas
7279 @subsection Darwin Pragmas
7281 The following pragmas are available for all architectures running the
7282 Darwin operating system. These are useful for compatibility with other
7286 @item mark @var{tokens}@dots{}
7287 @cindex pragma, mark
7288 This pragma is accepted, but has no effect.
7290 @item options align=@var{alignment}
7291 @cindex pragma, options align
7292 This pragma sets the alignment of fields in structures. The values of
7293 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
7294 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
7295 properly; to restore the previous setting, use @code{reset} for the
7298 @item segment @var{tokens}@dots{}
7299 @cindex pragma, segment
7300 This pragma is accepted, but has no effect.
7302 @item unused (@var{var} [, @var{var}]@dots{})
7303 @cindex pragma, unused
7304 This pragma declares variables to be possibly unused. GCC will not
7305 produce warnings for the listed variables. The effect is similar to
7306 that of the @code{unused} attribute, except that this pragma may appear
7307 anywhere within the variables' scopes.
7310 @node Symbol-Renaming Pragmas
7311 @subsection Symbol-Renaming Pragmas
7313 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
7314 supports two @code{#pragma} directives which change the name used in
7315 assembly for a given declaration. These pragmas are only available on
7316 platforms whose system headers need them. To get this effect on all
7317 platforms supported by GCC, use the asm labels extension (@pxref{Asm
7321 @item redefine_extname @var{oldname} @var{newname}
7322 @cindex pragma, redefine_extname
7324 This pragma gives the C function @var{oldname} the assembly symbol
7325 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
7326 will be defined if this pragma is available (currently only on
7329 @item extern_prefix @var{string}
7330 @cindex pragma, extern_prefix
7332 This pragma causes all subsequent external function and variable
7333 declarations to have @var{string} prepended to their assembly symbols.
7334 This effect may be terminated with another @code{extern_prefix} pragma
7335 whose argument is an empty string. The preprocessor macro
7336 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
7337 available (currently only on Tru64 UNIX).
7340 These pragmas and the asm labels extension interact in a complicated
7341 manner. Here are some corner cases you may want to be aware of.
7344 @item Both pragmas silently apply only to declarations with external
7345 linkage. Asm labels do not have this restriction.
7347 @item In C++, both pragmas silently apply only to declarations with
7348 ``C'' linkage. Again, asm labels do not have this restriction.
7350 @item If any of the three ways of changing the assembly name of a
7351 declaration is applied to a declaration whose assembly name has
7352 already been determined (either by a previous use of one of these
7353 features, or because the compiler needed the assembly name in order to
7354 generate code), and the new name is different, a warning issues and
7355 the name does not change.
7357 @item The @var{oldname} used by @code{#pragma redefine_extname} is
7358 always the C-language name.
7360 @item If @code{#pragma extern_prefix} is in effect, and a declaration
7361 occurs with an asm label attached, the prefix is silently ignored for
7364 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
7365 apply to the same declaration, whichever triggered first wins, and a
7366 warning issues if they contradict each other. (We would like to have
7367 @code{#pragma redefine_extname} always win, for consistency with asm
7368 labels, but if @code{#pragma extern_prefix} triggers first we have no
7369 way of knowing that that happened.)
7372 @node Unnamed Fields
7373 @section Unnamed struct/union fields within structs/unions.
7377 For compatibility with other compilers, GCC allows you to define
7378 a structure or union that contains, as fields, structures and unions
7379 without names. For example:
7392 In this example, the user would be able to access members of the unnamed
7393 union with code like @samp{foo.b}. Note that only unnamed structs and
7394 unions are allowed, you may not have, for example, an unnamed
7397 You must never create such structures that cause ambiguous field definitions.
7398 For example, this structure:
7409 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
7410 Such constructs are not supported and must be avoided. In the future,
7411 such constructs may be detected and treated as compilation errors.
7414 @section Thread-Local Storage
7415 @cindex Thread-Local Storage
7416 @cindex @acronym{TLS}
7419 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
7420 are allocated such that there is one instance of the variable per extant
7421 thread. The run-time model GCC uses to implement this originates
7422 in the IA-64 processor-specific ABI, but has since been migrated
7423 to other processors as well. It requires significant support from
7424 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
7425 system libraries (@file{libc.so} and @file{libpthread.so}), so it
7426 is not available everywhere.
7428 At the user level, the extension is visible with a new storage
7429 class keyword: @code{__thread}. For example:
7433 extern __thread struct state s;
7434 static __thread char *p;
7437 The @code{__thread} specifier may be used alone, with the @code{extern}
7438 or @code{static} specifiers, but with no other storage class specifier.
7439 When used with @code{extern} or @code{static}, @code{__thread} must appear
7440 immediately after the other storage class specifier.
7442 The @code{__thread} specifier may be applied to any global, file-scoped
7443 static, function-scoped static, or static data member of a class. It may
7444 not be applied to block-scoped automatic or non-static data member.
7446 When the address-of operator is applied to a thread-local variable, it is
7447 evaluated at run-time and returns the address of the current thread's
7448 instance of that variable. An address so obtained may be used by any
7449 thread. When a thread terminates, any pointers to thread-local variables
7450 in that thread become invalid.
7452 No static initialization may refer to the address of a thread-local variable.
7454 In C++, if an initializer is present for a thread-local variable, it must
7455 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
7458 See @uref{http://people.redhat.com/drepper/tls.pdf,
7459 ELF Handling For Thread-Local Storage} for a detailed explanation of
7460 the four thread-local storage addressing models, and how the run-time
7461 is expected to function.
7464 * C99 Thread-Local Edits::
7465 * C++98 Thread-Local Edits::
7468 @node C99 Thread-Local Edits
7469 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
7471 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
7472 that document the exact semantics of the language extension.
7476 @cite{5.1.2 Execution environments}
7478 Add new text after paragraph 1
7481 Within either execution environment, a @dfn{thread} is a flow of
7482 control within a program. It is implementation defined whether
7483 or not there may be more than one thread associated with a program.
7484 It is implementation defined how threads beyond the first are
7485 created, the name and type of the function called at thread
7486 startup, and how threads may be terminated. However, objects
7487 with thread storage duration shall be initialized before thread
7492 @cite{6.2.4 Storage durations of objects}
7494 Add new text before paragraph 3
7497 An object whose identifier is declared with the storage-class
7498 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
7499 Its lifetime is the entire execution of the thread, and its
7500 stored value is initialized only once, prior to thread startup.
7504 @cite{6.4.1 Keywords}
7506 Add @code{__thread}.
7509 @cite{6.7.1 Storage-class specifiers}
7511 Add @code{__thread} to the list of storage class specifiers in
7514 Change paragraph 2 to
7517 With the exception of @code{__thread}, at most one storage-class
7518 specifier may be given [@dots{}]. The @code{__thread} specifier may
7519 be used alone, or immediately following @code{extern} or
7523 Add new text after paragraph 6
7526 The declaration of an identifier for a variable that has
7527 block scope that specifies @code{__thread} shall also
7528 specify either @code{extern} or @code{static}.
7530 The @code{__thread} specifier shall be used only with
7535 @node C++98 Thread-Local Edits
7536 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
7538 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
7539 that document the exact semantics of the language extension.
7543 @b{[intro.execution]}
7545 New text after paragraph 4
7548 A @dfn{thread} is a flow of control within the abstract machine.
7549 It is implementation defined whether or not there may be more than
7553 New text after paragraph 7
7556 It is unspecified whether additional action must be taken to
7557 ensure when and whether side effects are visible to other threads.
7563 Add @code{__thread}.
7566 @b{[basic.start.main]}
7568 Add after paragraph 5
7571 The thread that begins execution at the @code{main} function is called
7572 the @dfn{main thread}. It is implementation defined how functions
7573 beginning threads other than the main thread are designated or typed.
7574 A function so designated, as well as the @code{main} function, is called
7575 a @dfn{thread startup function}. It is implementation defined what
7576 happens if a thread startup function returns. It is implementation
7577 defined what happens to other threads when any thread calls @code{exit}.
7581 @b{[basic.start.init]}
7583 Add after paragraph 4
7586 The storage for an object of thread storage duration shall be
7587 statically initialized before the first statement of the thread startup
7588 function. An object of thread storage duration shall not require
7589 dynamic initialization.
7593 @b{[basic.start.term]}
7595 Add after paragraph 3
7598 The type of an object with thread storage duration shall not have a
7599 non-trivial destructor, nor shall it be an array type whose elements
7600 (directly or indirectly) have non-trivial destructors.
7606 Add ``thread storage duration'' to the list in paragraph 1.
7611 Thread, static, and automatic storage durations are associated with
7612 objects introduced by declarations [@dots{}].
7615 Add @code{__thread} to the list of specifiers in paragraph 3.
7618 @b{[basic.stc.thread]}
7620 New section before @b{[basic.stc.static]}
7623 The keyword @code{__thread} applied to a non-local object gives the
7624 object thread storage duration.
7626 A local variable or class data member declared both @code{static}
7627 and @code{__thread} gives the variable or member thread storage
7632 @b{[basic.stc.static]}
7637 All objects which have neither thread storage duration, dynamic
7638 storage duration nor are local [@dots{}].
7644 Add @code{__thread} to the list in paragraph 1.
7649 With the exception of @code{__thread}, at most one
7650 @var{storage-class-specifier} shall appear in a given
7651 @var{decl-specifier-seq}. The @code{__thread} specifier may
7652 be used alone, or immediately following the @code{extern} or
7653 @code{static} specifiers. [@dots{}]
7656 Add after paragraph 5
7659 The @code{__thread} specifier can be applied only to the names of objects
7660 and to anonymous unions.
7666 Add after paragraph 6
7669 Non-@code{static} members shall not be @code{__thread}.
7673 @node C++ Extensions
7674 @chapter Extensions to the C++ Language
7675 @cindex extensions, C++ language
7676 @cindex C++ language extensions
7678 The GNU compiler provides these extensions to the C++ language (and you
7679 can also use most of the C language extensions in your C++ programs). If you
7680 want to write code that checks whether these features are available, you can
7681 test for the GNU compiler the same way as for C programs: check for a
7682 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
7683 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
7684 Predefined Macros,cpp,The GNU C Preprocessor}).
7687 * Min and Max:: C++ Minimum and maximum operators.
7688 * Volatiles:: What constitutes an access to a volatile object.
7689 * Restricted Pointers:: C99 restricted pointers and references.
7690 * Vague Linkage:: Where G++ puts inlines, vtables and such.
7691 * C++ Interface:: You can use a single C++ header file for both
7692 declarations and definitions.
7693 * Template Instantiation:: Methods for ensuring that exactly one copy of
7694 each needed template instantiation is emitted.
7695 * Bound member functions:: You can extract a function pointer to the
7696 method denoted by a @samp{->*} or @samp{.*} expression.
7697 * C++ Attributes:: Variable, function, and type attributes for C++ only.
7698 * Strong Using:: Strong using-directives for namespace composition.
7699 * Java Exceptions:: Tweaking exception handling to work with Java.
7700 * Deprecated Features:: Things will disappear from g++.
7701 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
7705 @section Minimum and Maximum Operators in C++
7707 It is very convenient to have operators which return the ``minimum'' or the
7708 ``maximum'' of two arguments. In GNU C++ (but not in GNU C),
7711 @item @var{a} <? @var{b}
7713 @cindex minimum operator
7714 is the @dfn{minimum}, returning the smaller of the numeric values
7715 @var{a} and @var{b};
7717 @item @var{a} >? @var{b}
7719 @cindex maximum operator
7720 is the @dfn{maximum}, returning the larger of the numeric values @var{a}
7724 These operations are not primitive in ordinary C++, since you can
7725 use a macro to return the minimum of two things in C++, as in the
7729 #define MIN(X,Y) ((X) < (Y) ? : (X) : (Y))
7733 You might then use @w{@samp{int min = MIN (i, j);}} to set @var{min} to
7734 the minimum value of variables @var{i} and @var{j}.
7736 However, side effects in @code{X} or @code{Y} may cause unintended
7737 behavior. For example, @code{MIN (i++, j++)} will fail, incrementing
7738 the smaller counter twice. The GNU C @code{typeof} extension allows you
7739 to write safe macros that avoid this kind of problem (@pxref{Typeof}).
7740 However, writing @code{MIN} and @code{MAX} as macros also forces you to
7741 use function-call notation for a fundamental arithmetic operation.
7742 Using GNU C++ extensions, you can write @w{@samp{int min = i <? j;}}
7745 Since @code{<?} and @code{>?} are built into the compiler, they properly
7746 handle expressions with side-effects; @w{@samp{int min = i++ <? j++;}}
7750 @section When is a Volatile Object Accessed?
7751 @cindex accessing volatiles
7752 @cindex volatile read
7753 @cindex volatile write
7754 @cindex volatile access
7756 Both the C and C++ standard have the concept of volatile objects. These
7757 are normally accessed by pointers and used for accessing hardware. The
7758 standards encourage compilers to refrain from optimizations
7759 concerning accesses to volatile objects that it might perform on
7760 non-volatile objects. The C standard leaves it implementation defined
7761 as to what constitutes a volatile access. The C++ standard omits to
7762 specify this, except to say that C++ should behave in a similar manner
7763 to C with respect to volatiles, where possible. The minimum either
7764 standard specifies is that at a sequence point all previous accesses to
7765 volatile objects have stabilized and no subsequent accesses have
7766 occurred. Thus an implementation is free to reorder and combine
7767 volatile accesses which occur between sequence points, but cannot do so
7768 for accesses across a sequence point. The use of volatiles does not
7769 allow you to violate the restriction on updating objects multiple times
7770 within a sequence point.
7772 In most expressions, it is intuitively obvious what is a read and what is
7773 a write. For instance
7776 volatile int *dst = @var{somevalue};
7777 volatile int *src = @var{someothervalue};
7782 will cause a read of the volatile object pointed to by @var{src} and stores the
7783 value into the volatile object pointed to by @var{dst}. There is no
7784 guarantee that these reads and writes are atomic, especially for objects
7785 larger than @code{int}.
7787 Less obvious expressions are where something which looks like an access
7788 is used in a void context. An example would be,
7791 volatile int *src = @var{somevalue};
7795 With C, such expressions are rvalues, and as rvalues cause a read of
7796 the object, GCC interprets this as a read of the volatile being pointed
7797 to. The C++ standard specifies that such expressions do not undergo
7798 lvalue to rvalue conversion, and that the type of the dereferenced
7799 object may be incomplete. The C++ standard does not specify explicitly
7800 that it is this lvalue to rvalue conversion which is responsible for
7801 causing an access. However, there is reason to believe that it is,
7802 because otherwise certain simple expressions become undefined. However,
7803 because it would surprise most programmers, G++ treats dereferencing a
7804 pointer to volatile object of complete type in a void context as a read
7805 of the object. When the object has incomplete type, G++ issues a
7810 struct T @{int m;@};
7811 volatile S *ptr1 = @var{somevalue};
7812 volatile T *ptr2 = @var{somevalue};
7817 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
7818 causes a read of the object pointed to. If you wish to force an error on
7819 the first case, you must force a conversion to rvalue with, for instance
7820 a static cast, @code{static_cast<S>(*ptr1)}.
7822 When using a reference to volatile, G++ does not treat equivalent
7823 expressions as accesses to volatiles, but instead issues a warning that
7824 no volatile is accessed. The rationale for this is that otherwise it
7825 becomes difficult to determine where volatile access occur, and not
7826 possible to ignore the return value from functions returning volatile
7827 references. Again, if you wish to force a read, cast the reference to
7830 @node Restricted Pointers
7831 @section Restricting Pointer Aliasing
7832 @cindex restricted pointers
7833 @cindex restricted references
7834 @cindex restricted this pointer
7836 As with the C front end, G++ understands the C99 feature of restricted pointers,
7837 specified with the @code{__restrict__}, or @code{__restrict} type
7838 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
7839 language flag, @code{restrict} is not a keyword in C++.
7841 In addition to allowing restricted pointers, you can specify restricted
7842 references, which indicate that the reference is not aliased in the local
7846 void fn (int *__restrict__ rptr, int &__restrict__ rref)
7853 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
7854 @var{rref} refers to a (different) unaliased integer.
7856 You may also specify whether a member function's @var{this} pointer is
7857 unaliased by using @code{__restrict__} as a member function qualifier.
7860 void T::fn () __restrict__
7867 Within the body of @code{T::fn}, @var{this} will have the effective
7868 definition @code{T *__restrict__ const this}. Notice that the
7869 interpretation of a @code{__restrict__} member function qualifier is
7870 different to that of @code{const} or @code{volatile} qualifier, in that it
7871 is applied to the pointer rather than the object. This is consistent with
7872 other compilers which implement restricted pointers.
7874 As with all outermost parameter qualifiers, @code{__restrict__} is
7875 ignored in function definition matching. This means you only need to
7876 specify @code{__restrict__} in a function definition, rather than
7877 in a function prototype as well.
7880 @section Vague Linkage
7881 @cindex vague linkage
7883 There are several constructs in C++ which require space in the object
7884 file but are not clearly tied to a single translation unit. We say that
7885 these constructs have ``vague linkage''. Typically such constructs are
7886 emitted wherever they are needed, though sometimes we can be more
7890 @item Inline Functions
7891 Inline functions are typically defined in a header file which can be
7892 included in many different compilations. Hopefully they can usually be
7893 inlined, but sometimes an out-of-line copy is necessary, if the address
7894 of the function is taken or if inlining fails. In general, we emit an
7895 out-of-line copy in all translation units where one is needed. As an
7896 exception, we only emit inline virtual functions with the vtable, since
7897 it will always require a copy.
7899 Local static variables and string constants used in an inline function
7900 are also considered to have vague linkage, since they must be shared
7901 between all inlined and out-of-line instances of the function.
7905 C++ virtual functions are implemented in most compilers using a lookup
7906 table, known as a vtable. The vtable contains pointers to the virtual
7907 functions provided by a class, and each object of the class contains a
7908 pointer to its vtable (or vtables, in some multiple-inheritance
7909 situations). If the class declares any non-inline, non-pure virtual
7910 functions, the first one is chosen as the ``key method'' for the class,
7911 and the vtable is only emitted in the translation unit where the key
7914 @emph{Note:} If the chosen key method is later defined as inline, the
7915 vtable will still be emitted in every translation unit which defines it.
7916 Make sure that any inline virtuals are declared inline in the class
7917 body, even if they are not defined there.
7919 @item type_info objects
7922 C++ requires information about types to be written out in order to
7923 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
7924 For polymorphic classes (classes with virtual functions), the type_info
7925 object is written out along with the vtable so that @samp{dynamic_cast}
7926 can determine the dynamic type of a class object at runtime. For all
7927 other types, we write out the type_info object when it is used: when
7928 applying @samp{typeid} to an expression, throwing an object, or
7929 referring to a type in a catch clause or exception specification.
7931 @item Template Instantiations
7932 Most everything in this section also applies to template instantiations,
7933 but there are other options as well.
7934 @xref{Template Instantiation,,Where's the Template?}.
7938 When used with GNU ld version 2.8 or later on an ELF system such as
7939 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
7940 these constructs will be discarded at link time. This is known as
7943 On targets that don't support COMDAT, but do support weak symbols, GCC
7944 will use them. This way one copy will override all the others, but
7945 the unused copies will still take up space in the executable.
7947 For targets which do not support either COMDAT or weak symbols,
7948 most entities with vague linkage will be emitted as local symbols to
7949 avoid duplicate definition errors from the linker. This will not happen
7950 for local statics in inlines, however, as having multiple copies will
7951 almost certainly break things.
7953 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
7954 another way to control placement of these constructs.
7957 @section #pragma interface and implementation
7959 @cindex interface and implementation headers, C++
7960 @cindex C++ interface and implementation headers
7961 @cindex pragmas, interface and implementation
7963 @code{#pragma interface} and @code{#pragma implementation} provide the
7964 user with a way of explicitly directing the compiler to emit entities
7965 with vague linkage (and debugging information) in a particular
7968 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
7969 most cases, because of COMDAT support and the ``key method'' heuristic
7970 mentioned in @ref{Vague Linkage}. Using them can actually cause your
7971 program to grow due to unnecesary out-of-line copies of inline
7972 functions. Currently (3.4) the only benefit of these
7973 @code{#pragma}s is reduced duplication of debugging information, and
7974 that should be addressed soon on DWARF 2 targets with the use of
7978 @item #pragma interface
7979 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
7980 @kindex #pragma interface
7981 Use this directive in @emph{header files} that define object classes, to save
7982 space in most of the object files that use those classes. Normally,
7983 local copies of certain information (backup copies of inline member
7984 functions, debugging information, and the internal tables that implement
7985 virtual functions) must be kept in each object file that includes class
7986 definitions. You can use this pragma to avoid such duplication. When a
7987 header file containing @samp{#pragma interface} is included in a
7988 compilation, this auxiliary information will not be generated (unless
7989 the main input source file itself uses @samp{#pragma implementation}).
7990 Instead, the object files will contain references to be resolved at link
7993 The second form of this directive is useful for the case where you have
7994 multiple headers with the same name in different directories. If you
7995 use this form, you must specify the same string to @samp{#pragma
7998 @item #pragma implementation
7999 @itemx #pragma implementation "@var{objects}.h"
8000 @kindex #pragma implementation
8001 Use this pragma in a @emph{main input file}, when you want full output from
8002 included header files to be generated (and made globally visible). The
8003 included header file, in turn, should use @samp{#pragma interface}.
8004 Backup copies of inline member functions, debugging information, and the
8005 internal tables used to implement virtual functions are all generated in
8006 implementation files.
8008 @cindex implied @code{#pragma implementation}
8009 @cindex @code{#pragma implementation}, implied
8010 @cindex naming convention, implementation headers
8011 If you use @samp{#pragma implementation} with no argument, it applies to
8012 an include file with the same basename@footnote{A file's @dfn{basename}
8013 was the name stripped of all leading path information and of trailing
8014 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
8015 file. For example, in @file{allclass.cc}, giving just
8016 @samp{#pragma implementation}
8017 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
8019 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
8020 an implementation file whenever you would include it from
8021 @file{allclass.cc} even if you never specified @samp{#pragma
8022 implementation}. This was deemed to be more trouble than it was worth,
8023 however, and disabled.
8025 Use the string argument if you want a single implementation file to
8026 include code from multiple header files. (You must also use
8027 @samp{#include} to include the header file; @samp{#pragma
8028 implementation} only specifies how to use the file---it doesn't actually
8031 There is no way to split up the contents of a single header file into
8032 multiple implementation files.
8035 @cindex inlining and C++ pragmas
8036 @cindex C++ pragmas, effect on inlining
8037 @cindex pragmas in C++, effect on inlining
8038 @samp{#pragma implementation} and @samp{#pragma interface} also have an
8039 effect on function inlining.
8041 If you define a class in a header file marked with @samp{#pragma
8042 interface}, the effect on an inline function defined in that class is
8043 similar to an explicit @code{extern} declaration---the compiler emits
8044 no code at all to define an independent version of the function. Its
8045 definition is used only for inlining with its callers.
8047 @opindex fno-implement-inlines
8048 Conversely, when you include the same header file in a main source file
8049 that declares it as @samp{#pragma implementation}, the compiler emits
8050 code for the function itself; this defines a version of the function
8051 that can be found via pointers (or by callers compiled without
8052 inlining). If all calls to the function can be inlined, you can avoid
8053 emitting the function by compiling with @option{-fno-implement-inlines}.
8054 If any calls were not inlined, you will get linker errors.
8056 @node Template Instantiation
8057 @section Where's the Template?
8058 @cindex template instantiation
8060 C++ templates are the first language feature to require more
8061 intelligence from the environment than one usually finds on a UNIX
8062 system. Somehow the compiler and linker have to make sure that each
8063 template instance occurs exactly once in the executable if it is needed,
8064 and not at all otherwise. There are two basic approaches to this
8065 problem, which are referred to as the Borland model and the Cfront model.
8069 Borland C++ solved the template instantiation problem by adding the code
8070 equivalent of common blocks to their linker; the compiler emits template
8071 instances in each translation unit that uses them, and the linker
8072 collapses them together. The advantage of this model is that the linker
8073 only has to consider the object files themselves; there is no external
8074 complexity to worry about. This disadvantage is that compilation time
8075 is increased because the template code is being compiled repeatedly.
8076 Code written for this model tends to include definitions of all
8077 templates in the header file, since they must be seen to be
8081 The AT&T C++ translator, Cfront, solved the template instantiation
8082 problem by creating the notion of a template repository, an
8083 automatically maintained place where template instances are stored. A
8084 more modern version of the repository works as follows: As individual
8085 object files are built, the compiler places any template definitions and
8086 instantiations encountered in the repository. At link time, the link
8087 wrapper adds in the objects in the repository and compiles any needed
8088 instances that were not previously emitted. The advantages of this
8089 model are more optimal compilation speed and the ability to use the
8090 system linker; to implement the Borland model a compiler vendor also
8091 needs to replace the linker. The disadvantages are vastly increased
8092 complexity, and thus potential for error; for some code this can be
8093 just as transparent, but in practice it can been very difficult to build
8094 multiple programs in one directory and one program in multiple
8095 directories. Code written for this model tends to separate definitions
8096 of non-inline member templates into a separate file, which should be
8097 compiled separately.
8100 When used with GNU ld version 2.8 or later on an ELF system such as
8101 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
8102 Borland model. On other systems, G++ implements neither automatic
8105 A future version of G++ will support a hybrid model whereby the compiler
8106 will emit any instantiations for which the template definition is
8107 included in the compile, and store template definitions and
8108 instantiation context information into the object file for the rest.
8109 The link wrapper will extract that information as necessary and invoke
8110 the compiler to produce the remaining instantiations. The linker will
8111 then combine duplicate instantiations.
8113 In the mean time, you have the following options for dealing with
8114 template instantiations:
8119 Compile your template-using code with @option{-frepo}. The compiler will
8120 generate files with the extension @samp{.rpo} listing all of the
8121 template instantiations used in the corresponding object files which
8122 could be instantiated there; the link wrapper, @samp{collect2}, will
8123 then update the @samp{.rpo} files to tell the compiler where to place
8124 those instantiations and rebuild any affected object files. The
8125 link-time overhead is negligible after the first pass, as the compiler
8126 will continue to place the instantiations in the same files.
8128 This is your best option for application code written for the Borland
8129 model, as it will just work. Code written for the Cfront model will
8130 need to be modified so that the template definitions are available at
8131 one or more points of instantiation; usually this is as simple as adding
8132 @code{#include <tmethods.cc>} to the end of each template header.
8134 For library code, if you want the library to provide all of the template
8135 instantiations it needs, just try to link all of its object files
8136 together; the link will fail, but cause the instantiations to be
8137 generated as a side effect. Be warned, however, that this may cause
8138 conflicts if multiple libraries try to provide the same instantiations.
8139 For greater control, use explicit instantiation as described in the next
8143 @opindex fno-implicit-templates
8144 Compile your code with @option{-fno-implicit-templates} to disable the
8145 implicit generation of template instances, and explicitly instantiate
8146 all the ones you use. This approach requires more knowledge of exactly
8147 which instances you need than do the others, but it's less
8148 mysterious and allows greater control. You can scatter the explicit
8149 instantiations throughout your program, perhaps putting them in the
8150 translation units where the instances are used or the translation units
8151 that define the templates themselves; you can put all of the explicit
8152 instantiations you need into one big file; or you can create small files
8159 template class Foo<int>;
8160 template ostream& operator <<
8161 (ostream&, const Foo<int>&);
8164 for each of the instances you need, and create a template instantiation
8167 If you are using Cfront-model code, you can probably get away with not
8168 using @option{-fno-implicit-templates} when compiling files that don't
8169 @samp{#include} the member template definitions.
8171 If you use one big file to do the instantiations, you may want to
8172 compile it without @option{-fno-implicit-templates} so you get all of the
8173 instances required by your explicit instantiations (but not by any
8174 other files) without having to specify them as well.
8176 G++ has extended the template instantiation syntax given in the ISO
8177 standard to allow forward declaration of explicit instantiations
8178 (with @code{extern}), instantiation of the compiler support data for a
8179 template class (i.e.@: the vtable) without instantiating any of its
8180 members (with @code{inline}), and instantiation of only the static data
8181 members of a template class, without the support data or member
8182 functions (with (@code{static}):
8185 extern template int max (int, int);
8186 inline template class Foo<int>;
8187 static template class Foo<int>;
8191 Do nothing. Pretend G++ does implement automatic instantiation
8192 management. Code written for the Borland model will work fine, but
8193 each translation unit will contain instances of each of the templates it
8194 uses. In a large program, this can lead to an unacceptable amount of code
8198 @node Bound member functions
8199 @section Extracting the function pointer from a bound pointer to member function
8201 @cindex pointer to member function
8202 @cindex bound pointer to member function
8204 In C++, pointer to member functions (PMFs) are implemented using a wide
8205 pointer of sorts to handle all the possible call mechanisms; the PMF
8206 needs to store information about how to adjust the @samp{this} pointer,
8207 and if the function pointed to is virtual, where to find the vtable, and
8208 where in the vtable to look for the member function. If you are using
8209 PMFs in an inner loop, you should really reconsider that decision. If
8210 that is not an option, you can extract the pointer to the function that
8211 would be called for a given object/PMF pair and call it directly inside
8212 the inner loop, to save a bit of time.
8214 Note that you will still be paying the penalty for the call through a
8215 function pointer; on most modern architectures, such a call defeats the
8216 branch prediction features of the CPU@. This is also true of normal
8217 virtual function calls.
8219 The syntax for this extension is
8223 extern int (A::*fp)();
8224 typedef int (*fptr)(A *);
8226 fptr p = (fptr)(a.*fp);
8229 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
8230 no object is needed to obtain the address of the function. They can be
8231 converted to function pointers directly:
8234 fptr p1 = (fptr)(&A::foo);
8237 @opindex Wno-pmf-conversions
8238 You must specify @option{-Wno-pmf-conversions} to use this extension.
8240 @node C++ Attributes
8241 @section C++-Specific Variable, Function, and Type Attributes
8243 Some attributes only make sense for C++ programs.
8246 @item init_priority (@var{priority})
8247 @cindex init_priority attribute
8250 In Standard C++, objects defined at namespace scope are guaranteed to be
8251 initialized in an order in strict accordance with that of their definitions
8252 @emph{in a given translation unit}. No guarantee is made for initializations
8253 across translation units. However, GNU C++ allows users to control the
8254 order of initialization of objects defined at namespace scope with the
8255 @code{init_priority} attribute by specifying a relative @var{priority},
8256 a constant integral expression currently bounded between 101 and 65535
8257 inclusive. Lower numbers indicate a higher priority.
8259 In the following example, @code{A} would normally be created before
8260 @code{B}, but the @code{init_priority} attribute has reversed that order:
8263 Some_Class A __attribute__ ((init_priority (2000)));
8264 Some_Class B __attribute__ ((init_priority (543)));
8268 Note that the particular values of @var{priority} do not matter; only their
8271 @item java_interface
8272 @cindex java_interface attribute
8274 This type attribute informs C++ that the class is a Java interface. It may
8275 only be applied to classes declared within an @code{extern "Java"} block.
8276 Calls to methods declared in this interface will be dispatched using GCJ's
8277 interface table mechanism, instead of regular virtual table dispatch.
8281 See also @xref{Strong Using}.
8284 @section Strong Using
8286 @strong{Caution:} The semantics of this extension are not fully
8287 defined. Users should refrain from using this extension as its
8288 semantics may change subtly over time. It is possible that this
8289 extension wil be removed in future versions of G++.
8291 A using-directive with @code{__attribute ((strong))} is stronger
8292 than a normal using-directive in two ways:
8296 Templates from the used namespace can be specialized as though they were members of the using namespace.
8299 The using namespace is considered an associated namespace of all
8300 templates in the used namespace for purposes of argument-dependent
8304 This is useful for composing a namespace transparently from
8305 implementation namespaces. For example:
8310 template <class T> struct A @{ @};
8312 using namespace debug __attribute ((__strong__));
8313 template <> struct A<int> @{ @}; // ok to specialize
8315 template <class T> void f (A<T>);
8320 f (std::A<float>()); // lookup finds std::f
8325 @node Java Exceptions
8326 @section Java Exceptions
8328 The Java language uses a slightly different exception handling model
8329 from C++. Normally, GNU C++ will automatically detect when you are
8330 writing C++ code that uses Java exceptions, and handle them
8331 appropriately. However, if C++ code only needs to execute destructors
8332 when Java exceptions are thrown through it, GCC will guess incorrectly.
8333 Sample problematic code is:
8336 struct S @{ ~S(); @};
8337 extern void bar(); // is written in Java, and may throw exceptions
8346 The usual effect of an incorrect guess is a link failure, complaining of
8347 a missing routine called @samp{__gxx_personality_v0}.
8349 You can inform the compiler that Java exceptions are to be used in a
8350 translation unit, irrespective of what it might think, by writing
8351 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
8352 @samp{#pragma} must appear before any functions that throw or catch
8353 exceptions, or run destructors when exceptions are thrown through them.
8355 You cannot mix Java and C++ exceptions in the same translation unit. It
8356 is believed to be safe to throw a C++ exception from one file through
8357 another file compiled for the Java exception model, or vice versa, but
8358 there may be bugs in this area.
8360 @node Deprecated Features
8361 @section Deprecated Features
8363 In the past, the GNU C++ compiler was extended to experiment with new
8364 features, at a time when the C++ language was still evolving. Now that
8365 the C++ standard is complete, some of those features are superseded by
8366 superior alternatives. Using the old features might cause a warning in
8367 some cases that the feature will be dropped in the future. In other
8368 cases, the feature might be gone already.
8370 While the list below is not exhaustive, it documents some of the options
8371 that are now deprecated:
8374 @item -fexternal-templates
8375 @itemx -falt-external-templates
8376 These are two of the many ways for G++ to implement template
8377 instantiation. @xref{Template Instantiation}. The C++ standard clearly
8378 defines how template definitions have to be organized across
8379 implementation units. G++ has an implicit instantiation mechanism that
8380 should work just fine for standard-conforming code.
8382 @item -fstrict-prototype
8383 @itemx -fno-strict-prototype
8384 Previously it was possible to use an empty prototype parameter list to
8385 indicate an unspecified number of parameters (like C), rather than no
8386 parameters, as C++ demands. This feature has been removed, except where
8387 it is required for backwards compatibility @xref{Backwards Compatibility}.
8390 The named return value extension has been deprecated, and is now
8393 The use of initializer lists with new expressions has been deprecated,
8394 and is now removed from G++.
8396 Floating and complex non-type template parameters have been deprecated,
8397 and are now removed from G++.
8399 The implicit typename extension has been deprecated and is now
8402 The use of default arguments in function pointers, function typedefs and
8403 and other places where they are not permitted by the standard is
8404 deprecated and will be removed from a future version of G++.
8406 @node Backwards Compatibility
8407 @section Backwards Compatibility
8408 @cindex Backwards Compatibility
8409 @cindex ARM [Annotated C++ Reference Manual]
8411 Now that there is a definitive ISO standard C++, G++ has a specification
8412 to adhere to. The C++ language evolved over time, and features that
8413 used to be acceptable in previous drafts of the standard, such as the ARM
8414 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
8415 compilation of C++ written to such drafts, G++ contains some backwards
8416 compatibilities. @emph{All such backwards compatibility features are
8417 liable to disappear in future versions of G++.} They should be considered
8418 deprecated @xref{Deprecated Features}.
8422 If a variable is declared at for scope, it used to remain in scope until
8423 the end of the scope which contained the for statement (rather than just
8424 within the for scope). G++ retains this, but issues a warning, if such a
8425 variable is accessed outside the for scope.
8427 @item Implicit C language
8428 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
8429 scope to set the language. On such systems, all header files are
8430 implicitly scoped inside a C language scope. Also, an empty prototype
8431 @code{()} will be treated as an unspecified number of arguments, rather
8432 than no arguments, as C++ demands.