1 @c Copyright (C) 1988,1989,1992,1993,1994,1996,1998,1999,2000,2001,2002 Free Software Foundation, Inc.
2 @c This is part of the GCC manual.
3 @c For copying conditions, see the file gcc.texi.
6 @chapter C Implementation-defined behavior
7 @cindex implementation-defined behavior, C language
9 A conforming implementation of ISO C is required to document its
10 choice of behavior in each of the areas that are designated
11 ``implementation defined.'' The following lists all such areas,
12 along with the section number from the ISO/IEC 9899:1999 standard.
15 * Translation implementation::
16 * Environment implementation::
17 * Identifiers implementation::
18 * Characters implementation::
19 * Integers implementation::
20 * Floating point implementation::
21 * Arrays and pointers implementation::
22 * Hints implementation::
23 * Structures unions enumerations and bit-fields implementation::
24 * Qualifiers implementation::
25 * Preprocessing directives implementation::
26 * Library functions implementation::
27 * Architecture implementation::
28 * Locale-specific behavior implementation::
31 @node Translation implementation
36 @cite{How a diagnostic is identified (3.10, 5.1.1.3).}
39 @cite{Whether each nonempty sequence of white-space characters other than
40 new-line is retained or replaced by one space character in translation
44 @node Environment implementation
47 The behavior of these points are dependent on the implementation
48 of the C library, and are not defined by GCC itself.
50 @node Identifiers implementation
55 @cite{Which additional multibyte characters may appear in identifiers
56 and their correspondence to universal character names (6.4.2).}
59 @cite{The number of significant initial characters in an identifier
63 @node Characters implementation
68 @cite{The number of bits in a byte (3.6).}
71 @cite{The values of the members of the execution character set (5.2.1).}
74 @cite{The unique value of the member of the execution character set produced
75 for each of the standard alphabetic escape sequences (5.2.2).}
78 @cite{The value of a @code{char} object into which has been stored any
79 character other than a member of the basic execution character set (6.2.5).}
82 @cite{Which of @code{signed char} or @code{unsigned char} has the same range,
83 representation, and behavior as ``plain'' @code{char} (6.2.5, 6.3.1.1).}
86 @cite{The mapping of members of the source character set (in character
87 constants and string literals) to members of the execution character
88 set (6.4.4.4, 5.1.1.2).}
91 @cite{The value of an integer character constant containing more than one
92 character or containing a character or escape sequence that does not map
93 to a single-byte execution character (6.4.4.4).}
96 @cite{The value of a wide character constant containing more than one
97 multibyte character, or containing a multibyte character or escape
98 sequence not represented in the extended execution character set (6.4.4.4).}
101 @cite{The current locale used to convert a wide character constant consisting
102 of a single multibyte character that maps to a member of the extended
103 execution character set into a corresponding wide character code (6.4.4.4).}
106 @cite{The current locale used to convert a wide string literal into
107 corresponding wide character codes (6.4.5).}
110 @cite{The value of a string literal containing a multibyte character or escape
111 sequence not represented in the execution character set (6.4.5).}
114 @node Integers implementation
119 @cite{Any extended integer types that exist in the implementation (6.2.5).}
122 @cite{Whether signed integer types are represented using sign and magnitude,
123 two's complement, or one's complement, and whether the extraordinary value
124 is a trap representation or an ordinary value (6.2.6.2).}
127 @cite{The rank of any extended integer type relative to another extended
128 integer type with the same precision (6.3.1.1).}
131 @cite{The result of, or the signal raised by, converting an integer to a
132 signed integer type when the value cannot be represented in an object of
133 that type (6.3.1.3).}
136 @cite{The results of some bitwise operations on signed integers (6.5).}
139 @node Floating point implementation
140 @section Floating point
144 @cite{The accuracy of the floating-point operations and of the library
145 functions in @code{<math.h>} and @code{<complex.h>} that return floating-point
146 results (5.2.4.2.2).}
149 @cite{The rounding behaviors characterized by non-standard values
150 of @code{FLT_ROUNDS} @gol
154 @cite{The evaluation methods characterized by non-standard negative
155 values of @code{FLT_EVAL_METHOD} (5.2.4.2.2).}
158 @cite{The direction of rounding when an integer is converted to a
159 floating-point number that cannot exactly represent the original
163 @cite{The direction of rounding when a floating-point number is
164 converted to a narrower floating-point number (6.3.1.5).}
167 @cite{How the nearest representable value or the larger or smaller
168 representable value immediately adjacent to the nearest representable
169 value is chosen for certain floating constants (6.4.4.2).}
172 @cite{Whether and how floating expressions are contracted when not
173 disallowed by the @code{FP_CONTRACT} pragma (6.5).}
176 @cite{The default state for the @code{FENV_ACCESS} pragma (7.6.1).}
179 @cite{Additional floating-point exceptions, rounding modes, environments,
180 and classifications, and their macro names (7.6, 7.12).}
183 @cite{The default state for the @code{FP_CONTRACT} pragma (7.12.2).}
186 @cite{Whether the ``inexact'' floating-point exception can be raised
187 when the rounded result actually does equal the mathematical result
188 in an IEC 60559 conformant implementation (F.9).}
191 @cite{Whether the ``underflow'' (and ``inexact'') floating-point
192 exception can be raised when a result is tiny but not inexact in an
193 IEC 60559 conformant implementation (F.9).}
197 @node Arrays and pointers implementation
198 @section Arrays and pointers
202 @cite{The result of converting a pointer to an integer or
203 vice versa (6.3.2.3).}
205 A cast from pointer to integer discards most-significant bits if the
206 pointer representation is larger than the integer type,
207 sign-extends@footnote{Future versions of GCC may zero-extend, or use
208 a target-defined @code{ptr_extend} pattern. Do not rely on sign extension.}
209 if the pointer representation is smaller than the integer type, otherwise
210 the bits are unchanged.
211 @c ??? We've always claimed that pointers were unsigned entities.
212 @c Shouldn't we therefore be doing zero-extension? If so, the bug
213 @c is in convert_to_integer, where we call type_for_size and request
214 @c a signed integral type. On the other hand, it might be most useful
215 @c for the target if we extend according to POINTERS_EXTEND_UNSIGNED.
217 A cast from integer to pointer discards most-significant bits if the
218 pointer representation is smaller than the integer type, extends according
219 to the signedness of the integer type if the pointer representation
220 is larger than the integer type, otherwise the bits are unchanged.
222 When casting from pointer to integer and back again, the resulting
223 pointer must reference the same object as the original pointer, otherwise
224 the behavior is undefined. That is, one may not use integer arithmetic to
225 avoid the undefined behavior of pointer arithmetic as proscribed in 6.5.6/8.
228 @cite{The size of the result of subtracting two pointers to elements
229 of the same array (6.5.6).}
233 @node Hints implementation
238 @cite{The extent to which suggestions made by using the @code{register}
239 storage-class specifier are effective (6.7.1).}
242 @cite{The extent to which suggestions made by using the inline function
243 specifier are effective (6.7.4).}
247 @node Structures unions enumerations and bit-fields implementation
248 @section Structures, unions, enumerations, and bit-fields
252 @cite{Whether a ``plain'' int bit-field is treated as a @code{signed int}
253 bit-field or as an @code{unsigned int} bit-field (6.7.2, 6.7.2.1).}
256 @cite{Allowable bit-field types other than @code{_Bool}, @code{signed int},
257 and @code{unsigned int} (6.7.2.1).}
260 @cite{Whether a bit-field can straddle a storage-unit boundary (6.7.2.1).}
263 @cite{The order of allocation of bit-fields within a unit (6.7.2.1).}
266 @cite{The alignment of non-bit-field members of structures (6.7.2.1).}
269 @cite{The integer type compatible with each enumerated type (6.7.2.2).}
273 @node Qualifiers implementation
278 @cite{What constitutes an access to an object that has volatile-qualified
283 @node Preprocessing directives implementation
284 @section Preprocessing directives
288 @cite{How sequences in both forms of header names are mapped to headers
289 or external source file names (6.4.7).}
292 @cite{Whether the value of a character constant in a constant expression
293 that controls conditional inclusion matches the value of the same character
294 constant in the execution character set (6.10.1).}
297 @cite{Whether the value of a single-character character constant in a
298 constant expression that controls conditional inclusion may have a
299 negative value (6.10.1).}
302 @cite{The places that are searched for an included @samp{<>} delimited
303 header, and how the places are specified or the header is
304 identified (6.10.2).}
307 @cite{How the named source file is searched for in an included @samp{""}
308 delimited header (6.10.2).}
311 @cite{The method by which preprocessing tokens (possibly resulting from
312 macro expansion) in a @code{#include} directive are combined into a header
316 @cite{The nesting limit for @code{#include} processing (6.10.2).}
319 @cite{Whether the @samp{#} operator inserts a @samp{\} character before
320 the @samp{\} character that begins a universal character name in a
321 character constant or string literal (6.10.3.2).}
324 @cite{The behavior on each recognized non-@code{STDC #pragma}
328 @cite{The definitions for @code{__DATE__} and @code{__TIME__} when
329 respectively, the date and time of translation are not available (6.10.8).}
333 @node Library functions implementation
334 @section Library functions
336 The behavior of these points are dependent on the implementation
337 of the C library, and are not defined by GCC itself.
339 @node Architecture implementation
340 @section Architecture
344 @cite{The values or expressions assigned to the macros specified in the
345 headers @code{<float.h>}, @code{<limits.h>}, and @code{<stdint.h>}
346 (5.2.4.2, 7.18.2, 7.18.3).}
349 @cite{The number, order, and encoding of bytes in any object
350 (when not explicitly specified in this International Standard) (6.2.6.1).}
353 @cite{The value of the result of the sizeof operator (6.5.3.4).}
357 @node Locale-specific behavior implementation
358 @section Locale-specific behavior
360 The behavior of these points are dependent on the implementation
361 of the C library, and are not defined by GCC itself.
364 @chapter Extensions to the C Language Family
365 @cindex extensions, C language
366 @cindex C language extensions
369 GNU C provides several language features not found in ISO standard C@.
370 (The @option{-pedantic} option directs GCC to print a warning message if
371 any of these features is used.) To test for the availability of these
372 features in conditional compilation, check for a predefined macro
373 @code{__GNUC__}, which is always defined under GCC@.
375 These extensions are available in C and Objective-C@. Most of them are
376 also available in C++. @xref{C++ Extensions,,Extensions to the
377 C++ Language}, for extensions that apply @emph{only} to C++.
379 Some features that are in ISO C99 but not C89 or C++ are also, as
380 extensions, accepted by GCC in C89 mode and in C++.
383 * Statement Exprs:: Putting statements and declarations inside expressions.
384 * Local Labels:: Labels local to a statement-expression.
385 * Labels as Values:: Getting pointers to labels, and computed gotos.
386 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
387 * Constructing Calls:: Dispatching a call to another function.
388 * Naming Types:: Giving a name to the type of some expression.
389 * Typeof:: @code{typeof}: referring to the type of an expression.
390 * Lvalues:: Using @samp{?:}, @samp{,} and casts in lvalues.
391 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
392 * Long Long:: Double-word integers---@code{long long int}.
393 * Complex:: Data types for complex numbers.
394 * Hex Floats:: Hexadecimal floating-point constants.
395 * Zero Length:: Zero-length arrays.
396 * Variable Length:: Arrays whose length is computed at run time.
397 * Variadic Macros:: Macros with a variable number of arguments.
398 * Escaped Newlines:: Slightly looser rules for escaped newlines.
399 * Multi-line Strings:: String literals with embedded newlines.
400 * Subscripting:: Any array can be subscripted, even if not an lvalue.
401 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
402 * Initializers:: Non-constant initializers.
403 * Compound Literals:: Compound literals give structures, unions
405 * Designated Inits:: Labeling elements of initializers.
406 * Cast to Union:: Casting to union type from any member of the union.
407 * Case Ranges:: `case 1 ... 9' and such.
408 * Mixed Declarations:: Mixing declarations and code.
409 * Function Attributes:: Declaring that functions have no side effects,
410 or that they can never return.
411 * Attribute Syntax:: Formal syntax for attributes.
412 * Function Prototypes:: Prototype declarations and old-style definitions.
413 * C++ Comments:: C++ comments are recognized.
414 * Dollar Signs:: Dollar sign is allowed in identifiers.
415 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
416 * Variable Attributes:: Specifying attributes of variables.
417 * Type Attributes:: Specifying attributes of types.
418 * Alignment:: Inquiring about the alignment of a type or variable.
419 * Inline:: Defining inline functions (as fast as macros).
420 * Extended Asm:: Assembler instructions with C expressions as operands.
421 (With them you can define ``built-in'' functions.)
422 * Constraints:: Constraints for asm operands
423 * Asm Labels:: Specifying the assembler name to use for a C symbol.
424 * Explicit Reg Vars:: Defining variables residing in specified registers.
425 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
426 * Incomplete Enums:: @code{enum foo;}, with details to follow.
427 * Function Names:: Printable strings which are the name of the current
429 * Return Address:: Getting the return or frame address of a function.
430 * Vector Extensions:: Using vector instructions through built-in functions.
431 * Other Builtins:: Other built-in functions.
432 * Target Builtins:: Built-in functions specific to particular targets.
433 * Pragmas:: Pragmas accepted by GCC.
434 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
435 * Thread-Local:: Per-thread variables.
438 @node Statement Exprs
439 @section Statements and Declarations in Expressions
440 @cindex statements inside expressions
441 @cindex declarations inside expressions
442 @cindex expressions containing statements
443 @cindex macros, statements in expressions
445 @c the above section title wrapped and causes an underfull hbox.. i
446 @c changed it from "within" to "in". --mew 4feb93
448 A compound statement enclosed in parentheses may appear as an expression
449 in GNU C@. This allows you to use loops, switches, and local variables
450 within an expression.
452 Recall that a compound statement is a sequence of statements surrounded
453 by braces; in this construct, parentheses go around the braces. For
457 (@{ int y = foo (); int z;
464 is a valid (though slightly more complex than necessary) expression
465 for the absolute value of @code{foo ()}.
467 The last thing in the compound statement should be an expression
468 followed by a semicolon; the value of this subexpression serves as the
469 value of the entire construct. (If you use some other kind of statement
470 last within the braces, the construct has type @code{void}, and thus
471 effectively no value.)
473 This feature is especially useful in making macro definitions ``safe'' (so
474 that they evaluate each operand exactly once). For example, the
475 ``maximum'' function is commonly defined as a macro in standard C as
479 #define max(a,b) ((a) > (b) ? (a) : (b))
483 @cindex side effects, macro argument
484 But this definition computes either @var{a} or @var{b} twice, with bad
485 results if the operand has side effects. In GNU C, if you know the
486 type of the operands (here let's assume @code{int}), you can define
487 the macro safely as follows:
490 #define maxint(a,b) \
491 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
494 Embedded statements are not allowed in constant expressions, such as
495 the value of an enumeration constant, the width of a bit-field, or
496 the initial value of a static variable.
498 If you don't know the type of the operand, you can still do this, but you
499 must use @code{typeof} (@pxref{Typeof}) or type naming (@pxref{Naming
502 Statement expressions are not supported fully in G++, and their fate
503 there is unclear. (It is possible that they will become fully supported
504 at some point, or that they will be deprecated, or that the bugs that
505 are present will continue to exist indefinitely.) Presently, statement
506 expressions do not work well as default arguments.
508 In addition, there are semantic issues with statement-expressions in
509 C++. If you try to use statement-expressions instead of inline
510 functions in C++, you may be surprised at the way object destruction is
511 handled. For example:
514 #define foo(a) (@{int b = (a); b + 3; @})
518 does not work the same way as:
521 inline int foo(int a) @{ int b = a; return b + 3; @}
525 In particular, if the expression passed into @code{foo} involves the
526 creation of temporaries, the destructors for those temporaries will be
527 run earlier in the case of the macro than in the case of the function.
529 These considerations mean that it is probably a bad idea to use
530 statement-expressions of this form in header files that are designed to
531 work with C++. (Note that some versions of the GNU C Library contained
532 header files using statement-expression that lead to precisely this
536 @section Locally Declared Labels
538 @cindex macros, local labels
540 Each statement expression is a scope in which @dfn{local labels} can be
541 declared. A local label is simply an identifier; you can jump to it
542 with an ordinary @code{goto} statement, but only from within the
543 statement expression it belongs to.
545 A local label declaration looks like this:
548 __label__ @var{label};
555 __label__ @var{label1}, @var{label2}, @dots{};
558 Local label declarations must come at the beginning of the statement
559 expression, right after the @samp{(@{}, before any ordinary
562 The label declaration defines the label @emph{name}, but does not define
563 the label itself. You must do this in the usual way, with
564 @code{@var{label}:}, within the statements of the statement expression.
566 The local label feature is useful because statement expressions are
567 often used in macros. If the macro contains nested loops, a @code{goto}
568 can be useful for breaking out of them. However, an ordinary label
569 whose scope is the whole function cannot be used: if the macro can be
570 expanded several times in one function, the label will be multiply
571 defined in that function. A local label avoids this problem. For
575 #define SEARCH(array, target) \
578 typeof (target) _SEARCH_target = (target); \
579 typeof (*(array)) *_SEARCH_array = (array); \
582 for (i = 0; i < max; i++) \
583 for (j = 0; j < max; j++) \
584 if (_SEARCH_array[i][j] == _SEARCH_target) \
585 @{ value = i; goto found; @} \
592 @node Labels as Values
593 @section Labels as Values
594 @cindex labels as values
595 @cindex computed gotos
596 @cindex goto with computed label
597 @cindex address of a label
599 You can get the address of a label defined in the current function
600 (or a containing function) with the unary operator @samp{&&}. The
601 value has type @code{void *}. This value is a constant and can be used
602 wherever a constant of that type is valid. For example:
610 To use these values, you need to be able to jump to one. This is done
611 with the computed goto statement@footnote{The analogous feature in
612 Fortran is called an assigned goto, but that name seems inappropriate in
613 C, where one can do more than simply store label addresses in label
614 variables.}, @code{goto *@var{exp};}. For example,
621 Any expression of type @code{void *} is allowed.
623 One way of using these constants is in initializing a static array that
624 will serve as a jump table:
627 static void *array[] = @{ &&foo, &&bar, &&hack @};
630 Then you can select a label with indexing, like this:
637 Note that this does not check whether the subscript is in bounds---array
638 indexing in C never does that.
640 Such an array of label values serves a purpose much like that of the
641 @code{switch} statement. The @code{switch} statement is cleaner, so
642 use that rather than an array unless the problem does not fit a
643 @code{switch} statement very well.
645 Another use of label values is in an interpreter for threaded code.
646 The labels within the interpreter function can be stored in the
647 threaded code for super-fast dispatching.
649 You may not use this mechanism to jump to code in a different function.
650 If you do that, totally unpredictable things will happen. The best way to
651 avoid this is to store the label address only in automatic variables and
652 never pass it as an argument.
654 An alternate way to write the above example is
657 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
659 goto *(&&foo + array[i]);
663 This is more friendly to code living in shared libraries, as it reduces
664 the number of dynamic relocations that are needed, and by consequence,
665 allows the data to be read-only.
667 @node Nested Functions
668 @section Nested Functions
669 @cindex nested functions
670 @cindex downward funargs
673 A @dfn{nested function} is a function defined inside another function.
674 (Nested functions are not supported for GNU C++.) The nested function's
675 name is local to the block where it is defined. For example, here we
676 define a nested function named @code{square}, and call it twice:
680 foo (double a, double b)
682 double square (double z) @{ return z * z; @}
684 return square (a) + square (b);
689 The nested function can access all the variables of the containing
690 function that are visible at the point of its definition. This is
691 called @dfn{lexical scoping}. For example, here we show a nested
692 function which uses an inherited variable named @code{offset}:
696 bar (int *array, int offset, int size)
698 int access (int *array, int index)
699 @{ return array[index + offset]; @}
702 for (i = 0; i < size; i++)
703 @dots{} access (array, i) @dots{}
708 Nested function definitions are permitted within functions in the places
709 where variable definitions are allowed; that is, in any block, before
710 the first statement in the block.
712 It is possible to call the nested function from outside the scope of its
713 name by storing its address or passing the address to another function:
716 hack (int *array, int size)
718 void store (int index, int value)
719 @{ array[index] = value; @}
721 intermediate (store, size);
725 Here, the function @code{intermediate} receives the address of
726 @code{store} as an argument. If @code{intermediate} calls @code{store},
727 the arguments given to @code{store} are used to store into @code{array}.
728 But this technique works only so long as the containing function
729 (@code{hack}, in this example) does not exit.
731 If you try to call the nested function through its address after the
732 containing function has exited, all hell will break loose. If you try
733 to call it after a containing scope level has exited, and if it refers
734 to some of the variables that are no longer in scope, you may be lucky,
735 but it's not wise to take the risk. If, however, the nested function
736 does not refer to anything that has gone out of scope, you should be
739 GCC implements taking the address of a nested function using a technique
740 called @dfn{trampolines}. A paper describing them is available as
743 @uref{http://people.debian.org/~karlheg/Usenix88-lexic.pdf}.
745 A nested function can jump to a label inherited from a containing
746 function, provided the label was explicitly declared in the containing
747 function (@pxref{Local Labels}). Such a jump returns instantly to the
748 containing function, exiting the nested function which did the
749 @code{goto} and any intermediate functions as well. Here is an example:
753 bar (int *array, int offset, int size)
756 int access (int *array, int index)
760 return array[index + offset];
764 for (i = 0; i < size; i++)
765 @dots{} access (array, i) @dots{}
769 /* @r{Control comes here from @code{access}
770 if it detects an error.} */
777 A nested function always has internal linkage. Declaring one with
778 @code{extern} is erroneous. If you need to declare the nested function
779 before its definition, use @code{auto} (which is otherwise meaningless
780 for function declarations).
783 bar (int *array, int offset, int size)
786 auto int access (int *, int);
788 int access (int *array, int index)
792 return array[index + offset];
798 @node Constructing Calls
799 @section Constructing Function Calls
800 @cindex constructing calls
801 @cindex forwarding calls
803 Using the built-in functions described below, you can record
804 the arguments a function received, and call another function
805 with the same arguments, without knowing the number or types
808 You can also record the return value of that function call,
809 and later return that value, without knowing what data type
810 the function tried to return (as long as your caller expects
813 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
814 This built-in function returns a pointer to data
815 describing how to perform a call with the same arguments as were passed
816 to the current function.
818 The function saves the arg pointer register, structure value address,
819 and all registers that might be used to pass arguments to a function
820 into a block of memory allocated on the stack. Then it returns the
821 address of that block.
824 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
825 This built-in function invokes @var{function}
826 with a copy of the parameters described by @var{arguments}
829 The value of @var{arguments} should be the value returned by
830 @code{__builtin_apply_args}. The argument @var{size} specifies the size
831 of the stack argument data, in bytes.
833 This function returns a pointer to data describing
834 how to return whatever value was returned by @var{function}. The data
835 is saved in a block of memory allocated on the stack.
837 It is not always simple to compute the proper value for @var{size}. The
838 value is used by @code{__builtin_apply} to compute the amount of data
839 that should be pushed on the stack and copied from the incoming argument
843 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
844 This built-in function returns the value described by @var{result} from
845 the containing function. You should specify, for @var{result}, a value
846 returned by @code{__builtin_apply}.
850 @section Naming an Expression's Type
853 You can give a name to the type of an expression using a @code{typedef}
854 declaration with an initializer. Here is how to define @var{name} as a
855 type name for the type of @var{exp}:
858 typedef @var{name} = @var{exp};
861 This is useful in conjunction with the statements-within-expressions
862 feature. Here is how the two together can be used to define a safe
863 ``maximum'' macro that operates on any arithmetic type:
867 (@{typedef _ta = (a), _tb = (b); \
868 _ta _a = (a); _tb _b = (b); \
869 _a > _b ? _a : _b; @})
872 @cindex underscores in variables in macros
873 @cindex @samp{_} in variables in macros
874 @cindex local variables in macros
875 @cindex variables, local, in macros
876 @cindex macros, local variables in
878 The reason for using names that start with underscores for the local
879 variables is to avoid conflicts with variable names that occur within the
880 expressions that are substituted for @code{a} and @code{b}. Eventually we
881 hope to design a new form of declaration syntax that allows you to declare
882 variables whose scopes start only after their initializers; this will be a
883 more reliable way to prevent such conflicts.
886 @section Referring to a Type with @code{typeof}
889 @cindex macros, types of arguments
891 Another way to refer to the type of an expression is with @code{typeof}.
892 The syntax of using of this keyword looks like @code{sizeof}, but the
893 construct acts semantically like a type name defined with @code{typedef}.
895 There are two ways of writing the argument to @code{typeof}: with an
896 expression or with a type. Here is an example with an expression:
903 This assumes that @code{x} is an array of pointers to functions;
904 the type described is that of the values of the functions.
906 Here is an example with a typename as the argument:
913 Here the type described is that of pointers to @code{int}.
915 If you are writing a header file that must work when included in ISO C
916 programs, write @code{__typeof__} instead of @code{typeof}.
917 @xref{Alternate Keywords}.
919 A @code{typeof}-construct can be used anywhere a typedef name could be
920 used. For example, you can use it in a declaration, in a cast, or inside
921 of @code{sizeof} or @code{typeof}.
925 This declares @code{y} with the type of what @code{x} points to.
932 This declares @code{y} as an array of such values.
939 This declares @code{y} as an array of pointers to characters:
942 typeof (typeof (char *)[4]) y;
946 It is equivalent to the following traditional C declaration:
952 To see the meaning of the declaration using @code{typeof}, and why it
953 might be a useful way to write, let's rewrite it with these macros:
956 #define pointer(T) typeof(T *)
957 #define array(T, N) typeof(T [N])
961 Now the declaration can be rewritten this way:
964 array (pointer (char), 4) y;
968 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
969 pointers to @code{char}.
973 @section Generalized Lvalues
974 @cindex compound expressions as lvalues
975 @cindex expressions, compound, as lvalues
976 @cindex conditional expressions as lvalues
977 @cindex expressions, conditional, as lvalues
978 @cindex casts as lvalues
979 @cindex generalized lvalues
980 @cindex lvalues, generalized
981 @cindex extensions, @code{?:}
982 @cindex @code{?:} extensions
983 Compound expressions, conditional expressions and casts are allowed as
984 lvalues provided their operands are lvalues. This means that you can take
985 their addresses or store values into them.
987 Standard C++ allows compound expressions and conditional expressions as
988 lvalues, and permits casts to reference type, so use of this extension
989 is deprecated for C++ code.
991 For example, a compound expression can be assigned, provided the last
992 expression in the sequence is an lvalue. These two expressions are
1000 Similarly, the address of the compound expression can be taken. These two
1001 expressions are equivalent:
1008 A conditional expression is a valid lvalue if its type is not void and the
1009 true and false branches are both valid lvalues. For example, these two
1010 expressions are equivalent:
1014 (a ? b = 5 : (c = 5))
1017 A cast is a valid lvalue if its operand is an lvalue. A simple
1018 assignment whose left-hand side is a cast works by converting the
1019 right-hand side first to the specified type, then to the type of the
1020 inner left-hand side expression. After this is stored, the value is
1021 converted back to the specified type to become the value of the
1022 assignment. Thus, if @code{a} has type @code{char *}, the following two
1023 expressions are equivalent:
1027 (int)(a = (char *)(int)5)
1030 An assignment-with-arithmetic operation such as @samp{+=} applied to a cast
1031 performs the arithmetic using the type resulting from the cast, and then
1032 continues as in the previous case. Therefore, these two expressions are
1037 (int)(a = (char *)(int) ((int)a + 5))
1040 You cannot take the address of an lvalue cast, because the use of its
1041 address would not work out coherently. Suppose that @code{&(int)f} were
1042 permitted, where @code{f} has type @code{float}. Then the following
1043 statement would try to store an integer bit-pattern where a floating
1044 point number belongs:
1050 This is quite different from what @code{(int)f = 1} would do---that
1051 would convert 1 to floating point and store it. Rather than cause this
1052 inconsistency, we think it is better to prohibit use of @samp{&} on a cast.
1054 If you really do want an @code{int *} pointer with the address of
1055 @code{f}, you can simply write @code{(int *)&f}.
1058 @section Conditionals with Omitted Operands
1059 @cindex conditional expressions, extensions
1060 @cindex omitted middle-operands
1061 @cindex middle-operands, omitted
1062 @cindex extensions, @code{?:}
1063 @cindex @code{?:} extensions
1065 The middle operand in a conditional expression may be omitted. Then
1066 if the first operand is nonzero, its value is the value of the conditional
1069 Therefore, the expression
1076 has the value of @code{x} if that is nonzero; otherwise, the value of
1079 This example is perfectly equivalent to
1085 @cindex side effect in ?:
1086 @cindex ?: side effect
1088 In this simple case, the ability to omit the middle operand is not
1089 especially useful. When it becomes useful is when the first operand does,
1090 or may (if it is a macro argument), contain a side effect. Then repeating
1091 the operand in the middle would perform the side effect twice. Omitting
1092 the middle operand uses the value already computed without the undesirable
1093 effects of recomputing it.
1096 @section Double-Word Integers
1097 @cindex @code{long long} data types
1098 @cindex double-word arithmetic
1099 @cindex multiprecision arithmetic
1100 @cindex @code{LL} integer suffix
1101 @cindex @code{ULL} integer suffix
1103 ISO C99 supports data types for integers that are at least 64 bits wide,
1104 and as an extension GCC supports them in C89 mode and in C++.
1105 Simply write @code{long long int} for a signed integer, or
1106 @code{unsigned long long int} for an unsigned integer. To make an
1107 integer constant of type @code{long long int}, add the suffix @samp{LL}
1108 to the integer. To make an integer constant of type @code{unsigned long
1109 long int}, add the suffix @samp{ULL} to the integer.
1111 You can use these types in arithmetic like any other integer types.
1112 Addition, subtraction, and bitwise boolean operations on these types
1113 are open-coded on all types of machines. Multiplication is open-coded
1114 if the machine supports fullword-to-doubleword a widening multiply
1115 instruction. Division and shifts are open-coded only on machines that
1116 provide special support. The operations that are not open-coded use
1117 special library routines that come with GCC@.
1119 There may be pitfalls when you use @code{long long} types for function
1120 arguments, unless you declare function prototypes. If a function
1121 expects type @code{int} for its argument, and you pass a value of type
1122 @code{long long int}, confusion will result because the caller and the
1123 subroutine will disagree about the number of bytes for the argument.
1124 Likewise, if the function expects @code{long long int} and you pass
1125 @code{int}. The best way to avoid such problems is to use prototypes.
1128 @section Complex Numbers
1129 @cindex complex numbers
1130 @cindex @code{_Complex} keyword
1131 @cindex @code{__complex__} keyword
1133 ISO C99 supports complex floating data types, and as an extension GCC
1134 supports them in C89 mode and in C++, and supports complex integer data
1135 types which are not part of ISO C99. You can declare complex types
1136 using the keyword @code{_Complex}. As an extension, the older GNU
1137 keyword @code{__complex__} is also supported.
1139 For example, @samp{_Complex double x;} declares @code{x} as a
1140 variable whose real part and imaginary part are both of type
1141 @code{double}. @samp{_Complex short int y;} declares @code{y} to
1142 have real and imaginary parts of type @code{short int}; this is not
1143 likely to be useful, but it shows that the set of complex types is
1146 To write a constant with a complex data type, use the suffix @samp{i} or
1147 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
1148 has type @code{_Complex float} and @code{3i} has type
1149 @code{_Complex int}. Such a constant always has a pure imaginary
1150 value, but you can form any complex value you like by adding one to a
1151 real constant. This is a GNU extension; if you have an ISO C99
1152 conforming C library (such as GNU libc), and want to construct complex
1153 constants of floating type, you should include @code{<complex.h>} and
1154 use the macros @code{I} or @code{_Complex_I} instead.
1156 @cindex @code{__real__} keyword
1157 @cindex @code{__imag__} keyword
1158 To extract the real part of a complex-valued expression @var{exp}, write
1159 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
1160 extract the imaginary part. This is a GNU extension; for values of
1161 floating type, you should use the ISO C99 functions @code{crealf},
1162 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
1163 @code{cimagl}, declared in @code{<complex.h>} and also provided as
1164 built-in functions by GCC@.
1166 @cindex complex conjugation
1167 The operator @samp{~} performs complex conjugation when used on a value
1168 with a complex type. This is a GNU extension; for values of
1169 floating type, you should use the ISO C99 functions @code{conjf},
1170 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
1171 provided as built-in functions by GCC@.
1173 GCC can allocate complex automatic variables in a noncontiguous
1174 fashion; it's even possible for the real part to be in a register while
1175 the imaginary part is on the stack (or vice-versa). None of the
1176 supported debugging info formats has a way to represent noncontiguous
1177 allocation like this, so GCC describes a noncontiguous complex
1178 variable as if it were two separate variables of noncomplex type.
1179 If the variable's actual name is @code{foo}, the two fictitious
1180 variables are named @code{foo$real} and @code{foo$imag}. You can
1181 examine and set these two fictitious variables with your debugger.
1183 A future version of GDB will know how to recognize such pairs and treat
1184 them as a single variable with a complex type.
1190 ISO C99 supports floating-point numbers written not only in the usual
1191 decimal notation, such as @code{1.55e1}, but also numbers such as
1192 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
1193 supports this in C89 mode (except in some cases when strictly
1194 conforming) and in C++. In that format the
1195 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
1196 mandatory. The exponent is a decimal number that indicates the power of
1197 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
1204 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
1205 is the same as @code{1.55e1}.
1207 Unlike for floating-point numbers in the decimal notation the exponent
1208 is always required in the hexadecimal notation. Otherwise the compiler
1209 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
1210 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
1211 extension for floating-point constants of type @code{float}.
1214 @section Arrays of Length Zero
1215 @cindex arrays of length zero
1216 @cindex zero-length arrays
1217 @cindex length-zero arrays
1218 @cindex flexible array members
1220 Zero-length arrays are allowed in GNU C@. They are very useful as the
1221 last element of a structure which is really a header for a variable-length
1230 struct line *thisline = (struct line *)
1231 malloc (sizeof (struct line) + this_length);
1232 thisline->length = this_length;
1235 In ISO C89, you would have to give @code{contents} a length of 1, which
1236 means either you waste space or complicate the argument to @code{malloc}.
1238 In ISO C99, you would use a @dfn{flexible array member}, which is
1239 slightly different in syntax and semantics:
1243 Flexible array members are written as @code{contents[]} without
1247 Flexible array members have incomplete type, and so the @code{sizeof}
1248 operator may not be applied. As a quirk of the original implementation
1249 of zero-length arrays, @code{sizeof} evaluates to zero.
1252 Flexible array members may only appear as the last member of a
1253 @code{struct} that is otherwise non-empty.
1256 GCC versions before 3.0 allowed zero-length arrays to be statically
1257 initialized, as if they were flexible arrays. In addition to those
1258 cases that were useful, it also allowed initializations in situations
1259 that would corrupt later data. Non-empty initialization of zero-length
1260 arrays is now treated like any case where there are more initializer
1261 elements than the array holds, in that a suitable warning about "excess
1262 elements in array" is given, and the excess elements (all of them, in
1263 this case) are ignored.
1265 Instead GCC allows static initialization of flexible array members.
1266 This is equivalent to defining a new structure containing the original
1267 structure followed by an array of sufficient size to contain the data.
1268 I.e.@: in the following, @code{f1} is constructed as if it were declared
1274 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
1277 struct f1 f1; int data[3];
1278 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1282 The convenience of this extension is that @code{f1} has the desired
1283 type, eliminating the need to consistently refer to @code{f2.f1}.
1285 This has symmetry with normal static arrays, in that an array of
1286 unknown size is also written with @code{[]}.
1288 Of course, this extension only makes sense if the extra data comes at
1289 the end of a top-level object, as otherwise we would be overwriting
1290 data at subsequent offsets. To avoid undue complication and confusion
1291 with initialization of deeply nested arrays, we simply disallow any
1292 non-empty initialization except when the structure is the top-level
1293 object. For example:
1296 struct foo @{ int x; int y[]; @};
1297 struct bar @{ struct foo z; @};
1299 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1300 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1301 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1302 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1305 @node Variable Length
1306 @section Arrays of Variable Length
1307 @cindex variable-length arrays
1308 @cindex arrays of variable length
1311 Variable-length automatic arrays are allowed in ISO C99, and as an
1312 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1313 implementation of variable-length arrays does not yet conform in detail
1314 to the ISO C99 standard.) These arrays are
1315 declared like any other automatic arrays, but with a length that is not
1316 a constant expression. The storage is allocated at the point of
1317 declaration and deallocated when the brace-level is exited. For
1322 concat_fopen (char *s1, char *s2, char *mode)
1324 char str[strlen (s1) + strlen (s2) + 1];
1327 return fopen (str, mode);
1331 @cindex scope of a variable length array
1332 @cindex variable-length array scope
1333 @cindex deallocating variable length arrays
1334 Jumping or breaking out of the scope of the array name deallocates the
1335 storage. Jumping into the scope is not allowed; you get an error
1338 @cindex @code{alloca} vs variable-length arrays
1339 You can use the function @code{alloca} to get an effect much like
1340 variable-length arrays. The function @code{alloca} is available in
1341 many other C implementations (but not in all). On the other hand,
1342 variable-length arrays are more elegant.
1344 There are other differences between these two methods. Space allocated
1345 with @code{alloca} exists until the containing @emph{function} returns.
1346 The space for a variable-length array is deallocated as soon as the array
1347 name's scope ends. (If you use both variable-length arrays and
1348 @code{alloca} in the same function, deallocation of a variable-length array
1349 will also deallocate anything more recently allocated with @code{alloca}.)
1351 You can also use variable-length arrays as arguments to functions:
1355 tester (int len, char data[len][len])
1361 The length of an array is computed once when the storage is allocated
1362 and is remembered for the scope of the array in case you access it with
1365 If you want to pass the array first and the length afterward, you can
1366 use a forward declaration in the parameter list---another GNU extension.
1370 tester (int len; char data[len][len], int len)
1376 @cindex parameter forward declaration
1377 The @samp{int len} before the semicolon is a @dfn{parameter forward
1378 declaration}, and it serves the purpose of making the name @code{len}
1379 known when the declaration of @code{data} is parsed.
1381 You can write any number of such parameter forward declarations in the
1382 parameter list. They can be separated by commas or semicolons, but the
1383 last one must end with a semicolon, which is followed by the ``real''
1384 parameter declarations. Each forward declaration must match a ``real''
1385 declaration in parameter name and data type. ISO C99 does not support
1386 parameter forward declarations.
1388 @node Variadic Macros
1389 @section Macros with a Variable Number of Arguments.
1390 @cindex variable number of arguments
1391 @cindex macro with variable arguments
1392 @cindex rest argument (in macro)
1393 @cindex variadic macros
1395 In the ISO C standard of 1999, a macro can be declared to accept a
1396 variable number of arguments much as a function can. The syntax for
1397 defining the macro is similar to that of a function. Here is an
1401 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1404 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1405 such a macro, it represents the zero or more tokens until the closing
1406 parenthesis that ends the invocation, including any commas. This set of
1407 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1408 wherever it appears. See the CPP manual for more information.
1410 GCC has long supported variadic macros, and used a different syntax that
1411 allowed you to give a name to the variable arguments just like any other
1412 argument. Here is an example:
1415 #define debug(format, args...) fprintf (stderr, format, args)
1418 This is in all ways equivalent to the ISO C example above, but arguably
1419 more readable and descriptive.
1421 GNU CPP has two further variadic macro extensions, and permits them to
1422 be used with either of the above forms of macro definition.
1424 In standard C, you are not allowed to leave the variable argument out
1425 entirely; but you are allowed to pass an empty argument. For example,
1426 this invocation is invalid in ISO C, because there is no comma after
1433 GNU CPP permits you to completely omit the variable arguments in this
1434 way. In the above examples, the compiler would complain, though since
1435 the expansion of the macro still has the extra comma after the format
1438 To help solve this problem, CPP behaves specially for variable arguments
1439 used with the token paste operator, @samp{##}. If instead you write
1442 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1445 and if the variable arguments are omitted or empty, the @samp{##}
1446 operator causes the preprocessor to remove the comma before it. If you
1447 do provide some variable arguments in your macro invocation, GNU CPP
1448 does not complain about the paste operation and instead places the
1449 variable arguments after the comma. Just like any other pasted macro
1450 argument, these arguments are not macro expanded.
1452 @node Escaped Newlines
1453 @section Slightly Looser Rules for Escaped Newlines
1454 @cindex escaped newlines
1455 @cindex newlines (escaped)
1457 Recently, the preprocessor has relaxed its treatment of escaped
1458 newlines. Previously, the newline had to immediately follow a
1459 backslash. The current implementation allows whitespace in the form of
1460 spaces, horizontal and vertical tabs, and form feeds between the
1461 backslash and the subsequent newline. The preprocessor issues a
1462 warning, but treats it as a valid escaped newline and combines the two
1463 lines to form a single logical line. This works within comments and
1464 tokens, including multi-line strings, as well as between tokens.
1465 Comments are @emph{not} treated as whitespace for the purposes of this
1466 relaxation, since they have not yet been replaced with spaces.
1468 @node Multi-line Strings
1469 @section String Literals with Embedded Newlines
1470 @cindex multi-line string literals
1472 As an extension, GNU CPP permits string literals to cross multiple lines
1473 without escaping the embedded newlines. Each embedded newline is
1474 replaced with a single @samp{\n} character in the resulting string
1475 literal, regardless of what form the newline took originally.
1477 CPP currently allows such strings in directives as well (other than the
1478 @samp{#include} family). This is deprecated and will eventually be
1482 @section Non-Lvalue Arrays May Have Subscripts
1483 @cindex subscripting
1484 @cindex arrays, non-lvalue
1486 @cindex subscripting and function values
1487 In ISO C99, arrays that are not lvalues still decay to pointers, and
1488 may be subscripted, although they may not be modified or used after
1489 the next sequence point and the unary @samp{&} operator may not be
1490 applied to them. As an extension, GCC allows such arrays to be
1491 subscripted in C89 mode, though otherwise they do not decay to
1492 pointers outside C99 mode. For example,
1493 this is valid in GNU C though not valid in C89:
1497 struct foo @{int a[4];@};
1503 return f().a[index];
1509 @section Arithmetic on @code{void}- and Function-Pointers
1510 @cindex void pointers, arithmetic
1511 @cindex void, size of pointer to
1512 @cindex function pointers, arithmetic
1513 @cindex function, size of pointer to
1515 In GNU C, addition and subtraction operations are supported on pointers to
1516 @code{void} and on pointers to functions. This is done by treating the
1517 size of a @code{void} or of a function as 1.
1519 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1520 and on function types, and returns 1.
1522 @opindex Wpointer-arith
1523 The option @option{-Wpointer-arith} requests a warning if these extensions
1527 @section Non-Constant Initializers
1528 @cindex initializers, non-constant
1529 @cindex non-constant initializers
1531 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1532 automatic variable are not required to be constant expressions in GNU C@.
1533 Here is an example of an initializer with run-time varying elements:
1536 foo (float f, float g)
1538 float beat_freqs[2] = @{ f-g, f+g @};
1543 @node Compound Literals
1544 @section Compound Literals
1545 @cindex constructor expressions
1546 @cindex initializations in expressions
1547 @cindex structures, constructor expression
1548 @cindex expressions, constructor
1549 @cindex compound literals
1550 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1552 ISO C99 supports compound literals. A compound literal looks like
1553 a cast containing an initializer. Its value is an object of the
1554 type specified in the cast, containing the elements specified in
1555 the initializer; it is an lvalue. As an extension, GCC supports
1556 compound literals in C89 mode and in C++.
1558 Usually, the specified type is a structure. Assume that
1559 @code{struct foo} and @code{structure} are declared as shown:
1562 struct foo @{int a; char b[2];@} structure;
1566 Here is an example of constructing a @code{struct foo} with a compound literal:
1569 structure = ((struct foo) @{x + y, 'a', 0@});
1573 This is equivalent to writing the following:
1577 struct foo temp = @{x + y, 'a', 0@};
1582 You can also construct an array. If all the elements of the compound literal
1583 are (made up of) simple constant expressions, suitable for use in
1584 initializers of objects of static storage duration, then the compound
1585 literal can be coerced to a pointer to its first element and used in
1586 such an initializer, as shown here:
1589 char **foo = (char *[]) @{ "x", "y", "z" @};
1592 Compound literals for scalar types and union types are is
1593 also allowed, but then the compound literal is equivalent
1596 As a GNU extension, GCC allows initialization of objects with static storage
1597 duration by compound literals (which is not possible in ISO C99, because
1598 the initializer is not a constant).
1599 It is handled as if the object was initialized only with the bracket
1600 enclosed list if compound literal's and object types match.
1601 The initializer list of the compound literal must be constant.
1602 If the object being initialized has array type of unknown size, the size is
1603 determined by compound literal size.
1606 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1607 static int y[] = (int []) @{1, 2, 3@};
1608 static int z[] = (int [3]) @{1@};
1612 The above lines are equivalent to the following:
1614 static struct foo x = @{1, 'a', 'b'@};
1615 static int y[] = @{1, 2, 3@};
1616 static int z[] = @{1, 0, 0@};
1619 @node Designated Inits
1620 @section Designated Initializers
1621 @cindex initializers with labeled elements
1622 @cindex labeled elements in initializers
1623 @cindex case labels in initializers
1624 @cindex designated initializers
1626 Standard C89 requires the elements of an initializer to appear in a fixed
1627 order, the same as the order of the elements in the array or structure
1630 In ISO C99 you can give the elements in any order, specifying the array
1631 indices or structure field names they apply to, and GNU C allows this as
1632 an extension in C89 mode as well. This extension is not
1633 implemented in GNU C++.
1635 To specify an array index, write
1636 @samp{[@var{index}] =} before the element value. For example,
1639 int a[6] = @{ [4] = 29, [2] = 15 @};
1646 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1650 The index values must be constant expressions, even if the array being
1651 initialized is automatic.
1653 An alternative syntax for this which has been obsolete since GCC 2.5 but
1654 GCC still accepts is to write @samp{[@var{index}]} before the element
1655 value, with no @samp{=}.
1657 To initialize a range of elements to the same value, write
1658 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1659 extension. For example,
1662 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1666 If the value in it has side-effects, the side-effects will happen only once,
1667 not for each initialized field by the range initializer.
1670 Note that the length of the array is the highest value specified
1673 In a structure initializer, specify the name of a field to initialize
1674 with @samp{.@var{fieldname} =} before the element value. For example,
1675 given the following structure,
1678 struct point @{ int x, y; @};
1682 the following initialization
1685 struct point p = @{ .y = yvalue, .x = xvalue @};
1692 struct point p = @{ xvalue, yvalue @};
1695 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1696 @samp{@var{fieldname}:}, as shown here:
1699 struct point p = @{ y: yvalue, x: xvalue @};
1703 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1704 @dfn{designator}. You can also use a designator (or the obsolete colon
1705 syntax) when initializing a union, to specify which element of the union
1706 should be used. For example,
1709 union foo @{ int i; double d; @};
1711 union foo f = @{ .d = 4 @};
1715 will convert 4 to a @code{double} to store it in the union using
1716 the second element. By contrast, casting 4 to type @code{union foo}
1717 would store it into the union as the integer @code{i}, since it is
1718 an integer. (@xref{Cast to Union}.)
1720 You can combine this technique of naming elements with ordinary C
1721 initialization of successive elements. Each initializer element that
1722 does not have a designator applies to the next consecutive element of the
1723 array or structure. For example,
1726 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1733 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1736 Labeling the elements of an array initializer is especially useful
1737 when the indices are characters or belong to an @code{enum} type.
1742 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1743 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1746 @cindex designator lists
1747 You can also write a series of @samp{.@var{fieldname}} and
1748 @samp{[@var{index}]} designators before an @samp{=} to specify a
1749 nested subobject to initialize; the list is taken relative to the
1750 subobject corresponding to the closest surrounding brace pair. For
1751 example, with the @samp{struct point} declaration above:
1754 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1758 If the same field is initialized multiple times, it will have value from
1759 the last initialization. If any such overridden initialization has
1760 side-effect, it is unspecified whether the side-effect happens or not.
1761 Currently, gcc will discard them and issue a warning.
1764 @section Case Ranges
1766 @cindex ranges in case statements
1768 You can specify a range of consecutive values in a single @code{case} label,
1772 case @var{low} ... @var{high}:
1776 This has the same effect as the proper number of individual @code{case}
1777 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1779 This feature is especially useful for ranges of ASCII character codes:
1785 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1786 it may be parsed wrong when you use it with integer values. For example,
1801 @section Cast to a Union Type
1802 @cindex cast to a union
1803 @cindex union, casting to a
1805 A cast to union type is similar to other casts, except that the type
1806 specified is a union type. You can specify the type either with
1807 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1808 a constructor though, not a cast, and hence does not yield an lvalue like
1809 normal casts. (@xref{Compound Literals}.)
1811 The types that may be cast to the union type are those of the members
1812 of the union. Thus, given the following union and variables:
1815 union foo @{ int i; double d; @};
1821 both @code{x} and @code{y} can be cast to type @code{union foo}.
1823 Using the cast as the right-hand side of an assignment to a variable of
1824 union type is equivalent to storing in a member of the union:
1829 u = (union foo) x @equiv{} u.i = x
1830 u = (union foo) y @equiv{} u.d = y
1833 You can also use the union cast as a function argument:
1836 void hack (union foo);
1838 hack ((union foo) x);
1841 @node Mixed Declarations
1842 @section Mixed Declarations and Code
1843 @cindex mixed declarations and code
1844 @cindex declarations, mixed with code
1845 @cindex code, mixed with declarations
1847 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1848 within compound statements. As an extension, GCC also allows this in
1849 C89 mode. For example, you could do:
1858 Each identifier is visible from where it is declared until the end of
1859 the enclosing block.
1861 @node Function Attributes
1862 @section Declaring Attributes of Functions
1863 @cindex function attributes
1864 @cindex declaring attributes of functions
1865 @cindex functions that never return
1866 @cindex functions that have no side effects
1867 @cindex functions in arbitrary sections
1868 @cindex functions that behave like malloc
1869 @cindex @code{volatile} applied to function
1870 @cindex @code{const} applied to function
1871 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1872 @cindex functions that are passed arguments in registers on the 386
1873 @cindex functions that pop the argument stack on the 386
1874 @cindex functions that do not pop the argument stack on the 386
1876 In GNU C, you declare certain things about functions called in your program
1877 which help the compiler optimize function calls and check your code more
1880 The keyword @code{__attribute__} allows you to specify special
1881 attributes when making a declaration. This keyword is followed by an
1882 attribute specification inside double parentheses. The following
1883 attributes are currently defined for functions on all targets:
1884 @code{noreturn}, @code{noinline}, @code{always_inline},
1885 @code{pure}, @code{const},
1886 @code{format}, @code{format_arg}, @code{no_instrument_function},
1887 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1888 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc}, and
1889 @code{alias}. Several other attributes are defined for functions on
1890 particular target systems. Other attributes, including @code{section}
1891 are supported for variables declarations (@pxref{Variable Attributes})
1892 and for types (@pxref{Type Attributes}).
1894 You may also specify attributes with @samp{__} preceding and following
1895 each keyword. This allows you to use them in header files without
1896 being concerned about a possible macro of the same name. For example,
1897 you may use @code{__noreturn__} instead of @code{noreturn}.
1899 @xref{Attribute Syntax}, for details of the exact syntax for using
1903 @cindex @code{noreturn} function attribute
1905 A few standard library functions, such as @code{abort} and @code{exit},
1906 cannot return. GCC knows this automatically. Some programs define
1907 their own functions that never return. You can declare them
1908 @code{noreturn} to tell the compiler this fact. For example,
1912 void fatal () __attribute__ ((noreturn));
1917 @dots{} /* @r{Print error message.} */ @dots{}
1923 The @code{noreturn} keyword tells the compiler to assume that
1924 @code{fatal} cannot return. It can then optimize without regard to what
1925 would happen if @code{fatal} ever did return. This makes slightly
1926 better code. More importantly, it helps avoid spurious warnings of
1927 uninitialized variables.
1929 Do not assume that registers saved by the calling function are
1930 restored before calling the @code{noreturn} function.
1932 It does not make sense for a @code{noreturn} function to have a return
1933 type other than @code{void}.
1935 The attribute @code{noreturn} is not implemented in GCC versions
1936 earlier than 2.5. An alternative way to declare that a function does
1937 not return, which works in the current version and in some older
1938 versions, is as follows:
1941 typedef void voidfn ();
1943 volatile voidfn fatal;
1946 @cindex @code{noinline} function attribute
1948 This function attribute prevents a function from being considered for
1951 @cindex @code{always_inline} function attribute
1953 Generally, functions are not inlined unless optimization is specified.
1954 For functions declared inline, this attribute inlines the function even
1955 if no optimization level was specified.
1957 @cindex @code{pure} function attribute
1959 Many functions have no effects except the return value and their
1960 return value depends only on the parameters and/or global variables.
1961 Such a function can be subject
1962 to common subexpression elimination and loop optimization just as an
1963 arithmetic operator would be. These functions should be declared
1964 with the attribute @code{pure}. For example,
1967 int square (int) __attribute__ ((pure));
1971 says that the hypothetical function @code{square} is safe to call
1972 fewer times than the program says.
1974 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
1975 Interesting non-pure functions are functions with infinite loops or those
1976 depending on volatile memory or other system resource, that may change between
1977 two consecutive calls (such as @code{feof} in a multithreading environment).
1979 The attribute @code{pure} is not implemented in GCC versions earlier
1981 @cindex @code{const} function attribute
1983 Many functions do not examine any values except their arguments, and
1984 have no effects except the return value. Basically this is just slightly
1985 more strict class than the @code{pure} attribute above, since function is not
1986 allowed to read global memory.
1988 @cindex pointer arguments
1989 Note that a function that has pointer arguments and examines the data
1990 pointed to must @emph{not} be declared @code{const}. Likewise, a
1991 function that calls a non-@code{const} function usually must not be
1992 @code{const}. It does not make sense for a @code{const} function to
1995 The attribute @code{const} is not implemented in GCC versions earlier
1996 than 2.5. An alternative way to declare that a function has no side
1997 effects, which works in the current version and in some older versions,
2001 typedef int intfn ();
2003 extern const intfn square;
2006 This approach does not work in GNU C++ from 2.6.0 on, since the language
2007 specifies that the @samp{const} must be attached to the return value.
2010 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2011 @cindex @code{format} function attribute
2013 The @code{format} attribute specifies that a function takes @code{printf},
2014 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
2015 should be type-checked against a format string. For example, the
2020 my_printf (void *my_object, const char *my_format, ...)
2021 __attribute__ ((format (printf, 2, 3)));
2025 causes the compiler to check the arguments in calls to @code{my_printf}
2026 for consistency with the @code{printf} style format string argument
2029 The parameter @var{archetype} determines how the format string is
2030 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
2031 or @code{strfmon}. (You can also use @code{__printf__},
2032 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
2033 parameter @var{string-index} specifies which argument is the format
2034 string argument (starting from 1), while @var{first-to-check} is the
2035 number of the first argument to check against the format string. For
2036 functions where the arguments are not available to be checked (such as
2037 @code{vprintf}), specify the third parameter as zero. In this case the
2038 compiler only checks the format string for consistency. For
2039 @code{strftime} formats, the third parameter is required to be zero.
2041 In the example above, the format string (@code{my_format}) is the second
2042 argument of the function @code{my_print}, and the arguments to check
2043 start with the third argument, so the correct parameters for the format
2044 attribute are 2 and 3.
2046 @opindex ffreestanding
2047 The @code{format} attribute allows you to identify your own functions
2048 which take format strings as arguments, so that GCC can check the
2049 calls to these functions for errors. The compiler always (unless
2050 @option{-ffreestanding} is used) checks formats
2051 for the standard library functions @code{printf}, @code{fprintf},
2052 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2053 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2054 warnings are requested (using @option{-Wformat}), so there is no need to
2055 modify the header file @file{stdio.h}. In C99 mode, the functions
2056 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2057 @code{vsscanf} are also checked. Except in strictly conforming C
2058 standard modes, the X/Open function @code{strfmon} is also checked as
2059 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2060 @xref{C Dialect Options,,Options Controlling C Dialect}.
2062 @item format_arg (@var{string-index})
2063 @cindex @code{format_arg} function attribute
2064 @opindex Wformat-nonliteral
2065 The @code{format_arg} attribute specifies that a function takes a format
2066 string for a @code{printf}, @code{scanf}, @code{strftime} or
2067 @code{strfmon} style function and modifies it (for example, to translate
2068 it into another language), so the result can be passed to a
2069 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2070 function (with the remaining arguments to the format function the same
2071 as they would have been for the unmodified string). For example, the
2076 my_dgettext (char *my_domain, const char *my_format)
2077 __attribute__ ((format_arg (2)));
2081 causes the compiler to check the arguments in calls to a @code{printf},
2082 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2083 format string argument is a call to the @code{my_dgettext} function, for
2084 consistency with the format string argument @code{my_format}. If the
2085 @code{format_arg} attribute had not been specified, all the compiler
2086 could tell in such calls to format functions would be that the format
2087 string argument is not constant; this would generate a warning when
2088 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2089 without the attribute.
2091 The parameter @var{string-index} specifies which argument is the format
2092 string argument (starting from 1).
2094 The @code{format-arg} attribute allows you to identify your own
2095 functions which modify format strings, so that GCC can check the
2096 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2097 type function whose operands are a call to one of your own function.
2098 The compiler always treats @code{gettext}, @code{dgettext}, and
2099 @code{dcgettext} in this manner except when strict ISO C support is
2100 requested by @option{-ansi} or an appropriate @option{-std} option, or
2101 @option{-ffreestanding} is used. @xref{C Dialect Options,,Options
2102 Controlling C Dialect}.
2104 @item no_instrument_function
2105 @cindex @code{no_instrument_function} function attribute
2106 @opindex finstrument-functions
2107 If @option{-finstrument-functions} is given, profiling function calls will
2108 be generated at entry and exit of most user-compiled functions.
2109 Functions with this attribute will not be so instrumented.
2111 @item section ("@var{section-name}")
2112 @cindex @code{section} function attribute
2113 Normally, the compiler places the code it generates in the @code{text} section.
2114 Sometimes, however, you need additional sections, or you need certain
2115 particular functions to appear in special sections. The @code{section}
2116 attribute specifies that a function lives in a particular section.
2117 For example, the declaration:
2120 extern void foobar (void) __attribute__ ((section ("bar")));
2124 puts the function @code{foobar} in the @code{bar} section.
2126 Some file formats do not support arbitrary sections so the @code{section}
2127 attribute is not available on all platforms.
2128 If you need to map the entire contents of a module to a particular
2129 section, consider using the facilities of the linker instead.
2133 @cindex @code{constructor} function attribute
2134 @cindex @code{destructor} function attribute
2135 The @code{constructor} attribute causes the function to be called
2136 automatically before execution enters @code{main ()}. Similarly, the
2137 @code{destructor} attribute causes the function to be called
2138 automatically after @code{main ()} has completed or @code{exit ()} has
2139 been called. Functions with these attributes are useful for
2140 initializing data that will be used implicitly during the execution of
2143 These attributes are not currently implemented for Objective-C@.
2145 @cindex @code{unused} attribute.
2147 This attribute, attached to a function, means that the function is meant
2148 to be possibly unused. GCC will not produce a warning for this
2149 function. GNU C++ does not currently support this attribute as
2150 definitions without parameters are valid in C++.
2152 @cindex @code{used} attribute.
2154 This attribute, attached to a function, means that code must be emitted
2155 for the function even if it appears that the function is not referenced.
2156 This is useful, for example, when the function is referenced only in
2159 @cindex @code{deprecated} attribute.
2161 The @code{deprecated} attribute results in a warning if the function
2162 is used anywhere in the source file. This is useful when identifying
2163 functions that are expected to be removed in a future version of a
2164 program. The warning also includes the location of the declaration
2165 of the deprecated function, to enable users to easily find further
2166 information about why the function is deprecated, or what they should
2167 do instead. Note that the warnings only occurs for uses:
2170 int old_fn () __attribute__ ((deprecated));
2172 int (*fn_ptr)() = old_fn;
2175 results in a warning on line 3 but not line 2.
2177 The @code{deprecated} attribute can also be used for variables and
2178 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2181 @cindex @code{weak} attribute
2182 The @code{weak} attribute causes the declaration to be emitted as a weak
2183 symbol rather than a global. This is primarily useful in defining
2184 library functions which can be overridden in user code, though it can
2185 also be used with non-function declarations. Weak symbols are supported
2186 for ELF targets, and also for a.out targets when using the GNU assembler
2190 @cindex @code{malloc} attribute
2191 The @code{malloc} attribute is used to tell the compiler that a function
2192 may be treated as if it were the malloc function. The compiler assumes
2193 that calls to malloc result in a pointers that cannot alias anything.
2194 This will often improve optimization.
2196 @item alias ("@var{target}")
2197 @cindex @code{alias} attribute
2198 The @code{alias} attribute causes the declaration to be emitted as an
2199 alias for another symbol, which must be specified. For instance,
2202 void __f () @{ /* @r{Do something.} */; @}
2203 void f () __attribute__ ((weak, alias ("__f")));
2206 declares @samp{f} to be a weak alias for @samp{__f}. In C++, the
2207 mangled name for the target must be used.
2209 Not all target machines support this attribute.
2211 @item visibility ("@var{visibility_type}")
2212 @cindex @code{visibility} attribute
2213 The @code{visibility} attribute on ELF targets causes the declaration
2214 to be emitted with hidden, protected or internal visibility.
2217 void __attribute__ ((visibility ("protected")))
2218 f () @{ /* @r{Do something.} */; @}
2219 int i __attribute__ ((visibility ("hidden")));
2222 See the ELF gABI for complete details, but the short story is
2226 Hidden visibility indicates that the symbol will not be placed into
2227 the dynamic symbol table, so no other @dfn{module} (executable or
2228 shared library) can reference it directly.
2231 Protected visibility indicates that the symbol will be placed in the
2232 dynamic symbol table, but that references within the defining module
2233 will bind to the local symbol. That is, the symbol cannot be overridden
2237 Internal visibility is like hidden visibility, but with additional
2238 processor specific semantics. Unless otherwise specified by the psABI,
2239 gcc defines internal visibility to mean that the function is @emph{never}
2240 called from another module. Note that hidden symbols, while then cannot
2241 be referenced directly by other modules, can be referenced indirectly via
2242 function pointers. By indicating that a symbol cannot be called from
2243 outside the module, gcc may for instance omit the load of a PIC register
2244 since it is known that the calling function loaded the correct value.
2247 Not all ELF targets support this attribute.
2249 @item regparm (@var{number})
2250 @cindex functions that are passed arguments in registers on the 386
2251 On the Intel 386, the @code{regparm} attribute causes the compiler to
2252 pass up to @var{number} integer arguments in registers EAX,
2253 EDX, and ECX instead of on the stack. Functions that take a
2254 variable number of arguments will continue to be passed all of their
2255 arguments on the stack.
2258 @cindex functions that pop the argument stack on the 386
2259 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2260 assume that the called function will pop off the stack space used to
2261 pass arguments, unless it takes a variable number of arguments.
2263 The PowerPC compiler for Windows NT currently ignores the @code{stdcall}
2267 @cindex functions that do pop the argument stack on the 386
2269 On the Intel 386, the @code{cdecl} attribute causes the compiler to
2270 assume that the calling function will pop off the stack space used to
2271 pass arguments. This is
2272 useful to override the effects of the @option{-mrtd} switch.
2274 The PowerPC compiler for Windows NT currently ignores the @code{cdecl}
2277 @item longcall/shortcall
2278 @cindex functions called via pointer on the RS/6000 and PowerPC
2279 On the RS/6000 and PowerPC, the @code{longcall} attribute causes the
2280 compiler to always call this function via a pointer, just as it would if
2281 the @option{-mlongcall} option had been specified. The @code{shortcall}
2282 attribute causes the compiler not to do this. These attributes override
2283 both the @option{-mlongcall} switch and the @code{#pragma longcall}
2286 @xref{RS/6000 and PowerPC Options}, for more information on when long
2287 calls are and are not necessary.
2289 @item long_call/short_call
2290 @cindex indirect calls on ARM
2291 This attribute allows to specify how to call a particular function on
2292 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2293 command line switch and @code{#pragma long_calls} settings. The
2294 @code{long_call} attribute causes the compiler to always call the
2295 function by first loading its address into a register and then using the
2296 contents of that register. The @code{short_call} attribute always places
2297 the offset to the function from the call site into the @samp{BL}
2298 instruction directly.
2301 @cindex functions which are imported from a dll on PowerPC Windows NT
2302 On the PowerPC running Windows NT, the @code{dllimport} attribute causes
2303 the compiler to call the function via a global pointer to the function
2304 pointer that is set up by the Windows NT dll library. The pointer name
2305 is formed by combining @code{__imp_} and the function name.
2308 @cindex functions which are exported from a dll on PowerPC Windows NT
2309 On the PowerPC running Windows NT, the @code{dllexport} attribute causes
2310 the compiler to provide a global pointer to the function pointer, so
2311 that it can be called with the @code{dllimport} attribute. The pointer
2312 name is formed by combining @code{__imp_} and the function name.
2314 @item exception (@var{except-func} [, @var{except-arg}])
2315 @cindex functions which specify exception handling on PowerPC Windows NT
2316 On the PowerPC running Windows NT, the @code{exception} attribute causes
2317 the compiler to modify the structured exception table entry it emits for
2318 the declared function. The string or identifier @var{except-func} is
2319 placed in the third entry of the structured exception table. It
2320 represents a function, which is called by the exception handling
2321 mechanism if an exception occurs. If it was specified, the string or
2322 identifier @var{except-arg} is placed in the fourth entry of the
2323 structured exception table.
2325 @item function_vector
2326 @cindex calling functions through the function vector on the H8/300 processors
2327 Use this attribute on the H8/300 and H8/300H to indicate that the specified
2328 function should be called through the function vector. Calling a
2329 function through the function vector will reduce code size, however;
2330 the function vector has a limited size (maximum 128 entries on the H8/300
2331 and 64 entries on the H8/300H) and shares space with the interrupt vector.
2333 You must use GAS and GLD from GNU binutils version 2.7 or later for
2334 this attribute to work correctly.
2337 @cindex interrupt handler functions
2338 Use this attribute on the ARM, AVR, M32R/D and Xstormy16 ports to indicate
2339 that the specified function is an interrupt handler. The compiler will
2340 generate function entry and exit sequences suitable for use in an
2341 interrupt handler when this attribute is present.
2343 Note, interrupt handlers for the H8/300, H8/300H and SH processors can
2344 be specified via the @code{interrupt_handler} attribute.
2346 Note, on the AVR interrupts will be enabled inside the function.
2348 Note, for the ARM you can specify the kind of interrupt to be handled by
2349 adding an optional parameter to the interrupt attribute like this:
2352 void f () __attribute__ ((interrupt ("IRQ")));
2355 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2357 @item interrupt_handler
2358 @cindex interrupt handler functions on the H8/300 and SH processors
2359 Use this attribute on the H8/300, H8/300H and SH to indicate that the
2360 specified function is an interrupt handler. The compiler will generate
2361 function entry and exit sequences suitable for use in an interrupt
2362 handler when this attribute is present.
2365 Use this attribute on the SH to indicate an @code{interrupt_handler}
2366 function should switch to an alternate stack. It expects a string
2367 argument that names a global variable holding the address of the
2372 void f () __attribute__ ((interrupt_handler,
2373 sp_switch ("alt_stack")));
2377 Use this attribute on the SH for an @code{interrupt_handle} to return using
2378 @code{trapa} instead of @code{rte}. This attribute expects an integer
2379 argument specifying the trap number to be used.
2382 @cindex eight bit data on the H8/300 and H8/300H
2383 Use this attribute on the H8/300 and H8/300H to indicate that the specified
2384 variable should be placed into the eight bit data section.
2385 The compiler will generate more efficient code for certain operations
2386 on data in the eight bit data area. Note the eight bit data area is limited to
2389 You must use GAS and GLD from GNU binutils version 2.7 or later for
2390 this attribute to work correctly.
2393 @cindex tiny data section on the H8/300H
2394 Use this attribute on the H8/300H to indicate that the specified
2395 variable should be placed into the tiny data section.
2396 The compiler will generate more efficient code for loads and stores
2397 on data in the tiny data section. Note the tiny data area is limited to
2398 slightly under 32kbytes of data.
2401 @cindex signal handler functions on the AVR processors
2402 Use this attribute on the AVR to indicate that the specified
2403 function is an signal handler. The compiler will generate function
2404 entry and exit sequences suitable for use in an signal handler when this
2405 attribute is present. Interrupts will be disabled inside function.
2408 @cindex function without a prologue/epilogue code
2409 Use this attribute on the ARM or AVR ports to indicate that the specified
2410 function do not need prologue/epilogue sequences generated by the
2411 compiler. It is up to the programmer to provide these sequences.
2413 @item model (@var{model-name})
2414 @cindex function addressability on the M32R/D
2415 Use this attribute on the M32R/D to set the addressability of an object,
2416 and the code generated for a function.
2417 The identifier @var{model-name} is one of @code{small}, @code{medium},
2418 or @code{large}, representing each of the code models.
2420 Small model objects live in the lower 16MB of memory (so that their
2421 addresses can be loaded with the @code{ld24} instruction), and are
2422 callable with the @code{bl} instruction.
2424 Medium model objects may live anywhere in the 32-bit address space (the
2425 compiler will generate @code{seth/add3} instructions to load their addresses),
2426 and are callable with the @code{bl} instruction.
2428 Large model objects may live anywhere in the 32-bit address space (the
2429 compiler will generate @code{seth/add3} instructions to load their addresses),
2430 and may not be reachable with the @code{bl} instruction (the compiler will
2431 generate the much slower @code{seth/add3/jl} instruction sequence).
2435 You can specify multiple attributes in a declaration by separating them
2436 by commas within the double parentheses or by immediately following an
2437 attribute declaration with another attribute declaration.
2439 @cindex @code{#pragma}, reason for not using
2440 @cindex pragma, reason for not using
2441 Some people object to the @code{__attribute__} feature, suggesting that
2442 ISO C's @code{#pragma} should be used instead. At the time
2443 @code{__attribute__} was designed, there were two reasons for not doing
2448 It is impossible to generate @code{#pragma} commands from a macro.
2451 There is no telling what the same @code{#pragma} might mean in another
2455 These two reasons applied to almost any application that might have been
2456 proposed for @code{#pragma}. It was basically a mistake to use
2457 @code{#pragma} for @emph{anything}.
2459 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2460 to be generated from macros. In addition, a @code{#pragma GCC}
2461 namespace is now in use for GCC-specific pragmas. However, it has been
2462 found convenient to use @code{__attribute__} to achieve a natural
2463 attachment of attributes to their corresponding declarations, whereas
2464 @code{#pragma GCC} is of use for constructs that do not naturally form
2465 part of the grammar. @xref{Other Directives,,Miscellaneous
2466 Preprocessing Directives, cpp, The C Preprocessor}.
2468 @node Attribute Syntax
2469 @section Attribute Syntax
2470 @cindex attribute syntax
2472 This section describes the syntax with which @code{__attribute__} may be
2473 used, and the constructs to which attribute specifiers bind, for the C
2474 language. Some details may vary for C++ and Objective-C@. Because of
2475 infelicities in the grammar for attributes, some forms described here
2476 may not be successfully parsed in all cases.
2478 There are some problems with the semantics of attributes in C++. For
2479 example, there are no manglings for attributes, although they may affect
2480 code generation, so problems may arise when attributed types are used in
2481 conjunction with templates or overloading. Similarly, @code{typeid}
2482 does not distinguish between types with different attributes. Support
2483 for attributes in C++ may be restricted in future to attributes on
2484 declarations only, but not on nested declarators.
2486 @xref{Function Attributes}, for details of the semantics of attributes
2487 applying to functions. @xref{Variable Attributes}, for details of the
2488 semantics of attributes applying to variables. @xref{Type Attributes},
2489 for details of the semantics of attributes applying to structure, union
2490 and enumerated types.
2492 An @dfn{attribute specifier} is of the form
2493 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2494 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2495 each attribute is one of the following:
2499 Empty. Empty attributes are ignored.
2502 A word (which may be an identifier such as @code{unused}, or a reserved
2503 word such as @code{const}).
2506 A word, followed by, in parentheses, parameters for the attribute.
2507 These parameters take one of the following forms:
2511 An identifier. For example, @code{mode} attributes use this form.
2514 An identifier followed by a comma and a non-empty comma-separated list
2515 of expressions. For example, @code{format} attributes use this form.
2518 A possibly empty comma-separated list of expressions. For example,
2519 @code{format_arg} attributes use this form with the list being a single
2520 integer constant expression, and @code{alias} attributes use this form
2521 with the list being a single string constant.
2525 An @dfn{attribute specifier list} is a sequence of one or more attribute
2526 specifiers, not separated by any other tokens.
2528 An attribute specifier list may appear after the colon following a
2529 label, other than a @code{case} or @code{default} label. The only
2530 attribute it makes sense to use after a label is @code{unused}. This
2531 feature is intended for code generated by programs which contains labels
2532 that may be unused but which is compiled with @option{-Wall}. It would
2533 not normally be appropriate to use in it human-written code, though it
2534 could be useful in cases where the code that jumps to the label is
2535 contained within an @code{#ifdef} conditional.
2537 An attribute specifier list may appear as part of a @code{struct},
2538 @code{union} or @code{enum} specifier. It may go either immediately
2539 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2540 the closing brace. It is ignored if the content of the structure, union
2541 or enumerated type is not defined in the specifier in which the
2542 attribute specifier list is used---that is, in usages such as
2543 @code{struct __attribute__((foo)) bar} with no following opening brace.
2544 Where attribute specifiers follow the closing brace, they are considered
2545 to relate to the structure, union or enumerated type defined, not to any
2546 enclosing declaration the type specifier appears in, and the type
2547 defined is not complete until after the attribute specifiers.
2548 @c Otherwise, there would be the following problems: a shift/reduce
2549 @c conflict between attributes binding the struct/union/enum and
2550 @c binding to the list of specifiers/qualifiers; and "aligned"
2551 @c attributes could use sizeof for the structure, but the size could be
2552 @c changed later by "packed" attributes.
2554 Otherwise, an attribute specifier appears as part of a declaration,
2555 counting declarations of unnamed parameters and type names, and relates
2556 to that declaration (which may be nested in another declaration, for
2557 example in the case of a parameter declaration), or to a particular declarator
2558 within a declaration. Where an
2559 attribute specifier is applied to a parameter declared as a function or
2560 an array, it should apply to the function or array rather than the
2561 pointer to which the parameter is implicitly converted, but this is not
2562 yet correctly implemented.
2564 Any list of specifiers and qualifiers at the start of a declaration may
2565 contain attribute specifiers, whether or not such a list may in that
2566 context contain storage class specifiers. (Some attributes, however,
2567 are essentially in the nature of storage class specifiers, and only make
2568 sense where storage class specifiers may be used; for example,
2569 @code{section}.) There is one necessary limitation to this syntax: the
2570 first old-style parameter declaration in a function definition cannot
2571 begin with an attribute specifier, because such an attribute applies to
2572 the function instead by syntax described below (which, however, is not
2573 yet implemented in this case). In some other cases, attribute
2574 specifiers are permitted by this grammar but not yet supported by the
2575 compiler. All attribute specifiers in this place relate to the
2576 declaration as a whole. In the obsolescent usage where a type of
2577 @code{int} is implied by the absence of type specifiers, such a list of
2578 specifiers and qualifiers may be an attribute specifier list with no
2579 other specifiers or qualifiers.
2581 An attribute specifier list may appear immediately before a declarator
2582 (other than the first) in a comma-separated list of declarators in a
2583 declaration of more than one identifier using a single list of
2584 specifiers and qualifiers. Such attribute specifiers apply
2585 only to the identifier before whose declarator they appear. For
2589 __attribute__((noreturn)) void d0 (void),
2590 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2595 the @code{noreturn} attribute applies to all the functions
2596 declared; the @code{format} attribute only applies to @code{d1}.
2598 An attribute specifier list may appear immediately before the comma,
2599 @code{=} or semicolon terminating the declaration of an identifier other
2600 than a function definition. At present, such attribute specifiers apply
2601 to the declared object or function, but in future they may attach to the
2602 outermost adjacent declarator. In simple cases there is no difference,
2603 but, for example, in
2606 void (****f)(void) __attribute__((noreturn));
2610 at present the @code{noreturn} attribute applies to @code{f}, which
2611 causes a warning since @code{f} is not a function, but in future it may
2612 apply to the function @code{****f}. The precise semantics of what
2613 attributes in such cases will apply to are not yet specified. Where an
2614 assembler name for an object or function is specified (@pxref{Asm
2615 Labels}), at present the attribute must follow the @code{asm}
2616 specification; in future, attributes before the @code{asm} specification
2617 may apply to the adjacent declarator, and those after it to the declared
2620 An attribute specifier list may, in future, be permitted to appear after
2621 the declarator in a function definition (before any old-style parameter
2622 declarations or the function body).
2624 Attribute specifiers may be mixed with type qualifiers appearing inside
2625 the @code{[]} of a parameter array declarator, in the C99 construct by
2626 which such qualifiers are applied to the pointer to which the array is
2627 implicitly converted. Such attribute specifiers apply to the pointer,
2628 not to the array, but at present this is not implemented and they are
2631 An attribute specifier list may appear at the start of a nested
2632 declarator. At present, there are some limitations in this usage: the
2633 attributes correctly apply to the declarator, but for most individual
2634 attributes the semantics this implies are not implemented.
2635 When attribute specifiers follow the @code{*} of a pointer
2636 declarator, they may be mixed with any type qualifiers present.
2637 The following describes the formal semantics of this syntax. It will make the
2638 most sense if you are familiar with the formal specification of
2639 declarators in the ISO C standard.
2641 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2642 D1}, where @code{T} contains declaration specifiers that specify a type
2643 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2644 contains an identifier @var{ident}. The type specified for @var{ident}
2645 for derived declarators whose type does not include an attribute
2646 specifier is as in the ISO C standard.
2648 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2649 and the declaration @code{T D} specifies the type
2650 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2651 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2652 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2654 If @code{D1} has the form @code{*
2655 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2656 declaration @code{T D} specifies the type
2657 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2658 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2659 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2665 void (__attribute__((noreturn)) ****f) (void);
2669 specifies the type ``pointer to pointer to pointer to pointer to
2670 non-returning function returning @code{void}''. As another example,
2673 char *__attribute__((aligned(8))) *f;
2677 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2678 Note again that this does not work with most attributes; for example,
2679 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2680 is not yet supported.
2682 For compatibility with existing code written for compiler versions that
2683 did not implement attributes on nested declarators, some laxity is
2684 allowed in the placing of attributes. If an attribute that only applies
2685 to types is applied to a declaration, it will be treated as applying to
2686 the type of that declaration. If an attribute that only applies to
2687 declarations is applied to the type of a declaration, it will be treated
2688 as applying to that declaration; and, for compatibility with code
2689 placing the attributes immediately before the identifier declared, such
2690 an attribute applied to a function return type will be treated as
2691 applying to the function type, and such an attribute applied to an array
2692 element type will be treated as applying to the array type. If an
2693 attribute that only applies to function types is applied to a
2694 pointer-to-function type, it will be treated as applying to the pointer
2695 target type; if such an attribute is applied to a function return type
2696 that is not a pointer-to-function type, it will be treated as applying
2697 to the function type.
2699 @node Function Prototypes
2700 @section Prototypes and Old-Style Function Definitions
2701 @cindex function prototype declarations
2702 @cindex old-style function definitions
2703 @cindex promotion of formal parameters
2705 GNU C extends ISO C to allow a function prototype to override a later
2706 old-style non-prototype definition. Consider the following example:
2709 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2716 /* @r{Prototype function declaration.} */
2717 int isroot P((uid_t));
2719 /* @r{Old-style function definition.} */
2721 isroot (x) /* ??? lossage here ??? */
2728 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2729 not allow this example, because subword arguments in old-style
2730 non-prototype definitions are promoted. Therefore in this example the
2731 function definition's argument is really an @code{int}, which does not
2732 match the prototype argument type of @code{short}.
2734 This restriction of ISO C makes it hard to write code that is portable
2735 to traditional C compilers, because the programmer does not know
2736 whether the @code{uid_t} type is @code{short}, @code{int}, or
2737 @code{long}. Therefore, in cases like these GNU C allows a prototype
2738 to override a later old-style definition. More precisely, in GNU C, a
2739 function prototype argument type overrides the argument type specified
2740 by a later old-style definition if the former type is the same as the
2741 latter type before promotion. Thus in GNU C the above example is
2742 equivalent to the following:
2755 GNU C++ does not support old-style function definitions, so this
2756 extension is irrelevant.
2759 @section C++ Style Comments
2761 @cindex C++ comments
2762 @cindex comments, C++ style
2764 In GNU C, you may use C++ style comments, which start with @samp{//} and
2765 continue until the end of the line. Many other C implementations allow
2766 such comments, and they are included in the 1999 C standard. However,
2767 C++ style comments are not recognized if you specify an @option{-std}
2768 option specifying a version of ISO C before C99, or @option{-ansi}
2769 (equivalent to @option{-std=c89}).
2772 @section Dollar Signs in Identifier Names
2774 @cindex dollar signs in identifier names
2775 @cindex identifier names, dollar signs in
2777 In GNU C, you may normally use dollar signs in identifier names.
2778 This is because many traditional C implementations allow such identifiers.
2779 However, dollar signs in identifiers are not supported on a few target
2780 machines, typically because the target assembler does not allow them.
2782 @node Character Escapes
2783 @section The Character @key{ESC} in Constants
2785 You can use the sequence @samp{\e} in a string or character constant to
2786 stand for the ASCII character @key{ESC}.
2789 @section Inquiring on Alignment of Types or Variables
2791 @cindex type alignment
2792 @cindex variable alignment
2794 The keyword @code{__alignof__} allows you to inquire about how an object
2795 is aligned, or the minimum alignment usually required by a type. Its
2796 syntax is just like @code{sizeof}.
2798 For example, if the target machine requires a @code{double} value to be
2799 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2800 This is true on many RISC machines. On more traditional machine
2801 designs, @code{__alignof__ (double)} is 4 or even 2.
2803 Some machines never actually require alignment; they allow reference to any
2804 data type even at an odd addresses. For these machines, @code{__alignof__}
2805 reports the @emph{recommended} alignment of a type.
2807 If the operand of @code{__alignof__} is an lvalue rather than a type,
2808 its value is the required alignment for its type, taking into account
2809 any minimum alignment specified with GCC's @code{__attribute__}
2810 extension (@pxref{Variable Attributes}). For example, after this
2814 struct foo @{ int x; char y; @} foo1;
2818 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2819 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2821 It is an error to ask for the alignment of an incomplete type.
2823 @node Variable Attributes
2824 @section Specifying Attributes of Variables
2825 @cindex attribute of variables
2826 @cindex variable attributes
2828 The keyword @code{__attribute__} allows you to specify special
2829 attributes of variables or structure fields. This keyword is followed
2830 by an attribute specification inside double parentheses. Ten
2831 attributes are currently defined for variables: @code{aligned},
2832 @code{mode}, @code{nocommon}, @code{packed}, @code{section},
2833 @code{transparent_union}, @code{unused}, @code{deprecated},
2834 @code{vector_size}, and @code{weak}. Some other attributes are defined
2835 for variables on particular target systems. Other attributes are
2836 available for functions (@pxref{Function Attributes}) and for types
2837 (@pxref{Type Attributes}). Other front ends might define more
2838 attributes (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2840 You may also specify attributes with @samp{__} preceding and following
2841 each keyword. This allows you to use them in header files without
2842 being concerned about a possible macro of the same name. For example,
2843 you may use @code{__aligned__} instead of @code{aligned}.
2845 @xref{Attribute Syntax}, for details of the exact syntax for using
2849 @cindex @code{aligned} attribute
2850 @item aligned (@var{alignment})
2851 This attribute specifies a minimum alignment for the variable or
2852 structure field, measured in bytes. For example, the declaration:
2855 int x __attribute__ ((aligned (16))) = 0;
2859 causes the compiler to allocate the global variable @code{x} on a
2860 16-byte boundary. On a 68040, this could be used in conjunction with
2861 an @code{asm} expression to access the @code{move16} instruction which
2862 requires 16-byte aligned operands.
2864 You can also specify the alignment of structure fields. For example, to
2865 create a double-word aligned @code{int} pair, you could write:
2868 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2872 This is an alternative to creating a union with a @code{double} member
2873 that forces the union to be double-word aligned.
2875 As in the preceding examples, you can explicitly specify the alignment
2876 (in bytes) that you wish the compiler to use for a given variable or
2877 structure field. Alternatively, you can leave out the alignment factor
2878 and just ask the compiler to align a variable or field to the maximum
2879 useful alignment for the target machine you are compiling for. For
2880 example, you could write:
2883 short array[3] __attribute__ ((aligned));
2886 Whenever you leave out the alignment factor in an @code{aligned} attribute
2887 specification, the compiler automatically sets the alignment for the declared
2888 variable or field to the largest alignment which is ever used for any data
2889 type on the target machine you are compiling for. Doing this can often make
2890 copy operations more efficient, because the compiler can use whatever
2891 instructions copy the biggest chunks of memory when performing copies to
2892 or from the variables or fields that you have aligned this way.
2894 The @code{aligned} attribute can only increase the alignment; but you
2895 can decrease it by specifying @code{packed} as well. See below.
2897 Note that the effectiveness of @code{aligned} attributes may be limited
2898 by inherent limitations in your linker. On many systems, the linker is
2899 only able to arrange for variables to be aligned up to a certain maximum
2900 alignment. (For some linkers, the maximum supported alignment may
2901 be very very small.) If your linker is only able to align variables
2902 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
2903 in an @code{__attribute__} will still only provide you with 8 byte
2904 alignment. See your linker documentation for further information.
2906 @item mode (@var{mode})
2907 @cindex @code{mode} attribute
2908 This attribute specifies the data type for the declaration---whichever
2909 type corresponds to the mode @var{mode}. This in effect lets you
2910 request an integer or floating point type according to its width.
2912 You may also specify a mode of @samp{byte} or @samp{__byte__} to
2913 indicate the mode corresponding to a one-byte integer, @samp{word} or
2914 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
2915 or @samp{__pointer__} for the mode used to represent pointers.
2918 @cindex @code{nocommon} attribute
2920 This attribute specifies requests GCC not to place a variable
2921 ``common'' but instead to allocate space for it directly. If you
2922 specify the @option{-fno-common} flag, GCC will do this for all
2925 Specifying the @code{nocommon} attribute for a variable provides an
2926 initialization of zeros. A variable may only be initialized in one
2930 @cindex @code{packed} attribute
2931 The @code{packed} attribute specifies that a variable or structure field
2932 should have the smallest possible alignment---one byte for a variable,
2933 and one bit for a field, unless you specify a larger value with the
2934 @code{aligned} attribute.
2936 Here is a structure in which the field @code{x} is packed, so that it
2937 immediately follows @code{a}:
2943 int x[2] __attribute__ ((packed));
2947 @item section ("@var{section-name}")
2948 @cindex @code{section} variable attribute
2949 Normally, the compiler places the objects it generates in sections like
2950 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
2951 or you need certain particular variables to appear in special sections,
2952 for example to map to special hardware. The @code{section}
2953 attribute specifies that a variable (or function) lives in a particular
2954 section. For example, this small program uses several specific section names:
2957 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
2958 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
2959 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
2960 int init_data __attribute__ ((section ("INITDATA"))) = 0;
2964 /* Initialize stack pointer */
2965 init_sp (stack + sizeof (stack));
2967 /* Initialize initialized data */
2968 memcpy (&init_data, &data, &edata - &data);
2970 /* Turn on the serial ports */
2977 Use the @code{section} attribute with an @emph{initialized} definition
2978 of a @emph{global} variable, as shown in the example. GCC issues
2979 a warning and otherwise ignores the @code{section} attribute in
2980 uninitialized variable declarations.
2982 You may only use the @code{section} attribute with a fully initialized
2983 global definition because of the way linkers work. The linker requires
2984 each object be defined once, with the exception that uninitialized
2985 variables tentatively go in the @code{common} (or @code{bss}) section
2986 and can be multiply ``defined''. You can force a variable to be
2987 initialized with the @option{-fno-common} flag or the @code{nocommon}
2990 Some file formats do not support arbitrary sections so the @code{section}
2991 attribute is not available on all platforms.
2992 If you need to map the entire contents of a module to a particular
2993 section, consider using the facilities of the linker instead.
2996 @cindex @code{shared} variable attribute
2997 On Windows NT, in addition to putting variable definitions in a named
2998 section, the section can also be shared among all running copies of an
2999 executable or DLL@. For example, this small program defines shared data
3000 by putting it in a named section @code{shared} and marking the section
3004 int foo __attribute__((section ("shared"), shared)) = 0;
3009 /* Read and write foo. All running
3010 copies see the same value. */
3016 You may only use the @code{shared} attribute along with @code{section}
3017 attribute with a fully initialized global definition because of the way
3018 linkers work. See @code{section} attribute for more information.
3020 The @code{shared} attribute is only available on Windows NT@.
3022 @item transparent_union
3023 This attribute, attached to a function parameter which is a union, means
3024 that the corresponding argument may have the type of any union member,
3025 but the argument is passed as if its type were that of the first union
3026 member. For more details see @xref{Type Attributes}. You can also use
3027 this attribute on a @code{typedef} for a union data type; then it
3028 applies to all function parameters with that type.
3031 This attribute, attached to a variable, means that the variable is meant
3032 to be possibly unused. GCC will not produce a warning for this
3036 The @code{deprecated} attribute results in a warning if the variable
3037 is used anywhere in the source file. This is useful when identifying
3038 variables that are expected to be removed in a future version of a
3039 program. The warning also includes the location of the declaration
3040 of the deprecated variable, to enable users to easily find further
3041 information about why the variable is deprecated, or what they should
3042 do instead. Note that the warnings only occurs for uses:
3045 extern int old_var __attribute__ ((deprecated));
3047 int new_fn () @{ return old_var; @}
3050 results in a warning on line 3 but not line 2.
3052 The @code{deprecated} attribute can also be used for functions and
3053 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3055 @item vector_size (@var{bytes})
3056 This attribute specifies the vector size for the variable, measured in
3057 bytes. For example, the declaration:
3060 int foo __attribute__ ((vector_size (16)));
3064 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3065 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3066 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3068 This attribute is only applicable to integral and float scalars,
3069 although arrays, pointers, and function return values are allowed in
3070 conjunction with this construct.
3072 Aggregates with this attribute are invalid, even if they are of the same
3073 size as a corresponding scalar. For example, the declaration:
3076 struct S @{ int a; @};
3077 struct S __attribute__ ((vector_size (16))) foo;
3081 is invalid even if the size of the structure is the same as the size of
3085 The @code{weak} attribute is described in @xref{Function Attributes}.
3087 @item model (@var{model-name})
3088 @cindex variable addressability on the M32R/D
3089 Use this attribute on the M32R/D to set the addressability of an object.
3090 The identifier @var{model-name} is one of @code{small}, @code{medium},
3091 or @code{large}, representing each of the code models.
3093 Small model objects live in the lower 16MB of memory (so that their
3094 addresses can be loaded with the @code{ld24} instruction).
3096 Medium and large model objects may live anywhere in the 32-bit address space
3097 (the compiler will generate @code{seth/add3} instructions to load their
3102 To specify multiple attributes, separate them by commas within the
3103 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3106 @node Type Attributes
3107 @section Specifying Attributes of Types
3108 @cindex attribute of types
3109 @cindex type attributes
3111 The keyword @code{__attribute__} allows you to specify special
3112 attributes of @code{struct} and @code{union} types when you define such
3113 types. This keyword is followed by an attribute specification inside
3114 double parentheses. Five attributes are currently defined for types:
3115 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3116 and @code{deprecated}. Other attributes are defined for functions
3117 (@pxref{Function Attributes}) and for variables (@pxref{Variable Attributes}).
3119 You may also specify any one of these attributes with @samp{__}
3120 preceding and following its keyword. This allows you to use these
3121 attributes in header files without being concerned about a possible
3122 macro of the same name. For example, you may use @code{__aligned__}
3123 instead of @code{aligned}.
3125 You may specify the @code{aligned} and @code{transparent_union}
3126 attributes either in a @code{typedef} declaration or just past the
3127 closing curly brace of a complete enum, struct or union type
3128 @emph{definition} and the @code{packed} attribute only past the closing
3129 brace of a definition.
3131 You may also specify attributes between the enum, struct or union
3132 tag and the name of the type rather than after the closing brace.
3134 @xref{Attribute Syntax}, for details of the exact syntax for using
3138 @cindex @code{aligned} attribute
3139 @item aligned (@var{alignment})
3140 This attribute specifies a minimum alignment (in bytes) for variables
3141 of the specified type. For example, the declarations:
3144 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3145 typedef int more_aligned_int __attribute__ ((aligned (8)));
3149 force the compiler to insure (as far as it can) that each variable whose
3150 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3151 aligned @emph{at least} on a 8-byte boundary. On a Sparc, having all
3152 variables of type @code{struct S} aligned to 8-byte boundaries allows
3153 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3154 store) instructions when copying one variable of type @code{struct S} to
3155 another, thus improving run-time efficiency.
3157 Note that the alignment of any given @code{struct} or @code{union} type
3158 is required by the ISO C standard to be at least a perfect multiple of
3159 the lowest common multiple of the alignments of all of the members of
3160 the @code{struct} or @code{union} in question. This means that you @emph{can}
3161 effectively adjust the alignment of a @code{struct} or @code{union}
3162 type by attaching an @code{aligned} attribute to any one of the members
3163 of such a type, but the notation illustrated in the example above is a
3164 more obvious, intuitive, and readable way to request the compiler to
3165 adjust the alignment of an entire @code{struct} or @code{union} type.
3167 As in the preceding example, you can explicitly specify the alignment
3168 (in bytes) that you wish the compiler to use for a given @code{struct}
3169 or @code{union} type. Alternatively, you can leave out the alignment factor
3170 and just ask the compiler to align a type to the maximum
3171 useful alignment for the target machine you are compiling for. For
3172 example, you could write:
3175 struct S @{ short f[3]; @} __attribute__ ((aligned));
3178 Whenever you leave out the alignment factor in an @code{aligned}
3179 attribute specification, the compiler automatically sets the alignment
3180 for the type to the largest alignment which is ever used for any data
3181 type on the target machine you are compiling for. Doing this can often
3182 make copy operations more efficient, because the compiler can use
3183 whatever instructions copy the biggest chunks of memory when performing
3184 copies to or from the variables which have types that you have aligned
3187 In the example above, if the size of each @code{short} is 2 bytes, then
3188 the size of the entire @code{struct S} type is 6 bytes. The smallest
3189 power of two which is greater than or equal to that is 8, so the
3190 compiler sets the alignment for the entire @code{struct S} type to 8
3193 Note that although you can ask the compiler to select a time-efficient
3194 alignment for a given type and then declare only individual stand-alone
3195 objects of that type, the compiler's ability to select a time-efficient
3196 alignment is primarily useful only when you plan to create arrays of
3197 variables having the relevant (efficiently aligned) type. If you
3198 declare or use arrays of variables of an efficiently-aligned type, then
3199 it is likely that your program will also be doing pointer arithmetic (or
3200 subscripting, which amounts to the same thing) on pointers to the
3201 relevant type, and the code that the compiler generates for these
3202 pointer arithmetic operations will often be more efficient for
3203 efficiently-aligned types than for other types.
3205 The @code{aligned} attribute can only increase the alignment; but you
3206 can decrease it by specifying @code{packed} as well. See below.
3208 Note that the effectiveness of @code{aligned} attributes may be limited
3209 by inherent limitations in your linker. On many systems, the linker is
3210 only able to arrange for variables to be aligned up to a certain maximum
3211 alignment. (For some linkers, the maximum supported alignment may
3212 be very very small.) If your linker is only able to align variables
3213 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3214 in an @code{__attribute__} will still only provide you with 8 byte
3215 alignment. See your linker documentation for further information.
3218 This attribute, attached to an @code{enum}, @code{struct}, or
3219 @code{union} type definition, specified that the minimum required memory
3220 be used to represent the type.
3222 @opindex fshort-enums
3223 Specifying this attribute for @code{struct} and @code{union} types is
3224 equivalent to specifying the @code{packed} attribute on each of the
3225 structure or union members. Specifying the @option{-fshort-enums}
3226 flag on the line is equivalent to specifying the @code{packed}
3227 attribute on all @code{enum} definitions.
3229 You may only specify this attribute after a closing curly brace on an
3230 @code{enum} definition, not in a @code{typedef} declaration, unless that
3231 declaration also contains the definition of the @code{enum}.
3233 @item transparent_union
3234 This attribute, attached to a @code{union} type definition, indicates
3235 that any function parameter having that union type causes calls to that
3236 function to be treated in a special way.
3238 First, the argument corresponding to a transparent union type can be of
3239 any type in the union; no cast is required. Also, if the union contains
3240 a pointer type, the corresponding argument can be a null pointer
3241 constant or a void pointer expression; and if the union contains a void
3242 pointer type, the corresponding argument can be any pointer expression.
3243 If the union member type is a pointer, qualifiers like @code{const} on
3244 the referenced type must be respected, just as with normal pointer
3247 Second, the argument is passed to the function using the calling
3248 conventions of first member of the transparent union, not the calling
3249 conventions of the union itself. All members of the union must have the
3250 same machine representation; this is necessary for this argument passing
3253 Transparent unions are designed for library functions that have multiple
3254 interfaces for compatibility reasons. For example, suppose the
3255 @code{wait} function must accept either a value of type @code{int *} to
3256 comply with Posix, or a value of type @code{union wait *} to comply with
3257 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3258 @code{wait} would accept both kinds of arguments, but it would also
3259 accept any other pointer type and this would make argument type checking
3260 less useful. Instead, @code{<sys/wait.h>} might define the interface
3268 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3270 pid_t wait (wait_status_ptr_t);
3273 This interface allows either @code{int *} or @code{union wait *}
3274 arguments to be passed, using the @code{int *} calling convention.
3275 The program can call @code{wait} with arguments of either type:
3278 int w1 () @{ int w; return wait (&w); @}
3279 int w2 () @{ union wait w; return wait (&w); @}
3282 With this interface, @code{wait}'s implementation might look like this:
3285 pid_t wait (wait_status_ptr_t p)
3287 return waitpid (-1, p.__ip, 0);
3292 When attached to a type (including a @code{union} or a @code{struct}),
3293 this attribute means that variables of that type are meant to appear
3294 possibly unused. GCC will not produce a warning for any variables of
3295 that type, even if the variable appears to do nothing. This is often
3296 the case with lock or thread classes, which are usually defined and then
3297 not referenced, but contain constructors and destructors that have
3298 nontrivial bookkeeping functions.
3301 The @code{deprecated} attribute results in a warning if the type
3302 is used anywhere in the source file. This is useful when identifying
3303 types that are expected to be removed in a future version of a program.
3304 If possible, the warning also includes the location of the declaration
3305 of the deprecated type, to enable users to easily find further
3306 information about why the type is deprecated, or what they should do
3307 instead. Note that the warnings only occur for uses and then only
3308 if the type is being applied to an identifier that itself is not being
3309 declared as deprecated.
3312 typedef int T1 __attribute__ ((deprecated));
3316 typedef T1 T3 __attribute__ ((deprecated));
3317 T3 z __attribute__ ((deprecated));
3320 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3321 warning is issued for line 4 because T2 is not explicitly
3322 deprecated. Line 5 has no warning because T3 is explicitly
3323 deprecated. Similarly for line 6.
3325 The @code{deprecated} attribute can also be used for functions and
3326 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3330 To specify multiple attributes, separate them by commas within the
3331 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3335 @section An Inline Function is As Fast As a Macro
3336 @cindex inline functions
3337 @cindex integrating function code
3339 @cindex macros, inline alternative
3341 By declaring a function @code{inline}, you can direct GCC to
3342 integrate that function's code into the code for its callers. This
3343 makes execution faster by eliminating the function-call overhead; in
3344 addition, if any of the actual argument values are constant, their known
3345 values may permit simplifications at compile time so that not all of the
3346 inline function's code needs to be included. The effect on code size is
3347 less predictable; object code may be larger or smaller with function
3348 inlining, depending on the particular case. Inlining of functions is an
3349 optimization and it really ``works'' only in optimizing compilation. If
3350 you don't use @option{-O}, no function is really inline.
3352 Inline functions are included in the ISO C99 standard, but there are
3353 currently substantial differences between what GCC implements and what
3354 the ISO C99 standard requires.
3356 To declare a function inline, use the @code{inline} keyword in its
3357 declaration, like this:
3367 (If you are writing a header file to be included in ISO C programs, write
3368 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3369 You can also make all ``simple enough'' functions inline with the option
3370 @option{-finline-functions}.
3373 Note that certain usages in a function definition can make it unsuitable
3374 for inline substitution. Among these usages are: use of varargs, use of
3375 alloca, use of variable sized data types (@pxref{Variable Length}),
3376 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3377 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3378 will warn when a function marked @code{inline} could not be substituted,
3379 and will give the reason for the failure.
3381 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3382 does not affect the linkage of the function.
3384 @cindex automatic @code{inline} for C++ member fns
3385 @cindex @code{inline} automatic for C++ member fns
3386 @cindex member fns, automatically @code{inline}
3387 @cindex C++ member fns, automatically @code{inline}
3388 @opindex fno-default-inline
3389 GCC automatically inlines member functions defined within the class
3390 body of C++ programs even if they are not explicitly declared
3391 @code{inline}. (You can override this with @option{-fno-default-inline};
3392 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3394 @cindex inline functions, omission of
3395 @opindex fkeep-inline-functions
3396 When a function is both inline and @code{static}, if all calls to the
3397 function are integrated into the caller, and the function's address is
3398 never used, then the function's own assembler code is never referenced.
3399 In this case, GCC does not actually output assembler code for the
3400 function, unless you specify the option @option{-fkeep-inline-functions}.
3401 Some calls cannot be integrated for various reasons (in particular,
3402 calls that precede the function's definition cannot be integrated, and
3403 neither can recursive calls within the definition). If there is a
3404 nonintegrated call, then the function is compiled to assembler code as
3405 usual. The function must also be compiled as usual if the program
3406 refers to its address, because that can't be inlined.
3408 @cindex non-static inline function
3409 When an inline function is not @code{static}, then the compiler must assume
3410 that there may be calls from other source files; since a global symbol can
3411 be defined only once in any program, the function must not be defined in
3412 the other source files, so the calls therein cannot be integrated.
3413 Therefore, a non-@code{static} inline function is always compiled on its
3414 own in the usual fashion.
3416 If you specify both @code{inline} and @code{extern} in the function
3417 definition, then the definition is used only for inlining. In no case
3418 is the function compiled on its own, not even if you refer to its
3419 address explicitly. Such an address becomes an external reference, as
3420 if you had only declared the function, and had not defined it.
3422 This combination of @code{inline} and @code{extern} has almost the
3423 effect of a macro. The way to use it is to put a function definition in
3424 a header file with these keywords, and put another copy of the
3425 definition (lacking @code{inline} and @code{extern}) in a library file.
3426 The definition in the header file will cause most calls to the function
3427 to be inlined. If any uses of the function remain, they will refer to
3428 the single copy in the library.
3430 For future compatibility with when GCC implements ISO C99 semantics for
3431 inline functions, it is best to use @code{static inline} only. (The
3432 existing semantics will remain available when @option{-std=gnu89} is
3433 specified, but eventually the default will be @option{-std=gnu99} and
3434 that will implement the C99 semantics, though it does not do so yet.)
3436 GCC does not inline any functions when not optimizing unless you specify
3437 the @samp{always_inline} attribute for the function, like this:
3441 inline void foo (const char) __attribute__((always_inline));
3445 @section Assembler Instructions with C Expression Operands
3446 @cindex extended @code{asm}
3447 @cindex @code{asm} expressions
3448 @cindex assembler instructions
3451 In an assembler instruction using @code{asm}, you can specify the
3452 operands of the instruction using C expressions. This means you need not
3453 guess which registers or memory locations will contain the data you want
3456 You must specify an assembler instruction template much like what
3457 appears in a machine description, plus an operand constraint string for
3460 For example, here is how to use the 68881's @code{fsinx} instruction:
3463 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3467 Here @code{angle} is the C expression for the input operand while
3468 @code{result} is that of the output operand. Each has @samp{"f"} as its
3469 operand constraint, saying that a floating point register is required.
3470 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3471 output operands' constraints must use @samp{=}. The constraints use the
3472 same language used in the machine description (@pxref{Constraints}).
3474 Each operand is described by an operand-constraint string followed by
3475 the C expression in parentheses. A colon separates the assembler
3476 template from the first output operand and another separates the last
3477 output operand from the first input, if any. Commas separate the
3478 operands within each group. The total number of operands is currently
3479 limited to 30; this limitation may be lifted in some future version of
3482 If there are no output operands but there are input operands, you must
3483 place two consecutive colons surrounding the place where the output
3486 As of GCC version 3.1, it is also possible to specify input and output
3487 operands using symbolic names which can be referenced within the
3488 assembler code. These names are specified inside square brackets
3489 preceding the constraint string, and can be referenced inside the
3490 assembler code using @code{%[@var{name}]} instead of a percentage sign
3491 followed by the operand number. Using named operands the above example
3495 asm ("fsinx %[angle],%[output]"
3496 : [output] "=f" (result)
3497 : [angle] "f" (angle));
3501 Note that the symbolic operand names have no relation whatsoever to
3502 other C identifiers. You may use any name you like, even those of
3503 existing C symbols, but must ensure that no two operands within the same
3504 assembler construct use the same symbolic name.
3506 Output operand expressions must be lvalues; the compiler can check this.
3507 The input operands need not be lvalues. The compiler cannot check
3508 whether the operands have data types that are reasonable for the
3509 instruction being executed. It does not parse the assembler instruction
3510 template and does not know what it means or even whether it is valid
3511 assembler input. The extended @code{asm} feature is most often used for
3512 machine instructions the compiler itself does not know exist. If
3513 the output expression cannot be directly addressed (for example, it is a
3514 bit-field), your constraint must allow a register. In that case, GCC
3515 will use the register as the output of the @code{asm}, and then store
3516 that register into the output.
3518 The ordinary output operands must be write-only; GCC will assume that
3519 the values in these operands before the instruction are dead and need
3520 not be generated. Extended asm supports input-output or read-write
3521 operands. Use the constraint character @samp{+} to indicate such an
3522 operand and list it with the output operands.
3524 When the constraints for the read-write operand (or the operand in which
3525 only some of the bits are to be changed) allows a register, you may, as
3526 an alternative, logically split its function into two separate operands,
3527 one input operand and one write-only output operand. The connection
3528 between them is expressed by constraints which say they need to be in
3529 the same location when the instruction executes. You can use the same C
3530 expression for both operands, or different expressions. For example,
3531 here we write the (fictitious) @samp{combine} instruction with
3532 @code{bar} as its read-only source operand and @code{foo} as its
3533 read-write destination:
3536 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3540 The constraint @samp{"0"} for operand 1 says that it must occupy the
3541 same location as operand 0. A number in constraint is allowed only in
3542 an input operand and it must refer to an output operand.
3544 Only a number in the constraint can guarantee that one operand will be in
3545 the same place as another. The mere fact that @code{foo} is the value
3546 of both operands is not enough to guarantee that they will be in the
3547 same place in the generated assembler code. The following would not
3551 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3554 Various optimizations or reloading could cause operands 0 and 1 to be in
3555 different registers; GCC knows no reason not to do so. For example, the
3556 compiler might find a copy of the value of @code{foo} in one register and
3557 use it for operand 1, but generate the output operand 0 in a different
3558 register (copying it afterward to @code{foo}'s own address). Of course,
3559 since the register for operand 1 is not even mentioned in the assembler
3560 code, the result will not work, but GCC can't tell that.
3562 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3563 the operand number for a matching constraint. For example:
3566 asm ("cmoveq %1,%2,%[result]"
3567 : [result] "=r"(result)
3568 : "r" (test), "r"(new), "[result]"(old));
3571 Some instructions clobber specific hard registers. To describe this,
3572 write a third colon after the input operands, followed by the names of
3573 the clobbered hard registers (given as strings). Here is a realistic
3574 example for the VAX:
3577 asm volatile ("movc3 %0,%1,%2"
3579 : "g" (from), "g" (to), "g" (count)
3580 : "r0", "r1", "r2", "r3", "r4", "r5");
3583 You may not write a clobber description in a way that overlaps with an
3584 input or output operand. For example, you may not have an operand
3585 describing a register class with one member if you mention that register
3586 in the clobber list. There is no way for you to specify that an input
3587 operand is modified without also specifying it as an output
3588 operand. Note that if all the output operands you specify are for this
3589 purpose (and hence unused), you will then also need to specify
3590 @code{volatile} for the @code{asm} construct, as described below, to
3591 prevent GCC from deleting the @code{asm} statement as unused.
3593 If you refer to a particular hardware register from the assembler code,
3594 you will probably have to list the register after the third colon to
3595 tell the compiler the register's value is modified. In some assemblers,
3596 the register names begin with @samp{%}; to produce one @samp{%} in the
3597 assembler code, you must write @samp{%%} in the input.
3599 If your assembler instruction can alter the condition code register, add
3600 @samp{cc} to the list of clobbered registers. GCC on some machines
3601 represents the condition codes as a specific hardware register;
3602 @samp{cc} serves to name this register. On other machines, the
3603 condition code is handled differently, and specifying @samp{cc} has no
3604 effect. But it is valid no matter what the machine.
3606 If your assembler instruction modifies memory in an unpredictable
3607 fashion, add @samp{memory} to the list of clobbered registers. This
3608 will cause GCC to not keep memory values cached in registers across
3609 the assembler instruction. You will also want to add the
3610 @code{volatile} keyword if the memory affected is not listed in the
3611 inputs or outputs of the @code{asm}, as the @samp{memory} clobber does
3612 not count as a side-effect of the @code{asm}.
3614 You can put multiple assembler instructions together in a single
3615 @code{asm} template, separated by the characters normally used in assembly
3616 code for the system. A combination that works in most places is a newline
3617 to break the line, plus a tab character to move to the instruction field
3618 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
3619 assembler allows semicolons as a line-breaking character. Note that some
3620 assembler dialects use semicolons to start a comment.
3621 The input operands are guaranteed not to use any of the clobbered
3622 registers, and neither will the output operands' addresses, so you can
3623 read and write the clobbered registers as many times as you like. Here
3624 is an example of multiple instructions in a template; it assumes the
3625 subroutine @code{_foo} accepts arguments in registers 9 and 10:
3628 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
3630 : "g" (from), "g" (to)
3634 Unless an output operand has the @samp{&} constraint modifier, GCC
3635 may allocate it in the same register as an unrelated input operand, on
3636 the assumption the inputs are consumed before the outputs are produced.
3637 This assumption may be false if the assembler code actually consists of
3638 more than one instruction. In such a case, use @samp{&} for each output
3639 operand that may not overlap an input. @xref{Modifiers}.
3641 If you want to test the condition code produced by an assembler
3642 instruction, you must include a branch and a label in the @code{asm}
3643 construct, as follows:
3646 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
3652 This assumes your assembler supports local labels, as the GNU assembler
3653 and most Unix assemblers do.
3655 Speaking of labels, jumps from one @code{asm} to another are not
3656 supported. The compiler's optimizers do not know about these jumps, and
3657 therefore they cannot take account of them when deciding how to
3660 @cindex macros containing @code{asm}
3661 Usually the most convenient way to use these @code{asm} instructions is to
3662 encapsulate them in macros that look like functions. For example,
3666 (@{ double __value, __arg = (x); \
3667 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
3672 Here the variable @code{__arg} is used to make sure that the instruction
3673 operates on a proper @code{double} value, and to accept only those
3674 arguments @code{x} which can convert automatically to a @code{double}.
3676 Another way to make sure the instruction operates on the correct data
3677 type is to use a cast in the @code{asm}. This is different from using a
3678 variable @code{__arg} in that it converts more different types. For
3679 example, if the desired type were @code{int}, casting the argument to
3680 @code{int} would accept a pointer with no complaint, while assigning the
3681 argument to an @code{int} variable named @code{__arg} would warn about
3682 using a pointer unless the caller explicitly casts it.
3684 If an @code{asm} has output operands, GCC assumes for optimization
3685 purposes the instruction has no side effects except to change the output
3686 operands. This does not mean instructions with a side effect cannot be
3687 used, but you must be careful, because the compiler may eliminate them
3688 if the output operands aren't used, or move them out of loops, or
3689 replace two with one if they constitute a common subexpression. Also,
3690 if your instruction does have a side effect on a variable that otherwise
3691 appears not to change, the old value of the variable may be reused later
3692 if it happens to be found in a register.
3694 You can prevent an @code{asm} instruction from being deleted, moved
3695 significantly, or combined, by writing the keyword @code{volatile} after
3696 the @code{asm}. For example:
3699 #define get_and_set_priority(new) \
3701 asm volatile ("get_and_set_priority %0, %1" \
3702 : "=g" (__old) : "g" (new)); \
3707 If you write an @code{asm} instruction with no outputs, GCC will know
3708 the instruction has side-effects and will not delete the instruction or
3709 move it outside of loops.
3711 The @code{volatile} keyword indicates that the instruction has
3712 important side-effects. GCC will not delete a volatile @code{asm} if
3713 it is reachable. (The instruction can still be deleted if GCC can
3714 prove that control-flow will never reach the location of the
3715 instruction.) In addition, GCC will not reschedule instructions
3716 across a volatile @code{asm} instruction. For example:
3719 *(volatile int *)addr = foo;
3720 asm volatile ("eieio" : : );
3724 Assume @code{addr} contains the address of a memory mapped device
3725 register. The PowerPC @code{eieio} instruction (Enforce In-order
3726 Execution of I/O) tells the CPU to make sure that the store to that
3727 device register happens before it issues any other I/O@.
3729 Note that even a volatile @code{asm} instruction can be moved in ways
3730 that appear insignificant to the compiler, such as across jump
3731 instructions. You can't expect a sequence of volatile @code{asm}
3732 instructions to remain perfectly consecutive. If you want consecutive
3733 output, use a single @code{asm}. Also, GCC will perform some
3734 optimizations across a volatile @code{asm} instruction; GCC does not
3735 ``forget everything'' when it encounters a volatile @code{asm}
3736 instruction the way some other compilers do.
3738 An @code{asm} instruction without any operands or clobbers (an ``old
3739 style'' @code{asm}) will be treated identically to a volatile
3740 @code{asm} instruction.
3742 It is a natural idea to look for a way to give access to the condition
3743 code left by the assembler instruction. However, when we attempted to
3744 implement this, we found no way to make it work reliably. The problem
3745 is that output operands might need reloading, which would result in
3746 additional following ``store'' instructions. On most machines, these
3747 instructions would alter the condition code before there was time to
3748 test it. This problem doesn't arise for ordinary ``test'' and
3749 ``compare'' instructions because they don't have any output operands.
3751 For reasons similar to those described above, it is not possible to give
3752 an assembler instruction access to the condition code left by previous
3755 If you are writing a header file that should be includable in ISO C
3756 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
3759 @subsection i386 floating point asm operands
3761 There are several rules on the usage of stack-like regs in
3762 asm_operands insns. These rules apply only to the operands that are
3767 Given a set of input regs that die in an asm_operands, it is
3768 necessary to know which are implicitly popped by the asm, and
3769 which must be explicitly popped by gcc.
3771 An input reg that is implicitly popped by the asm must be
3772 explicitly clobbered, unless it is constrained to match an
3776 For any input reg that is implicitly popped by an asm, it is
3777 necessary to know how to adjust the stack to compensate for the pop.
3778 If any non-popped input is closer to the top of the reg-stack than
3779 the implicitly popped reg, it would not be possible to know what the
3780 stack looked like---it's not clear how the rest of the stack ``slides
3783 All implicitly popped input regs must be closer to the top of
3784 the reg-stack than any input that is not implicitly popped.
3786 It is possible that if an input dies in an insn, reload might
3787 use the input reg for an output reload. Consider this example:
3790 asm ("foo" : "=t" (a) : "f" (b));
3793 This asm says that input B is not popped by the asm, and that
3794 the asm pushes a result onto the reg-stack, i.e., the stack is one
3795 deeper after the asm than it was before. But, it is possible that
3796 reload will think that it can use the same reg for both the input and
3797 the output, if input B dies in this insn.
3799 If any input operand uses the @code{f} constraint, all output reg
3800 constraints must use the @code{&} earlyclobber.
3802 The asm above would be written as
3805 asm ("foo" : "=&t" (a) : "f" (b));
3809 Some operands need to be in particular places on the stack. All
3810 output operands fall in this category---there is no other way to
3811 know which regs the outputs appear in unless the user indicates
3812 this in the constraints.
3814 Output operands must specifically indicate which reg an output
3815 appears in after an asm. @code{=f} is not allowed: the operand
3816 constraints must select a class with a single reg.
3819 Output operands may not be ``inserted'' between existing stack regs.
3820 Since no 387 opcode uses a read/write operand, all output operands
3821 are dead before the asm_operands, and are pushed by the asm_operands.
3822 It makes no sense to push anywhere but the top of the reg-stack.
3824 Output operands must start at the top of the reg-stack: output
3825 operands may not ``skip'' a reg.
3828 Some asm statements may need extra stack space for internal
3829 calculations. This can be guaranteed by clobbering stack registers
3830 unrelated to the inputs and outputs.
3834 Here are a couple of reasonable asms to want to write. This asm
3835 takes one input, which is internally popped, and produces two outputs.
3838 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
3841 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
3842 and replaces them with one output. The user must code the @code{st(1)}
3843 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
3846 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
3852 @section Controlling Names Used in Assembler Code
3853 @cindex assembler names for identifiers
3854 @cindex names used in assembler code
3855 @cindex identifiers, names in assembler code
3857 You can specify the name to be used in the assembler code for a C
3858 function or variable by writing the @code{asm} (or @code{__asm__})
3859 keyword after the declarator as follows:
3862 int foo asm ("myfoo") = 2;
3866 This specifies that the name to be used for the variable @code{foo} in
3867 the assembler code should be @samp{myfoo} rather than the usual
3870 On systems where an underscore is normally prepended to the name of a C
3871 function or variable, this feature allows you to define names for the
3872 linker that do not start with an underscore.
3874 It does not make sense to use this feature with a non-static local
3875 variable since such variables do not have assembler names. If you are
3876 trying to put the variable in a particular register, see @ref{Explicit
3877 Reg Vars}. GCC presently accepts such code with a warning, but will
3878 probably be changed to issue an error, rather than a warning, in the
3881 You cannot use @code{asm} in this way in a function @emph{definition}; but
3882 you can get the same effect by writing a declaration for the function
3883 before its definition and putting @code{asm} there, like this:
3886 extern func () asm ("FUNC");
3893 It is up to you to make sure that the assembler names you choose do not
3894 conflict with any other assembler symbols. Also, you must not use a
3895 register name; that would produce completely invalid assembler code. GCC
3896 does not as yet have the ability to store static variables in registers.
3897 Perhaps that will be added.
3899 @node Explicit Reg Vars
3900 @section Variables in Specified Registers
3901 @cindex explicit register variables
3902 @cindex variables in specified registers
3903 @cindex specified registers
3904 @cindex registers, global allocation
3906 GNU C allows you to put a few global variables into specified hardware
3907 registers. You can also specify the register in which an ordinary
3908 register variable should be allocated.
3912 Global register variables reserve registers throughout the program.
3913 This may be useful in programs such as programming language
3914 interpreters which have a couple of global variables that are accessed
3918 Local register variables in specific registers do not reserve the
3919 registers. The compiler's data flow analysis is capable of determining
3920 where the specified registers contain live values, and where they are
3921 available for other uses. Stores into local register variables may be deleted
3922 when they appear to be dead according to dataflow analysis. References
3923 to local register variables may be deleted or moved or simplified.
3925 These local variables are sometimes convenient for use with the extended
3926 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
3927 output of the assembler instruction directly into a particular register.
3928 (This will work provided the register you specify fits the constraints
3929 specified for that operand in the @code{asm}.)
3937 @node Global Reg Vars
3938 @subsection Defining Global Register Variables
3939 @cindex global register variables
3940 @cindex registers, global variables in
3942 You can define a global register variable in GNU C like this:
3945 register int *foo asm ("a5");
3949 Here @code{a5} is the name of the register which should be used. Choose a
3950 register which is normally saved and restored by function calls on your
3951 machine, so that library routines will not clobber it.
3953 Naturally the register name is cpu-dependent, so you would need to
3954 conditionalize your program according to cpu type. The register
3955 @code{a5} would be a good choice on a 68000 for a variable of pointer
3956 type. On machines with register windows, be sure to choose a ``global''
3957 register that is not affected magically by the function call mechanism.
3959 In addition, operating systems on one type of cpu may differ in how they
3960 name the registers; then you would need additional conditionals. For
3961 example, some 68000 operating systems call this register @code{%a5}.
3963 Eventually there may be a way of asking the compiler to choose a register
3964 automatically, but first we need to figure out how it should choose and
3965 how to enable you to guide the choice. No solution is evident.
3967 Defining a global register variable in a certain register reserves that
3968 register entirely for this use, at least within the current compilation.
3969 The register will not be allocated for any other purpose in the functions
3970 in the current compilation. The register will not be saved and restored by
3971 these functions. Stores into this register are never deleted even if they
3972 would appear to be dead, but references may be deleted or moved or
3975 It is not safe to access the global register variables from signal
3976 handlers, or from more than one thread of control, because the system
3977 library routines may temporarily use the register for other things (unless
3978 you recompile them specially for the task at hand).
3980 @cindex @code{qsort}, and global register variables
3981 It is not safe for one function that uses a global register variable to
3982 call another such function @code{foo} by way of a third function
3983 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
3984 different source file in which the variable wasn't declared). This is
3985 because @code{lose} might save the register and put some other value there.
3986 For example, you can't expect a global register variable to be available in
3987 the comparison-function that you pass to @code{qsort}, since @code{qsort}
3988 might have put something else in that register. (If you are prepared to
3989 recompile @code{qsort} with the same global register variable, you can
3990 solve this problem.)
3992 If you want to recompile @code{qsort} or other source files which do not
3993 actually use your global register variable, so that they will not use that
3994 register for any other purpose, then it suffices to specify the compiler
3995 option @option{-ffixed-@var{reg}}. You need not actually add a global
3996 register declaration to their source code.
3998 A function which can alter the value of a global register variable cannot
3999 safely be called from a function compiled without this variable, because it
4000 could clobber the value the caller expects to find there on return.
4001 Therefore, the function which is the entry point into the part of the
4002 program that uses the global register variable must explicitly save and
4003 restore the value which belongs to its caller.
4005 @cindex register variable after @code{longjmp}
4006 @cindex global register after @code{longjmp}
4007 @cindex value after @code{longjmp}
4010 On most machines, @code{longjmp} will restore to each global register
4011 variable the value it had at the time of the @code{setjmp}. On some
4012 machines, however, @code{longjmp} will not change the value of global
4013 register variables. To be portable, the function that called @code{setjmp}
4014 should make other arrangements to save the values of the global register
4015 variables, and to restore them in a @code{longjmp}. This way, the same
4016 thing will happen regardless of what @code{longjmp} does.
4018 All global register variable declarations must precede all function
4019 definitions. If such a declaration could appear after function
4020 definitions, the declaration would be too late to prevent the register from
4021 being used for other purposes in the preceding functions.
4023 Global register variables may not have initial values, because an
4024 executable file has no means to supply initial contents for a register.
4026 On the Sparc, there are reports that g3 @dots{} g7 are suitable
4027 registers, but certain library functions, such as @code{getwd}, as well
4028 as the subroutines for division and remainder, modify g3 and g4. g1 and
4029 g2 are local temporaries.
4031 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4032 Of course, it will not do to use more than a few of those.
4034 @node Local Reg Vars
4035 @subsection Specifying Registers for Local Variables
4036 @cindex local variables, specifying registers
4037 @cindex specifying registers for local variables
4038 @cindex registers for local variables
4040 You can define a local register variable with a specified register
4044 register int *foo asm ("a5");
4048 Here @code{a5} is the name of the register which should be used. Note
4049 that this is the same syntax used for defining global register
4050 variables, but for a local variable it would appear within a function.
4052 Naturally the register name is cpu-dependent, but this is not a
4053 problem, since specific registers are most often useful with explicit
4054 assembler instructions (@pxref{Extended Asm}). Both of these things
4055 generally require that you conditionalize your program according to
4058 In addition, operating systems on one type of cpu may differ in how they
4059 name the registers; then you would need additional conditionals. For
4060 example, some 68000 operating systems call this register @code{%a5}.
4062 Defining such a register variable does not reserve the register; it
4063 remains available for other uses in places where flow control determines
4064 the variable's value is not live. However, these registers are made
4065 unavailable for use in the reload pass; excessive use of this feature
4066 leaves the compiler too few available registers to compile certain
4069 This option does not guarantee that GCC will generate code that has
4070 this variable in the register you specify at all times. You may not
4071 code an explicit reference to this register in an @code{asm} statement
4072 and assume it will always refer to this variable.
4074 Stores into local register variables may be deleted when they appear to be dead
4075 according to dataflow analysis. References to local register variables may
4076 be deleted or moved or simplified.
4078 @node Alternate Keywords
4079 @section Alternate Keywords
4080 @cindex alternate keywords
4081 @cindex keywords, alternate
4083 @option{-ansi} and the various @option{-std} options disable certain
4084 keywords. This causes trouble when you want to use GNU C extensions, or
4085 a general-purpose header file that should be usable by all programs,
4086 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4087 @code{inline} are not available in programs compiled with
4088 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4089 program compiled with @option{-std=c99}). The ISO C99 keyword
4090 @code{restrict} is only available when @option{-std=gnu99} (which will
4091 eventually be the default) or @option{-std=c99} (or the equivalent
4092 @option{-std=iso9899:1999}) is used.
4094 The way to solve these problems is to put @samp{__} at the beginning and
4095 end of each problematical keyword. For example, use @code{__asm__}
4096 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4098 Other C compilers won't accept these alternative keywords; if you want to
4099 compile with another compiler, you can define the alternate keywords as
4100 macros to replace them with the customary keywords. It looks like this:
4108 @findex __extension__
4110 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4112 prevent such warnings within one expression by writing
4113 @code{__extension__} before the expression. @code{__extension__} has no
4114 effect aside from this.
4116 @node Incomplete Enums
4117 @section Incomplete @code{enum} Types
4119 You can define an @code{enum} tag without specifying its possible values.
4120 This results in an incomplete type, much like what you get if you write
4121 @code{struct foo} without describing the elements. A later declaration
4122 which does specify the possible values completes the type.
4124 You can't allocate variables or storage using the type while it is
4125 incomplete. However, you can work with pointers to that type.
4127 This extension may not be very useful, but it makes the handling of
4128 @code{enum} more consistent with the way @code{struct} and @code{union}
4131 This extension is not supported by GNU C++.
4133 @node Function Names
4134 @section Function Names as Strings
4135 @cindex @code{__FUNCTION__} identifier
4136 @cindex @code{__PRETTY_FUNCTION__} identifier
4137 @cindex @code{__func__} identifier
4139 GCC predefines two magic identifiers to hold the name of the current
4140 function. The identifier @code{__FUNCTION__} holds the name of the function
4141 as it appears in the source. The identifier @code{__PRETTY_FUNCTION__}
4142 holds the name of the function pretty printed in a language specific
4145 These names are always the same in a C function, but in a C++ function
4146 they may be different. For example, this program:
4150 extern int printf (char *, ...);
4157 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4158 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4176 __PRETTY_FUNCTION__ = int a::sub (int)
4179 The compiler automagically replaces the identifiers with a string
4180 literal containing the appropriate name. Thus, they are neither
4181 preprocessor macros, like @code{__FILE__} and @code{__LINE__}, nor
4182 variables. This means that they catenate with other string literals, and
4183 that they can be used to initialize char arrays. For example
4186 char here[] = "Function " __FUNCTION__ " in " __FILE__;
4189 On the other hand, @samp{#ifdef __FUNCTION__} does not have any special
4190 meaning inside a function, since the preprocessor does not do anything
4191 special with the identifier @code{__FUNCTION__}.
4193 Note that these semantics are deprecated, and that GCC 3.2 will handle
4194 @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__} the same way as
4195 @code{__func__}. @code{__func__} is defined by the ISO standard C99:
4198 The identifier @code{__func__} is implicitly declared by the translator
4199 as if, immediately following the opening brace of each function
4200 definition, the declaration
4203 static const char __func__[] = "function-name";
4206 appeared, where function-name is the name of the lexically-enclosing
4207 function. This name is the unadorned name of the function.
4210 By this definition, @code{__func__} is a variable, not a string literal.
4211 In particular, @code{__func__} does not catenate with other string
4214 In @code{C++}, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__} are
4215 variables, declared in the same way as @code{__func__}.
4217 @node Return Address
4218 @section Getting the Return or Frame Address of a Function
4220 These functions may be used to get information about the callers of a
4223 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4224 This function returns the return address of the current function, or of
4225 one of its callers. The @var{level} argument is number of frames to
4226 scan up the call stack. A value of @code{0} yields the return address
4227 of the current function, a value of @code{1} yields the return address
4228 of the caller of the current function, and so forth.
4230 The @var{level} argument must be a constant integer.
4232 On some machines it may be impossible to determine the return address of
4233 any function other than the current one; in such cases, or when the top
4234 of the stack has been reached, this function will return @code{0} or a
4235 random value. In addition, @code{__builtin_frame_address} may be used
4236 to determine if the top of the stack has been reached.
4238 This function should only be used with a nonzero argument for debugging
4242 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4243 This function is similar to @code{__builtin_return_address}, but it
4244 returns the address of the function frame rather than the return address
4245 of the function. Calling @code{__builtin_frame_address} with a value of
4246 @code{0} yields the frame address of the current function, a value of
4247 @code{1} yields the frame address of the caller of the current function,
4250 The frame is the area on the stack which holds local variables and saved
4251 registers. The frame address is normally the address of the first word
4252 pushed on to the stack by the function. However, the exact definition
4253 depends upon the processor and the calling convention. If the processor
4254 has a dedicated frame pointer register, and the function has a frame,
4255 then @code{__builtin_frame_address} will return the value of the frame
4258 On some machines it may be impossible to determine the frame address of
4259 any function other than the current one; in such cases, or when the top
4260 of the stack has been reached, this function will return @code{0} if
4261 the first frame pointer is properly initialized by the startup code.
4263 This function should only be used with a nonzero argument for debugging
4267 @node Vector Extensions
4268 @section Using vector instructions through built-in functions
4270 On some targets, the instruction set contains SIMD vector instructions that
4271 operate on multiple values contained in one large register at the same time.
4272 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4275 The first step in using these extensions is to provide the necessary data
4276 types. This should be done using an appropriate @code{typedef}:
4279 typedef int v4si __attribute__ ((mode(V4SI)));
4282 The base type @code{int} is effectively ignored by the compiler, the
4283 actual properties of the new type @code{v4si} are defined by the
4284 @code{__attribute__}. It defines the machine mode to be used; for vector
4285 types these have the form @code{V@var{n}@var{B}}; @var{n} should be the
4286 number of elements in the vector, and @var{B} should be the base mode of the
4287 individual elements. The following can be used as base modes:
4291 An integer that is as wide as the smallest addressable unit, usually 8 bits.
4293 An integer, twice as wide as a QI mode integer, usually 16 bits.
4295 An integer, four times as wide as a QI mode integer, usually 32 bits.
4297 An integer, eight times as wide as a QI mode integer, usually 64 bits.
4299 A floating point value, as wide as a SI mode integer, usually 32 bits.
4301 A floating point value, as wide as a DI mode integer, usually 64 bits.
4304 Not all base types or combinations are always valid; which modes can be used
4305 is determined by the target machine. For example, if targetting the i386 MMX
4306 extensions, only @code{V8QI}, @code{V4HI} and @code{V2SI} are allowed modes.
4308 There are no @code{V1xx} vector modes - they would be identical to the
4309 corresponding base mode.
4311 There is no distinction between signed and unsigned vector modes. This
4312 distinction is made by the operations that perform on the vectors, not
4315 The types defined in this manner are somewhat special, they cannot be
4316 used with most normal C operations (i.e., a vector addition can @emph{not}
4317 be represented by a normal addition of two vector type variables). You
4318 can declare only variables and use them in function calls and returns, as
4319 well as in assignments and some casts. It is possible to cast from one
4320 vector type to another, provided they are of the same size (in fact, you
4321 can also cast vectors to and from other datatypes of the same size).
4323 A port that supports vector operations provides a set of built-in functions
4324 that can be used to operate on vectors. For example, a function to add two
4325 vectors and multiply the result by a third could look like this:
4328 v4si f (v4si a, v4si b, v4si c)
4330 v4si tmp = __builtin_addv4si (a, b);
4331 return __builtin_mulv4si (tmp, c);
4336 @node Other Builtins
4337 @section Other built-in functions provided by GCC
4338 @cindex built-in functions
4339 @findex __builtin_isgreater
4340 @findex __builtin_isgreaterequal
4341 @findex __builtin_isless
4342 @findex __builtin_islessequal
4343 @findex __builtin_islessgreater
4344 @findex __builtin_isunordered
4370 @findex fprintf_unlocked
4372 @findex fputs_unlocked
4381 @findex printf_unlocked
4403 GCC provides a large number of built-in functions other than the ones
4404 mentioned above. Some of these are for internal use in the processing
4405 of exceptions or variable-length argument lists and will not be
4406 documented here because they may change from time to time; we do not
4407 recommend general use of these functions.
4409 The remaining functions are provided for optimization purposes.
4411 @opindex fno-builtin
4412 GCC includes built-in versions of many of the functions in the standard
4413 C library. The versions prefixed with @code{__builtin_} will always be
4414 treated as having the same meaning as the C library function even if you
4415 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
4416 Many of these functions are only optimized in certain cases; if they are
4417 not optimized in a particular case, a call to the library function will
4422 The functions @code{abort}, @code{exit}, @code{_Exit} and @code{_exit}
4423 are recognized and presumed not to return, but otherwise are not built
4424 in. @code{_exit} is not recognized in strict ISO C mode (@option{-ansi},
4425 @option{-std=c89} or @option{-std=c99}). @code{_Exit} is not recognized in
4426 strict C89 mode (@option{-ansi} or @option{-std=c89}).
4428 Outside strict ISO C mode, the functions @code{alloca}, @code{bcmp},
4429 @code{bzero}, @code{index}, @code{rindex}, @code{ffs}, @code{fputs_unlocked},
4430 @code{printf_unlocked} and @code{fprintf_unlocked} may be handled as
4431 built-in functions. All these functions have corresponding versions
4432 prefixed with @code{__builtin_}, which may be used even in strict C89
4435 The ISO C99 functions @code{conj}, @code{conjf}, @code{conjl},
4436 @code{creal}, @code{crealf}, @code{creall}, @code{cimag}, @code{cimagf},
4437 @code{cimagl}, @code{llabs} and @code{imaxabs} are handled as built-in
4438 functions except in strict ISO C89 mode. There are also built-in
4439 versions of the ISO C99 functions @code{cosf}, @code{cosl},
4440 @code{fabsf}, @code{fabsl}, @code{sinf}, @code{sinl}, @code{sqrtf}, and
4441 @code{sqrtl}, that are recognized in any mode since ISO C89 reserves
4442 these names for the purpose to which ISO C99 puts them. All these
4443 functions have corresponding versions prefixed with @code{__builtin_}.
4445 The ISO C89 functions @code{abs}, @code{cos}, @code{fabs},
4446 @code{fprintf}, @code{fputs}, @code{labs}, @code{memcmp}, @code{memcpy},
4447 @code{memset}, @code{printf}, @code{sin}, @code{sqrt}, @code{strcat},
4448 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
4449 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
4450 @code{strpbrk}, @code{strrchr}, @code{strspn}, and @code{strstr} are all
4451 recognized as built-in functions unless @option{-fno-builtin} is
4452 specified (or @option{-fno-builtin-@var{function}} is specified for an
4453 individual function). All of these functions have corresponding
4454 versions prefixed with @code{__builtin_}.
4456 GCC provides built-in versions of the ISO C99 floating point comparison
4457 macros that avoid raising exceptions for unordered operands. They have
4458 the same names as the standard macros ( @code{isgreater},
4459 @code{isgreaterequal}, @code{isless}, @code{islessequal},
4460 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
4461 prefixed. We intend for a library implementor to be able to simply
4462 @code{#define} each standard macro to its built-in equivalent.
4464 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
4466 You can use the built-in function @code{__builtin_types_compatible_p} to
4467 determine whether two types are the same.
4469 This built-in function returns 1 if the unqualified versions of the
4470 types @var{type1} and @var{type2} (which are types, not expressions) are
4471 compatible, 0 otherwise. The result of this built-in function can be
4472 used in integer constant expressions.
4474 This built-in function ignores top level qualifiers (e.g., @code{const},
4475 @code{volatile}). For example, @code{int} is equivalent to @code{const
4478 The type @code{int[]} and @code{int[5]} are compatible. On the other
4479 hand, @code{int} and @code{char *} are not compatible, even if the size
4480 of their types, on the particular architecture are the same. Also, the
4481 amount of pointer indirection is taken into account when determining
4482 similarity. Consequently, @code{short *} is not similar to
4483 @code{short **}. Furthermore, two types that are typedefed are
4484 considered compatible if their underlying types are compatible.
4486 An @code{enum} type is considered to be compatible with another
4487 @code{enum} type. For example, @code{enum @{foo, bar@}} is similar to
4488 @code{enum @{hot, dog@}}.
4490 You would typically use this function in code whose execution varies
4491 depending on the arguments' types. For example:
4497 if (__builtin_types_compatible_p (typeof (x), long double)) \
4498 tmp = foo_long_double (tmp); \
4499 else if (__builtin_types_compatible_p (typeof (x), double)) \
4500 tmp = foo_double (tmp); \
4501 else if (__builtin_types_compatible_p (typeof (x), float)) \
4502 tmp = foo_float (tmp); \
4509 @emph{Note:} This construct is only available for C.
4513 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
4515 You can use the built-in function @code{__builtin_choose_expr} to
4516 evaluate code depending on the value of a constant expression. This
4517 built-in function returns @var{exp1} if @var{const_exp}, which is a
4518 constant expression that must be able to be determined at compile time,
4519 is nonzero. Otherwise it returns 0.
4521 This built-in function is analogous to the @samp{? :} operator in C,
4522 except that the expression returned has its type unaltered by promotion
4523 rules. Also, the built-in function does not evaluate the expression
4524 that was not chosen. For example, if @var{const_exp} evaluates to true,
4525 @var{exp2} is not evaluated even if it has side-effects.
4527 This built-in function can return an lvalue if the chosen argument is an
4530 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
4531 type. Similarly, if @var{exp2} is returned, its return type is the same
4538 __builtin_choose_expr (__builtin_types_compatible_p (typeof (x), double), \
4540 __builtin_choose_expr (__builtin_types_compatible_p (typeof (x), float), \
4542 /* @r{The void expression results in a compile-time error} \
4543 @r{when assigning the result to something.} */ \
4547 @emph{Note:} This construct is only available for C. Furthermore, the
4548 unused expression (@var{exp1} or @var{exp2} depending on the value of
4549 @var{const_exp}) may still generate syntax errors. This may change in
4554 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
4555 You can use the built-in function @code{__builtin_constant_p} to
4556 determine if a value is known to be constant at compile-time and hence
4557 that GCC can perform constant-folding on expressions involving that
4558 value. The argument of the function is the value to test. The function
4559 returns the integer 1 if the argument is known to be a compile-time
4560 constant and 0 if it is not known to be a compile-time constant. A
4561 return of 0 does not indicate that the value is @emph{not} a constant,
4562 but merely that GCC cannot prove it is a constant with the specified
4563 value of the @option{-O} option.
4565 You would typically use this function in an embedded application where
4566 memory was a critical resource. If you have some complex calculation,
4567 you may want it to be folded if it involves constants, but need to call
4568 a function if it does not. For example:
4571 #define Scale_Value(X) \
4572 (__builtin_constant_p (X) \
4573 ? ((X) * SCALE + OFFSET) : Scale (X))
4576 You may use this built-in function in either a macro or an inline
4577 function. However, if you use it in an inlined function and pass an
4578 argument of the function as the argument to the built-in, GCC will
4579 never return 1 when you call the inline function with a string constant
4580 or compound literal (@pxref{Compound Literals}) and will not return 1
4581 when you pass a constant numeric value to the inline function unless you
4582 specify the @option{-O} option.
4584 You may also use @code{__builtin_constant_p} in initializers for static
4585 data. For instance, you can write
4588 static const int table[] = @{
4589 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
4595 This is an acceptable initializer even if @var{EXPRESSION} is not a
4596 constant expression. GCC must be more conservative about evaluating the
4597 built-in in this case, because it has no opportunity to perform
4600 Previous versions of GCC did not accept this built-in in data
4601 initializers. The earliest version where it is completely safe is
4605 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
4606 @opindex fprofile-arcs
4607 You may use @code{__builtin_expect} to provide the compiler with
4608 branch prediction information. In general, you should prefer to
4609 use actual profile feedback for this (@option{-fprofile-arcs}), as
4610 programmers are notoriously bad at predicting how their programs
4611 actually perform. However, there are applications in which this
4612 data is hard to collect.
4614 The return value is the value of @var{exp}, which should be an
4615 integral expression. The value of @var{c} must be a compile-time
4616 constant. The semantics of the built-in are that it is expected
4617 that @var{exp} == @var{c}. For example:
4620 if (__builtin_expect (x, 0))
4625 would indicate that we do not expect to call @code{foo}, since
4626 we expect @code{x} to be zero. Since you are limited to integral
4627 expressions for @var{exp}, you should use constructions such as
4630 if (__builtin_expect (ptr != NULL, 1))
4635 when testing pointer or floating-point values.
4638 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
4639 This function is used to minimize cache-miss latency by moving data into
4640 a cache before it is accessed.
4641 You can insert calls to @code{__builtin_prefetch} into code for which
4642 you know addresses of data in memory that is likely to be accessed soon.
4643 If the target supports them, data prefetch instructions will be generated.
4644 If the prefetch is done early enough before the access then the data will
4645 be in the cache by the time it is accessed.
4647 The value of @var{addr} is the address of the memory to prefetch.
4648 There are two optional arguments, @var{rw} and @var{locality}.
4649 The value of @var{rw} is a compile-time constant one or zero; one
4650 means that the prefetch is preparing for a write to the memory address
4651 and zero, the default, means that the prefetch is preparing for a read.
4652 The value @var{locality} must be a compile-time constant integer between
4653 zero and three. A value of zero means that the data has no temporal
4654 locality, so it need not be left in the cache after the access. A value
4655 of three means that the data has a high degree of temporal locality and
4656 should be left in all levels of cache possible. Values of one and two
4657 mean, respectively, a low or moderate degree of temporal locality. The
4661 for (i = 0; i < n; i++)
4664 __builtin_prefetch (&a[i+j], 1, 1);
4665 __builtin_prefetch (&b[i+j], 0, 1);
4670 Data prefetch does not generate faults if @var{addr} is invalid, but
4671 the address expression itself must be valid. For example, a prefetch
4672 of @code{p->next} will not fault if @code{p->next} is not a valid
4673 address, but evaluation will fault if @code{p} is not a valid address.
4675 If the target does not support data prefetch, the address expression
4676 is evaluated if it includes side effects but no other code is generated
4677 and GCC does not issue a warning.
4680 @node Target Builtins
4681 @section Built-in Functions Specific to Particular Target Machines
4683 On some target machines, GCC supports many built-in functions specific
4684 to those machines. Generally these generate calls to specific machine
4685 instructions, but allow the compiler to schedule those calls.
4688 * X86 Built-in Functions::
4689 * PowerPC AltiVec Built-in Functions::
4692 @node X86 Built-in Functions
4693 @subsection X86 Built-in Functions
4695 These built-in functions are available for the i386 and x86-64 family
4696 of computers, depending on the command-line switches used.
4698 The following machine modes are available for use with MMX built-in functions
4699 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
4700 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
4701 vector of eight 8-bit integers. Some of the built-in functions operate on
4702 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
4704 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
4705 of two 32-bit floating point values.
4707 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
4708 floating point values. Some instructions use a vector of four 32-bit
4709 integers, these use @code{V4SI}. Finally, some instructions operate on an
4710 entire vector register, interpreting it as a 128-bit integer, these use mode
4713 The following built-in functions are made available by @option{-mmmx}.
4714 All of them generate the machine instruction that is part of the name.
4717 v8qi __builtin_ia32_paddb (v8qi, v8qi)
4718 v4hi __builtin_ia32_paddw (v4hi, v4hi)
4719 v2si __builtin_ia32_paddd (v2si, v2si)
4720 v8qi __builtin_ia32_psubb (v8qi, v8qi)
4721 v4hi __builtin_ia32_psubw (v4hi, v4hi)
4722 v2si __builtin_ia32_psubd (v2si, v2si)
4723 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
4724 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
4725 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
4726 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
4727 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
4728 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
4729 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
4730 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
4731 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
4732 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
4733 di __builtin_ia32_pand (di, di)
4734 di __builtin_ia32_pandn (di,di)
4735 di __builtin_ia32_por (di, di)
4736 di __builtin_ia32_pxor (di, di)
4737 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
4738 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
4739 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
4740 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
4741 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
4742 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
4743 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
4744 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
4745 v2si __builtin_ia32_punpckhdq (v2si, v2si)
4746 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
4747 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
4748 v2si __builtin_ia32_punpckldq (v2si, v2si)
4749 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
4750 v4hi __builtin_ia32_packssdw (v2si, v2si)
4751 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
4754 The following built-in functions are made available either with
4755 @option{-msse}, or with a combination of @option{-m3dnow} and
4756 @option{-march=athlon}. All of them generate the machine
4757 instruction that is part of the name.
4760 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
4761 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
4762 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
4763 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
4764 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
4765 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
4766 v8qi __builtin_ia32_pminub (v8qi, v8qi)
4767 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
4768 int __builtin_ia32_pextrw (v4hi, int)
4769 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
4770 int __builtin_ia32_pmovmskb (v8qi)
4771 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
4772 void __builtin_ia32_movntq (di *, di)
4773 void __builtin_ia32_sfence (void)
4776 The following built-in functions are available when @option{-msse} is used.
4777 All of them generate the machine instruction that is part of the name.
4780 int __builtin_ia32_comieq (v4sf, v4sf)
4781 int __builtin_ia32_comineq (v4sf, v4sf)
4782 int __builtin_ia32_comilt (v4sf, v4sf)
4783 int __builtin_ia32_comile (v4sf, v4sf)
4784 int __builtin_ia32_comigt (v4sf, v4sf)
4785 int __builtin_ia32_comige (v4sf, v4sf)
4786 int __builtin_ia32_ucomieq (v4sf, v4sf)
4787 int __builtin_ia32_ucomineq (v4sf, v4sf)
4788 int __builtin_ia32_ucomilt (v4sf, v4sf)
4789 int __builtin_ia32_ucomile (v4sf, v4sf)
4790 int __builtin_ia32_ucomigt (v4sf, v4sf)
4791 int __builtin_ia32_ucomige (v4sf, v4sf)
4792 v4sf __builtin_ia32_addps (v4sf, v4sf)
4793 v4sf __builtin_ia32_subps (v4sf, v4sf)
4794 v4sf __builtin_ia32_mulps (v4sf, v4sf)
4795 v4sf __builtin_ia32_divps (v4sf, v4sf)
4796 v4sf __builtin_ia32_addss (v4sf, v4sf)
4797 v4sf __builtin_ia32_subss (v4sf, v4sf)
4798 v4sf __builtin_ia32_mulss (v4sf, v4sf)
4799 v4sf __builtin_ia32_divss (v4sf, v4sf)
4800 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
4801 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
4802 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
4803 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
4804 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
4805 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
4806 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
4807 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
4808 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
4809 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
4810 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
4811 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
4812 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
4813 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
4814 v4si __builtin_ia32_cmpless (v4sf, v4sf)
4815 v4si __builtin_ia32_cmpgtss (v4sf, v4sf)
4816 v4si __builtin_ia32_cmpgess (v4sf, v4sf)
4817 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
4818 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
4819 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
4820 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
4821 v4si __builtin_ia32_cmpngtss (v4sf, v4sf)
4822 v4si __builtin_ia32_cmpngess (v4sf, v4sf)
4823 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
4824 v4sf __builtin_ia32_maxps (v4sf, v4sf)
4825 v4sf __builtin_ia32_maxss (v4sf, v4sf)
4826 v4sf __builtin_ia32_minps (v4sf, v4sf)
4827 v4sf __builtin_ia32_minss (v4sf, v4sf)
4828 v4sf __builtin_ia32_andps (v4sf, v4sf)
4829 v4sf __builtin_ia32_andnps (v4sf, v4sf)
4830 v4sf __builtin_ia32_orps (v4sf, v4sf)
4831 v4sf __builtin_ia32_xorps (v4sf, v4sf)
4832 v4sf __builtin_ia32_movss (v4sf, v4sf)
4833 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
4834 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
4835 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
4836 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
4837 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
4838 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
4839 v2si __builtin_ia32_cvtps2pi (v4sf)
4840 int __builtin_ia32_cvtss2si (v4sf)
4841 v2si __builtin_ia32_cvttps2pi (v4sf)
4842 int __builtin_ia32_cvttss2si (v4sf)
4843 v4sf __builtin_ia32_rcpps (v4sf)
4844 v4sf __builtin_ia32_rsqrtps (v4sf)
4845 v4sf __builtin_ia32_sqrtps (v4sf)
4846 v4sf __builtin_ia32_rcpss (v4sf)
4847 v4sf __builtin_ia32_rsqrtss (v4sf)
4848 v4sf __builtin_ia32_sqrtss (v4sf)
4849 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
4850 void __builtin_ia32_movntps (float *, v4sf)
4851 int __builtin_ia32_movmskps (v4sf)
4854 The following built-in functions are available when @option{-msse} is used.
4857 @item v4sf __builtin_ia32_loadaps (float *)
4858 Generates the @code{movaps} machine instruction as a load from memory.
4859 @item void __builtin_ia32_storeaps (float *, v4sf)
4860 Generates the @code{movaps} machine instruction as a store to memory.
4861 @item v4sf __builtin_ia32_loadups (float *)
4862 Generates the @code{movups} machine instruction as a load from memory.
4863 @item void __builtin_ia32_storeups (float *, v4sf)
4864 Generates the @code{movups} machine instruction as a store to memory.
4865 @item v4sf __builtin_ia32_loadsss (float *)
4866 Generates the @code{movss} machine instruction as a load from memory.
4867 @item void __builtin_ia32_storess (float *, v4sf)
4868 Generates the @code{movss} machine instruction as a store to memory.
4869 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
4870 Generates the @code{movhps} machine instruction as a load from memory.
4871 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
4872 Generates the @code{movlps} machine instruction as a load from memory
4873 @item void __builtin_ia32_storehps (v4sf, v2si *)
4874 Generates the @code{movhps} machine instruction as a store to memory.
4875 @item void __builtin_ia32_storelps (v4sf, v2si *)
4876 Generates the @code{movlps} machine instruction as a store to memory.
4879 The following built-in functions are available when @option{-m3dnow} is used.
4880 All of them generate the machine instruction that is part of the name.
4883 void __builtin_ia32_femms (void)
4884 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
4885 v2si __builtin_ia32_pf2id (v2sf)
4886 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
4887 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
4888 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
4889 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
4890 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
4891 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
4892 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
4893 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
4894 v2sf __builtin_ia32_pfrcp (v2sf)
4895 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
4896 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
4897 v2sf __builtin_ia32_pfrsqrt (v2sf)
4898 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
4899 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
4900 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
4901 v2sf __builtin_ia32_pi2fd (v2si)
4902 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
4905 The following built-in functions are available when both @option{-m3dnow}
4906 and @option{-march=athlon} are used. All of them generate the machine
4907 instruction that is part of the name.
4910 v2si __builtin_ia32_pf2iw (v2sf)
4911 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
4912 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
4913 v2sf __builtin_ia32_pi2fw (v2si)
4914 v2sf __builtin_ia32_pswapdsf (v2sf)
4915 v2si __builtin_ia32_pswapdsi (v2si)
4918 @node PowerPC AltiVec Built-in Functions
4919 @subsection PowerPC AltiVec Built-in Functions
4921 These built-in functions are available for the PowerPC family
4922 of computers, depending on the command-line switches used.
4924 The following machine modes are available for use with AltiVec built-in
4925 functions (@pxref{Vector Extensions}): @code{V4SI} for a vector of four
4926 32-bit integers, @code{V4SF} for a vector of four 32-bit floating point
4927 numbers, @code{V8HI} for a vector of eight 16-bit integers, and
4928 @code{V16QI} for a vector of sixteen 8-bit integers.
4930 The following functions are made available by including
4931 @code{<altivec.h>} and using @option{-maltivec} and
4932 @option{-mabi=altivec}. The functions implement the functionality
4933 described in Motorola's AltiVec Programming Interface Manual.
4935 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
4936 Internally, GCC uses built-in functions to achieve the functionality in
4937 the aforementioned header file, but they are not supported and are
4938 subject to change without notice.
4941 vector signed char vec_abs (vector signed char, vector signed char);
4942 vector signed short vec_abs (vector signed short, vector signed short);
4943 vector signed int vec_abs (vector signed int, vector signed int);
4944 vector signed float vec_abs (vector signed float, vector signed float);
4946 vector signed char vec_abss (vector signed char, vector signed char);
4947 vector signed short vec_abss (vector signed short, vector signed short);
4949 vector signed char vec_add (vector signed char, vector signed char);
4950 vector unsigned char vec_add (vector signed char, vector unsigned char);
4952 vector unsigned char vec_add (vector unsigned char, vector signed char);
4954 vector unsigned char vec_add (vector unsigned char,
4955 vector unsigned char);
4956 vector signed short vec_add (vector signed short, vector signed short);
4957 vector unsigned short vec_add (vector signed short,
4958 vector unsigned short);
4959 vector unsigned short vec_add (vector unsigned short,
4960 vector signed short);
4961 vector unsigned short vec_add (vector unsigned short,
4962 vector unsigned short);
4963 vector signed int vec_add (vector signed int, vector signed int);
4964 vector unsigned int vec_add (vector signed int, vector unsigned int);
4965 vector unsigned int vec_add (vector unsigned int, vector signed int);
4966 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
4967 vector float vec_add (vector float, vector float);
4969 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
4971 vector unsigned char vec_adds (vector signed char,
4972 vector unsigned char);
4973 vector unsigned char vec_adds (vector unsigned char,
4974 vector signed char);
4975 vector unsigned char vec_adds (vector unsigned char,
4976 vector unsigned char);
4977 vector signed char vec_adds (vector signed char, vector signed char);
4978 vector unsigned short vec_adds (vector signed short,
4979 vector unsigned short);
4980 vector unsigned short vec_adds (vector unsigned short,
4981 vector signed short);
4982 vector unsigned short vec_adds (vector unsigned short,
4983 vector unsigned short);
4984 vector signed short vec_adds (vector signed short, vector signed short);
4986 vector unsigned int vec_adds (vector signed int, vector unsigned int);
4987 vector unsigned int vec_adds (vector unsigned int, vector signed int);
4988 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
4990 vector signed int vec_adds (vector signed int, vector signed int);
4992 vector float vec_and (vector float, vector float);
4993 vector float vec_and (vector float, vector signed int);
4994 vector float vec_and (vector signed int, vector float);
4995 vector signed int vec_and (vector signed int, vector signed int);
4996 vector unsigned int vec_and (vector signed int, vector unsigned int);
4997 vector unsigned int vec_and (vector unsigned int, vector signed int);
4998 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
4999 vector signed short vec_and (vector signed short, vector signed short);
5000 vector unsigned short vec_and (vector signed short,
5001 vector unsigned short);
5002 vector unsigned short vec_and (vector unsigned short,
5003 vector signed short);
5004 vector unsigned short vec_and (vector unsigned short,
5005 vector unsigned short);
5006 vector signed char vec_and (vector signed char, vector signed char);
5007 vector unsigned char vec_and (vector signed char, vector unsigned char);
5009 vector unsigned char vec_and (vector unsigned char, vector signed char);
5011 vector unsigned char vec_and (vector unsigned char,
5012 vector unsigned char);
5014 vector float vec_andc (vector float, vector float);
5015 vector float vec_andc (vector float, vector signed int);
5016 vector float vec_andc (vector signed int, vector float);
5017 vector signed int vec_andc (vector signed int, vector signed int);
5018 vector unsigned int vec_andc (vector signed int, vector unsigned int);
5019 vector unsigned int vec_andc (vector unsigned int, vector signed int);
5020 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
5022 vector signed short vec_andc (vector signed short, vector signed short);
5024 vector unsigned short vec_andc (vector signed short,
5025 vector unsigned short);
5026 vector unsigned short vec_andc (vector unsigned short,
5027 vector signed short);
5028 vector unsigned short vec_andc (vector unsigned short,
5029 vector unsigned short);
5030 vector signed char vec_andc (vector signed char, vector signed char);
5031 vector unsigned char vec_andc (vector signed char,
5032 vector unsigned char);
5033 vector unsigned char vec_andc (vector unsigned char,
5034 vector signed char);
5035 vector unsigned char vec_andc (vector unsigned char,
5036 vector unsigned char);
5038 vector unsigned char vec_avg (vector unsigned char,
5039 vector unsigned char);
5040 vector signed char vec_avg (vector signed char, vector signed char);
5041 vector unsigned short vec_avg (vector unsigned short,
5042 vector unsigned short);
5043 vector signed short vec_avg (vector signed short, vector signed short);
5044 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
5045 vector signed int vec_avg (vector signed int, vector signed int);
5047 vector float vec_ceil (vector float);
5049 vector signed int vec_cmpb (vector float, vector float);
5051 vector signed char vec_cmpeq (vector signed char, vector signed char);
5052 vector signed char vec_cmpeq (vector unsigned char,
5053 vector unsigned char);
5054 vector signed short vec_cmpeq (vector signed short,
5055 vector signed short);
5056 vector signed short vec_cmpeq (vector unsigned short,
5057 vector unsigned short);
5058 vector signed int vec_cmpeq (vector signed int, vector signed int);
5059 vector signed int vec_cmpeq (vector unsigned int, vector unsigned int);
5060 vector signed int vec_cmpeq (vector float, vector float);
5062 vector signed int vec_cmpge (vector float, vector float);
5064 vector signed char vec_cmpgt (vector unsigned char,
5065 vector unsigned char);
5066 vector signed char vec_cmpgt (vector signed char, vector signed char);
5067 vector signed short vec_cmpgt (vector unsigned short,
5068 vector unsigned short);
5069 vector signed short vec_cmpgt (vector signed short,
5070 vector signed short);
5071 vector signed int vec_cmpgt (vector unsigned int, vector unsigned int);
5072 vector signed int vec_cmpgt (vector signed int, vector signed int);
5073 vector signed int vec_cmpgt (vector float, vector float);
5075 vector signed int vec_cmple (vector float, vector float);
5077 vector signed char vec_cmplt (vector unsigned char,
5078 vector unsigned char);
5079 vector signed char vec_cmplt (vector signed char, vector signed char);
5080 vector signed short vec_cmplt (vector unsigned short,
5081 vector unsigned short);
5082 vector signed short vec_cmplt (vector signed short,
5083 vector signed short);
5084 vector signed int vec_cmplt (vector unsigned int, vector unsigned int);
5085 vector signed int vec_cmplt (vector signed int, vector signed int);
5086 vector signed int vec_cmplt (vector float, vector float);
5088 vector float vec_ctf (vector unsigned int, const char);
5089 vector float vec_ctf (vector signed int, const char);
5091 vector signed int vec_cts (vector float, const char);
5093 vector unsigned int vec_ctu (vector float, const char);
5095 void vec_dss (const char);
5097 void vec_dssall (void);
5099 void vec_dst (void *, int, const char);
5101 void vec_dstst (void *, int, const char);
5103 void vec_dststt (void *, int, const char);
5105 void vec_dstt (void *, int, const char);
5107 vector float vec_expte (vector float, vector float);
5109 vector float vec_floor (vector float, vector float);
5111 vector float vec_ld (int, vector float *);
5112 vector float vec_ld (int, float *):
5113 vector signed int vec_ld (int, int *);
5114 vector signed int vec_ld (int, vector signed int *);
5115 vector unsigned int vec_ld (int, vector unsigned int *);
5116 vector unsigned int vec_ld (int, unsigned int *);
5117 vector signed short vec_ld (int, short *, vector signed short *);
5118 vector unsigned short vec_ld (int, unsigned short *,
5119 vector unsigned short *);
5120 vector signed char vec_ld (int, signed char *);
5121 vector signed char vec_ld (int, vector signed char *);
5122 vector unsigned char vec_ld (int, unsigned char *);
5123 vector unsigned char vec_ld (int, vector unsigned char *);
5125 vector signed char vec_lde (int, signed char *);
5126 vector unsigned char vec_lde (int, unsigned char *);
5127 vector signed short vec_lde (int, short *);
5128 vector unsigned short vec_lde (int, unsigned short *);
5129 vector float vec_lde (int, float *);
5130 vector signed int vec_lde (int, int *);
5131 vector unsigned int vec_lde (int, unsigned int *);
5133 void float vec_ldl (int, float *);
5134 void float vec_ldl (int, vector float *);
5135 void signed int vec_ldl (int, vector signed int *);
5136 void signed int vec_ldl (int, int *);
5137 void unsigned int vec_ldl (int, unsigned int *);
5138 void unsigned int vec_ldl (int, vector unsigned int *);
5139 void signed short vec_ldl (int, vector signed short *);
5140 void signed short vec_ldl (int, short *);
5141 void unsigned short vec_ldl (int, vector unsigned short *);
5142 void unsigned short vec_ldl (int, unsigned short *);
5143 void signed char vec_ldl (int, vector signed char *);
5144 void signed char vec_ldl (int, signed char *);
5145 void unsigned char vec_ldl (int, vector unsigned char *);
5146 void unsigned char vec_ldl (int, unsigned char *);
5148 vector float vec_loge (vector float);
5150 vector unsigned char vec_lvsl (int, void *, int *);
5152 vector unsigned char vec_lvsr (int, void *, int *);
5154 vector float vec_madd (vector float, vector float, vector float);
5156 vector signed short vec_madds (vector signed short, vector signed short,
5157 vector signed short);
5159 vector unsigned char vec_max (vector signed char, vector unsigned char);
5161 vector unsigned char vec_max (vector unsigned char, vector signed char);
5163 vector unsigned char vec_max (vector unsigned char,
5164 vector unsigned char);
5165 vector signed char vec_max (vector signed char, vector signed char);
5166 vector unsigned short vec_max (vector signed short,
5167 vector unsigned short);
5168 vector unsigned short vec_max (vector unsigned short,
5169 vector signed short);
5170 vector unsigned short vec_max (vector unsigned short,
5171 vector unsigned short);
5172 vector signed short vec_max (vector signed short, vector signed short);
5173 vector unsigned int vec_max (vector signed int, vector unsigned int);
5174 vector unsigned int vec_max (vector unsigned int, vector signed int);
5175 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
5176 vector signed int vec_max (vector signed int, vector signed int);
5177 vector float vec_max (vector float, vector float);
5179 vector signed char vec_mergeh (vector signed char, vector signed char);
5180 vector unsigned char vec_mergeh (vector unsigned char,
5181 vector unsigned char);
5182 vector signed short vec_mergeh (vector signed short,
5183 vector signed short);
5184 vector unsigned short vec_mergeh (vector unsigned short,
5185 vector unsigned short);
5186 vector float vec_mergeh (vector float, vector float);
5187 vector signed int vec_mergeh (vector signed int, vector signed int);
5188 vector unsigned int vec_mergeh (vector unsigned int,
5189 vector unsigned int);
5191 vector signed char vec_mergel (vector signed char, vector signed char);
5192 vector unsigned char vec_mergel (vector unsigned char,
5193 vector unsigned char);
5194 vector signed short vec_mergel (vector signed short,
5195 vector signed short);
5196 vector unsigned short vec_mergel (vector unsigned short,
5197 vector unsigned short);
5198 vector float vec_mergel (vector float, vector float);
5199 vector signed int vec_mergel (vector signed int, vector signed int);
5200 vector unsigned int vec_mergel (vector unsigned int,
5201 vector unsigned int);
5203 vector unsigned short vec_mfvscr (void);
5205 vector unsigned char vec_min (vector signed char, vector unsigned char);
5207 vector unsigned char vec_min (vector unsigned char, vector signed char);
5209 vector unsigned char vec_min (vector unsigned char,
5210 vector unsigned char);
5211 vector signed char vec_min (vector signed char, vector signed char);
5212 vector unsigned short vec_min (vector signed short,
5213 vector unsigned short);
5214 vector unsigned short vec_min (vector unsigned short,
5215 vector signed short);
5216 vector unsigned short vec_min (vector unsigned short,
5217 vector unsigned short);
5218 vector signed short vec_min (vector signed short, vector signed short);
5219 vector unsigned int vec_min (vector signed int, vector unsigned int);
5220 vector unsigned int vec_min (vector unsigned int, vector signed int);
5221 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
5222 vector signed int vec_min (vector signed int, vector signed int);
5223 vector float vec_min (vector float, vector float);
5225 vector signed short vec_mladd (vector signed short, vector signed short,
5226 vector signed short);
5227 vector signed short vec_mladd (vector signed short,
5228 vector unsigned short,
5229 vector unsigned short);
5230 vector signed short vec_mladd (vector unsigned short,
5231 vector signed short,
5232 vector signed short);
5233 vector unsigned short vec_mladd (vector unsigned short,
5234 vector unsigned short,
5235 vector unsigned short);
5237 vector signed short vec_mradds (vector signed short,
5238 vector signed short,
5239 vector signed short);
5241 vector unsigned int vec_msum (vector unsigned char,
5242 vector unsigned char,
5243 vector unsigned int);
5244 vector signed int vec_msum (vector signed char, vector unsigned char,
5246 vector unsigned int vec_msum (vector unsigned short,
5247 vector unsigned short,
5248 vector unsigned int);
5249 vector signed int vec_msum (vector signed short, vector signed short,
5252 vector unsigned int vec_msums (vector unsigned short,
5253 vector unsigned short,
5254 vector unsigned int);
5255 vector signed int vec_msums (vector signed short, vector signed short,
5258 void vec_mtvscr (vector signed int);
5259 void vec_mtvscr (vector unsigned int);
5260 void vec_mtvscr (vector signed short);
5261 void vec_mtvscr (vector unsigned short);
5262 void vec_mtvscr (vector signed char);
5263 void vec_mtvscr (vector unsigned char);
5265 vector unsigned short vec_mule (vector unsigned char,
5266 vector unsigned char);
5267 vector signed short vec_mule (vector signed char, vector signed char);
5268 vector unsigned int vec_mule (vector unsigned short,
5269 vector unsigned short);
5270 vector signed int vec_mule (vector signed short, vector signed short);
5272 vector unsigned short vec_mulo (vector unsigned char,
5273 vector unsigned char);
5274 vector signed short vec_mulo (vector signed char, vector signed char);
5275 vector unsigned int vec_mulo (vector unsigned short,
5276 vector unsigned short);
5277 vector signed int vec_mulo (vector signed short, vector signed short);
5279 vector float vec_nmsub (vector float, vector float, vector float);
5281 vector float vec_nor (vector float, vector float);
5282 vector signed int vec_nor (vector signed int, vector signed int);
5283 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
5284 vector signed short vec_nor (vector signed short, vector signed short);
5285 vector unsigned short vec_nor (vector unsigned short,
5286 vector unsigned short);
5287 vector signed char vec_nor (vector signed char, vector signed char);
5288 vector unsigned char vec_nor (vector unsigned char,
5289 vector unsigned char);
5291 vector float vec_or (vector float, vector float);
5292 vector float vec_or (vector float, vector signed int);
5293 vector float vec_or (vector signed int, vector float);
5294 vector signed int vec_or (vector signed int, vector signed int);
5295 vector unsigned int vec_or (vector signed int, vector unsigned int);
5296 vector unsigned int vec_or (vector unsigned int, vector signed int);
5297 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
5298 vector signed short vec_or (vector signed short, vector signed short);
5299 vector unsigned short vec_or (vector signed short,
5300 vector unsigned short);
5301 vector unsigned short vec_or (vector unsigned short,
5302 vector signed short);
5303 vector unsigned short vec_or (vector unsigned short,
5304 vector unsigned short);
5305 vector signed char vec_or (vector signed char, vector signed char);
5306 vector unsigned char vec_or (vector signed char, vector unsigned char);
5307 vector unsigned char vec_or (vector unsigned char, vector signed char);
5308 vector unsigned char vec_or (vector unsigned char,
5309 vector unsigned char);
5311 vector signed char vec_pack (vector signed short, vector signed short);
5312 vector unsigned char vec_pack (vector unsigned short,
5313 vector unsigned short);
5314 vector signed short vec_pack (vector signed int, vector signed int);
5315 vector unsigned short vec_pack (vector unsigned int,
5316 vector unsigned int);
5318 vector signed short vec_packpx (vector unsigned int,
5319 vector unsigned int);
5321 vector unsigned char vec_packs (vector unsigned short,
5322 vector unsigned short);
5323 vector signed char vec_packs (vector signed short, vector signed short);
5325 vector unsigned short vec_packs (vector unsigned int,
5326 vector unsigned int);
5327 vector signed short vec_packs (vector signed int, vector signed int);
5329 vector unsigned char vec_packsu (vector unsigned short,
5330 vector unsigned short);
5331 vector unsigned char vec_packsu (vector signed short,
5332 vector signed short);
5333 vector unsigned short vec_packsu (vector unsigned int,
5334 vector unsigned int);
5335 vector unsigned short vec_packsu (vector signed int, vector signed int);
5337 vector float vec_perm (vector float, vector float,
5338 vector unsigned char);
5339 vector signed int vec_perm (vector signed int, vector signed int,
5340 vector unsigned char);
5341 vector unsigned int vec_perm (vector unsigned int, vector unsigned int,
5342 vector unsigned char);
5343 vector signed short vec_perm (vector signed short, vector signed short,
5344 vector unsigned char);
5345 vector unsigned short vec_perm (vector unsigned short,
5346 vector unsigned short,
5347 vector unsigned char);
5348 vector signed char vec_perm (vector signed char, vector signed char,
5349 vector unsigned char);
5350 vector unsigned char vec_perm (vector unsigned char,
5351 vector unsigned char,
5352 vector unsigned char);
5354 vector float vec_re (vector float);
5356 vector signed char vec_rl (vector signed char, vector unsigned char);
5357 vector unsigned char vec_rl (vector unsigned char,
5358 vector unsigned char);
5359 vector signed short vec_rl (vector signed short, vector unsigned short);
5361 vector unsigned short vec_rl (vector unsigned short,
5362 vector unsigned short);
5363 vector signed int vec_rl (vector signed int, vector unsigned int);
5364 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
5366 vector float vec_round (vector float);
5368 vector float vec_rsqrte (vector float);
5370 vector float vec_sel (vector float, vector float, vector signed int);
5371 vector float vec_sel (vector float, vector float, vector unsigned int);
5372 vector signed int vec_sel (vector signed int, vector signed int,
5374 vector signed int vec_sel (vector signed int, vector signed int,
5375 vector unsigned int);
5376 vector unsigned int vec_sel (vector unsigned int, vector unsigned int,
5378 vector unsigned int vec_sel (vector unsigned int, vector unsigned int,
5379 vector unsigned int);
5380 vector signed short vec_sel (vector signed short, vector signed short,
5381 vector signed short);
5382 vector signed short vec_sel (vector signed short, vector signed short,
5383 vector unsigned short);
5384 vector unsigned short vec_sel (vector unsigned short,
5385 vector unsigned short,
5386 vector signed short);
5387 vector unsigned short vec_sel (vector unsigned short,
5388 vector unsigned short,
5389 vector unsigned short);
5390 vector signed char vec_sel (vector signed char, vector signed char,
5391 vector signed char);
5392 vector signed char vec_sel (vector signed char, vector signed char,
5393 vector unsigned char);
5394 vector unsigned char vec_sel (vector unsigned char,
5395 vector unsigned char,
5396 vector signed char);
5397 vector unsigned char vec_sel (vector unsigned char,
5398 vector unsigned char,
5399 vector unsigned char);
5401 vector signed char vec_sl (vector signed char, vector unsigned char);
5402 vector unsigned char vec_sl (vector unsigned char,
5403 vector unsigned char);
5404 vector signed short vec_sl (vector signed short, vector unsigned short);
5406 vector unsigned short vec_sl (vector unsigned short,
5407 vector unsigned short);
5408 vector signed int vec_sl (vector signed int, vector unsigned int);
5409 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
5411 vector float vec_sld (vector float, vector float, const char);
5412 vector signed int vec_sld (vector signed int, vector signed int,
5414 vector unsigned int vec_sld (vector unsigned int, vector unsigned int,
5416 vector signed short vec_sld (vector signed short, vector signed short,
5418 vector unsigned short vec_sld (vector unsigned short,
5419 vector unsigned short, const char);
5420 vector signed char vec_sld (vector signed char, vector signed char,
5422 vector unsigned char vec_sld (vector unsigned char,
5423 vector unsigned char,
5426 vector signed int vec_sll (vector signed int, vector unsigned int);
5427 vector signed int vec_sll (vector signed int, vector unsigned short);
5428 vector signed int vec_sll (vector signed int, vector unsigned char);
5429 vector unsigned int vec_sll (vector unsigned int, vector unsigned int);
5430 vector unsigned int vec_sll (vector unsigned int,
5431 vector unsigned short);
5432 vector unsigned int vec_sll (vector unsigned int, vector unsigned char);
5434 vector signed short vec_sll (vector signed short, vector unsigned int);
5435 vector signed short vec_sll (vector signed short,
5436 vector unsigned short);
5437 vector signed short vec_sll (vector signed short, vector unsigned char);
5439 vector unsigned short vec_sll (vector unsigned short,
5440 vector unsigned int);
5441 vector unsigned short vec_sll (vector unsigned short,
5442 vector unsigned short);
5443 vector unsigned short vec_sll (vector unsigned short,
5444 vector unsigned char);
5445 vector signed char vec_sll (vector signed char, vector unsigned int);
5446 vector signed char vec_sll (vector signed char, vector unsigned short);
5447 vector signed char vec_sll (vector signed char, vector unsigned char);
5448 vector unsigned char vec_sll (vector unsigned char,
5449 vector unsigned int);
5450 vector unsigned char vec_sll (vector unsigned char,
5451 vector unsigned short);
5452 vector unsigned char vec_sll (vector unsigned char,
5453 vector unsigned char);
5455 vector float vec_slo (vector float, vector signed char);
5456 vector float vec_slo (vector float, vector unsigned char);
5457 vector signed int vec_slo (vector signed int, vector signed char);
5458 vector signed int vec_slo (vector signed int, vector unsigned char);
5459 vector unsigned int vec_slo (vector unsigned int, vector signed char);
5460 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
5462 vector signed short vec_slo (vector signed short, vector signed char);
5463 vector signed short vec_slo (vector signed short, vector unsigned char);
5465 vector unsigned short vec_slo (vector unsigned short,
5466 vector signed char);
5467 vector unsigned short vec_slo (vector unsigned short,
5468 vector unsigned char);
5469 vector signed char vec_slo (vector signed char, vector signed char);
5470 vector signed char vec_slo (vector signed char, vector unsigned char);
5471 vector unsigned char vec_slo (vector unsigned char, vector signed char);
5473 vector unsigned char vec_slo (vector unsigned char,
5474 vector unsigned char);
5476 vector signed char vec_splat (vector signed char, const char);
5477 vector unsigned char vec_splat (vector unsigned char, const char);
5478 vector signed short vec_splat (vector signed short, const char);
5479 vector unsigned short vec_splat (vector unsigned short, const char);
5480 vector float vec_splat (vector float, const char);
5481 vector signed int vec_splat (vector signed int, const char);
5482 vector unsigned int vec_splat (vector unsigned int, const char);
5484 vector signed char vec_splat_s8 (const char);
5486 vector signed short vec_splat_s16 (const char);
5488 vector signed int vec_splat_s32 (const char);
5490 vector unsigned char vec_splat_u8 (const char);
5492 vector unsigned short vec_splat_u16 (const char);
5494 vector unsigned int vec_splat_u32 (const char);
5496 vector signed char vec_sr (vector signed char, vector unsigned char);
5497 vector unsigned char vec_sr (vector unsigned char,
5498 vector unsigned char);
5499 vector signed short vec_sr (vector signed short, vector unsigned short);
5501 vector unsigned short vec_sr (vector unsigned short,
5502 vector unsigned short);
5503 vector signed int vec_sr (vector signed int, vector unsigned int);
5504 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
5506 vector signed char vec_sra (vector signed char, vector unsigned char);
5507 vector unsigned char vec_sra (vector unsigned char,
5508 vector unsigned char);
5509 vector signed short vec_sra (vector signed short,
5510 vector unsigned short);
5511 vector unsigned short vec_sra (vector unsigned short,
5512 vector unsigned short);
5513 vector signed int vec_sra (vector signed int, vector unsigned int);
5514 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
5516 vector signed int vec_srl (vector signed int, vector unsigned int);
5517 vector signed int vec_srl (vector signed int, vector unsigned short);
5518 vector signed int vec_srl (vector signed int, vector unsigned char);
5519 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
5520 vector unsigned int vec_srl (vector unsigned int,
5521 vector unsigned short);
5522 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
5524 vector signed short vec_srl (vector signed short, vector unsigned int);
5525 vector signed short vec_srl (vector signed short,
5526 vector unsigned short);
5527 vector signed short vec_srl (vector signed short, vector unsigned char);
5529 vector unsigned short vec_srl (vector unsigned short,
5530 vector unsigned int);
5531 vector unsigned short vec_srl (vector unsigned short,
5532 vector unsigned short);
5533 vector unsigned short vec_srl (vector unsigned short,
5534 vector unsigned char);
5535 vector signed char vec_srl (vector signed char, vector unsigned int);
5536 vector signed char vec_srl (vector signed char, vector unsigned short);
5537 vector signed char vec_srl (vector signed char, vector unsigned char);
5538 vector unsigned char vec_srl (vector unsigned char,
5539 vector unsigned int);
5540 vector unsigned char vec_srl (vector unsigned char,
5541 vector unsigned short);
5542 vector unsigned char vec_srl (vector unsigned char,
5543 vector unsigned char);
5545 vector float vec_sro (vector float, vector signed char);
5546 vector float vec_sro (vector float, vector unsigned char);
5547 vector signed int vec_sro (vector signed int, vector signed char);
5548 vector signed int vec_sro (vector signed int, vector unsigned char);
5549 vector unsigned int vec_sro (vector unsigned int, vector signed char);
5550 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
5552 vector signed short vec_sro (vector signed short, vector signed char);
5553 vector signed short vec_sro (vector signed short, vector unsigned char);
5555 vector unsigned short vec_sro (vector unsigned short,
5556 vector signed char);
5557 vector unsigned short vec_sro (vector unsigned short,
5558 vector unsigned char);
5559 vector signed char vec_sro (vector signed char, vector signed char);
5560 vector signed char vec_sro (vector signed char, vector unsigned char);
5561 vector unsigned char vec_sro (vector unsigned char, vector signed char);
5563 vector unsigned char vec_sro (vector unsigned char,
5564 vector unsigned char);
5566 void vec_st (vector float, int, float *);
5567 void vec_st (vector float, int, vector float *);
5568 void vec_st (vector signed int, int, int *);
5569 void vec_st (vector signed int, int, unsigned int *);
5570 void vec_st (vector unsigned int, int, unsigned int *);
5571 void vec_st (vector unsigned int, int, vector unsigned int *);
5572 void vec_st (vector signed short, int, short *);
5573 void vec_st (vector signed short, int, vector unsigned short *);
5574 void vec_st (vector signed short, int, vector signed short *);
5575 void vec_st (vector unsigned short, int, unsigned short *);
5576 void vec_st (vector unsigned short, int, vector unsigned short *);
5577 void vec_st (vector signed char, int, signed char *);
5578 void vec_st (vector signed char, int, unsigned char *);
5579 void vec_st (vector signed char, int, vector signed char *);
5580 void vec_st (vector unsigned char, int, unsigned char *);
5581 void vec_st (vector unsigned char, int, vector unsigned char *);
5583 void vec_ste (vector signed char, int, unsigned char *);
5584 void vec_ste (vector signed char, int, signed char *);
5585 void vec_ste (vector unsigned char, int, unsigned char *);
5586 void vec_ste (vector signed short, int, short *);
5587 void vec_ste (vector signed short, int, unsigned short *);
5588 void vec_ste (vector unsigned short, int, void *);
5589 void vec_ste (vector signed int, int, unsigned int *);
5590 void vec_ste (vector signed int, int, int *);
5591 void vec_ste (vector unsigned int, int, unsigned int *);
5592 void vec_ste (vector float, int, float *);
5594 void vec_stl (vector float, int, vector float *);
5595 void vec_stl (vector float, int, float *);
5596 void vec_stl (vector signed int, int, vector signed int *);
5597 void vec_stl (vector signed int, int, int *);
5598 void vec_stl (vector signed int, int, unsigned int *);
5599 void vec_stl (vector unsigned int, int, vector unsigned int *);
5600 void vec_stl (vector unsigned int, int, unsigned int *);
5601 void vec_stl (vector signed short, int, short *);
5602 void vec_stl (vector signed short, int, unsigned short *);
5603 void vec_stl (vector signed short, int, vector signed short *);
5604 void vec_stl (vector unsigned short, int, unsigned short *);
5605 void vec_stl (vector unsigned short, int, vector signed short *);
5606 void vec_stl (vector signed char, int, signed char *);
5607 void vec_stl (vector signed char, int, unsigned char *);
5608 void vec_stl (vector signed char, int, vector signed char *);
5609 void vec_stl (vector unsigned char, int, unsigned char *);
5610 void vec_stl (vector unsigned char, int, vector unsigned char *);
5612 vector signed char vec_sub (vector signed char, vector signed char);
5613 vector unsigned char vec_sub (vector signed char, vector unsigned char);
5615 vector unsigned char vec_sub (vector unsigned char, vector signed char);
5617 vector unsigned char vec_sub (vector unsigned char,
5618 vector unsigned char);
5619 vector signed short vec_sub (vector signed short, vector signed short);
5620 vector unsigned short vec_sub (vector signed short,
5621 vector unsigned short);
5622 vector unsigned short vec_sub (vector unsigned short,
5623 vector signed short);
5624 vector unsigned short vec_sub (vector unsigned short,
5625 vector unsigned short);
5626 vector signed int vec_sub (vector signed int, vector signed int);
5627 vector unsigned int vec_sub (vector signed int, vector unsigned int);
5628 vector unsigned int vec_sub (vector unsigned int, vector signed int);
5629 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
5630 vector float vec_sub (vector float, vector float);
5632 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
5634 vector unsigned char vec_subs (vector signed char,
5635 vector unsigned char);
5636 vector unsigned char vec_subs (vector unsigned char,
5637 vector signed char);
5638 vector unsigned char vec_subs (vector unsigned char,
5639 vector unsigned char);
5640 vector signed char vec_subs (vector signed char, vector signed char);
5641 vector unsigned short vec_subs (vector signed short,
5642 vector unsigned short);
5643 vector unsigned short vec_subs (vector unsigned short,
5644 vector signed short);
5645 vector unsigned short vec_subs (vector unsigned short,
5646 vector unsigned short);
5647 vector signed short vec_subs (vector signed short, vector signed short);
5649 vector unsigned int vec_subs (vector signed int, vector unsigned int);
5650 vector unsigned int vec_subs (vector unsigned int, vector signed int);
5651 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
5653 vector signed int vec_subs (vector signed int, vector signed int);
5655 vector unsigned int vec_sum4s (vector unsigned char,
5656 vector unsigned int);
5657 vector signed int vec_sum4s (vector signed char, vector signed int);
5658 vector signed int vec_sum4s (vector signed short, vector signed int);
5660 vector signed int vec_sum2s (vector signed int, vector signed int);
5662 vector signed int vec_sums (vector signed int, vector signed int);
5664 vector float vec_trunc (vector float);
5666 vector signed short vec_unpackh (vector signed char);
5667 vector unsigned int vec_unpackh (vector signed short);
5668 vector signed int vec_unpackh (vector signed short);
5670 vector signed short vec_unpackl (vector signed char);
5671 vector unsigned int vec_unpackl (vector signed short);
5672 vector signed int vec_unpackl (vector signed short);
5674 vector float vec_xor (vector float, vector float);
5675 vector float vec_xor (vector float, vector signed int);
5676 vector float vec_xor (vector signed int, vector float);
5677 vector signed int vec_xor (vector signed int, vector signed int);
5678 vector unsigned int vec_xor (vector signed int, vector unsigned int);
5679 vector unsigned int vec_xor (vector unsigned int, vector signed int);
5680 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
5681 vector signed short vec_xor (vector signed short, vector signed short);
5682 vector unsigned short vec_xor (vector signed short,
5683 vector unsigned short);
5684 vector unsigned short vec_xor (vector unsigned short,
5685 vector signed short);
5686 vector unsigned short vec_xor (vector unsigned short,
5687 vector unsigned short);
5688 vector signed char vec_xor (vector signed char, vector signed char);
5689 vector unsigned char vec_xor (vector signed char, vector unsigned char);
5691 vector unsigned char vec_xor (vector unsigned char, vector signed char);
5693 vector unsigned char vec_xor (vector unsigned char,
5694 vector unsigned char);
5696 vector signed int vec_all_eq (vector signed char, vector unsigned char);
5698 vector signed int vec_all_eq (vector signed char, vector signed char);
5699 vector signed int vec_all_eq (vector unsigned char, vector signed char);
5701 vector signed int vec_all_eq (vector unsigned char,
5702 vector unsigned char);
5703 vector signed int vec_all_eq (vector signed short,
5704 vector unsigned short);
5705 vector signed int vec_all_eq (vector signed short, vector signed short);
5707 vector signed int vec_all_eq (vector unsigned short,
5708 vector signed short);
5709 vector signed int vec_all_eq (vector unsigned short,
5710 vector unsigned short);
5711 vector signed int vec_all_eq (vector signed int, vector unsigned int);
5712 vector signed int vec_all_eq (vector signed int, vector signed int);
5713 vector signed int vec_all_eq (vector unsigned int, vector signed int);
5714 vector signed int vec_all_eq (vector unsigned int, vector unsigned int);
5716 vector signed int vec_all_eq (vector float, vector float);
5718 vector signed int vec_all_ge (vector signed char, vector unsigned char);
5720 vector signed int vec_all_ge (vector unsigned char, vector signed char);
5722 vector signed int vec_all_ge (vector unsigned char,
5723 vector unsigned char);
5724 vector signed int vec_all_ge (vector signed char, vector signed char);
5725 vector signed int vec_all_ge (vector signed short,
5726 vector unsigned short);
5727 vector signed int vec_all_ge (vector unsigned short,
5728 vector signed short);
5729 vector signed int vec_all_ge (vector unsigned short,
5730 vector unsigned short);
5731 vector signed int vec_all_ge (vector signed short, vector signed short);
5733 vector signed int vec_all_ge (vector signed int, vector unsigned int);
5734 vector signed int vec_all_ge (vector unsigned int, vector signed int);
5735 vector signed int vec_all_ge (vector unsigned int, vector unsigned int);
5737 vector signed int vec_all_ge (vector signed int, vector signed int);
5738 vector signed int vec_all_ge (vector float, vector float);
5740 vector signed int vec_all_gt (vector signed char, vector unsigned char);
5742 vector signed int vec_all_gt (vector unsigned char, vector signed char);
5744 vector signed int vec_all_gt (vector unsigned char,
5745 vector unsigned char);
5746 vector signed int vec_all_gt (vector signed char, vector signed char);
5747 vector signed int vec_all_gt (vector signed short,
5748 vector unsigned short);
5749 vector signed int vec_all_gt (vector unsigned short,
5750 vector signed short);
5751 vector signed int vec_all_gt (vector unsigned short,
5752 vector unsigned short);
5753 vector signed int vec_all_gt (vector signed short, vector signed short);
5755 vector signed int vec_all_gt (vector signed int, vector unsigned int);
5756 vector signed int vec_all_gt (vector unsigned int, vector signed int);
5757 vector signed int vec_all_gt (vector unsigned int, vector unsigned int);
5759 vector signed int vec_all_gt (vector signed int, vector signed int);
5760 vector signed int vec_all_gt (vector float, vector float);
5762 vector signed int vec_all_in (vector float, vector float);
5764 vector signed int vec_all_le (vector signed char, vector unsigned char);
5766 vector signed int vec_all_le (vector unsigned char, vector signed char);
5768 vector signed int vec_all_le (vector unsigned char,
5769 vector unsigned char);
5770 vector signed int vec_all_le (vector signed char, vector signed char);
5771 vector signed int vec_all_le (vector signed short,
5772 vector unsigned short);
5773 vector signed int vec_all_le (vector unsigned short,
5774 vector signed short);
5775 vector signed int vec_all_le (vector unsigned short,
5776 vector unsigned short);
5777 vector signed int vec_all_le (vector signed short, vector signed short);
5779 vector signed int vec_all_le (vector signed int, vector unsigned int);
5780 vector signed int vec_all_le (vector unsigned int, vector signed int);
5781 vector signed int vec_all_le (vector unsigned int, vector unsigned int);
5783 vector signed int vec_all_le (vector signed int, vector signed int);
5784 vector signed int vec_all_le (vector float, vector float);
5786 vector signed int vec_all_lt (vector signed char, vector unsigned char);
5788 vector signed int vec_all_lt (vector unsigned char, vector signed char);
5790 vector signed int vec_all_lt (vector unsigned char,
5791 vector unsigned char);
5792 vector signed int vec_all_lt (vector signed char, vector signed char);
5793 vector signed int vec_all_lt (vector signed short,
5794 vector unsigned short);
5795 vector signed int vec_all_lt (vector unsigned short,
5796 vector signed short);
5797 vector signed int vec_all_lt (vector unsigned short,
5798 vector unsigned short);
5799 vector signed int vec_all_lt (vector signed short, vector signed short);
5801 vector signed int vec_all_lt (vector signed int, vector unsigned int);
5802 vector signed int vec_all_lt (vector unsigned int, vector signed int);
5803 vector signed int vec_all_lt (vector unsigned int, vector unsigned int);
5805 vector signed int vec_all_lt (vector signed int, vector signed int);
5806 vector signed int vec_all_lt (vector float, vector float);
5808 vector signed int vec_all_nan (vector float);
5810 vector signed int vec_all_ne (vector signed char, vector unsigned char);
5812 vector signed int vec_all_ne (vector signed char, vector signed char);
5813 vector signed int vec_all_ne (vector unsigned char, vector signed char);
5815 vector signed int vec_all_ne (vector unsigned char,
5816 vector unsigned char);
5817 vector signed int vec_all_ne (vector signed short,
5818 vector unsigned short);
5819 vector signed int vec_all_ne (vector signed short, vector signed short);
5821 vector signed int vec_all_ne (vector unsigned short,
5822 vector signed short);
5823 vector signed int vec_all_ne (vector unsigned short,
5824 vector unsigned short);
5825 vector signed int vec_all_ne (vector signed int, vector unsigned int);
5826 vector signed int vec_all_ne (vector signed int, vector signed int);
5827 vector signed int vec_all_ne (vector unsigned int, vector signed int);
5828 vector signed int vec_all_ne (vector unsigned int, vector unsigned int);
5830 vector signed int vec_all_ne (vector float, vector float);
5832 vector signed int vec_all_nge (vector float, vector float);
5834 vector signed int vec_all_ngt (vector float, vector float);
5836 vector signed int vec_all_nle (vector float, vector float);
5838 vector signed int vec_all_nlt (vector float, vector float);
5840 vector signed int vec_all_numeric (vector float);
5842 vector signed int vec_any_eq (vector signed char, vector unsigned char);
5844 vector signed int vec_any_eq (vector signed char, vector signed char);
5845 vector signed int vec_any_eq (vector unsigned char, vector signed char);
5847 vector signed int vec_any_eq (vector unsigned char,
5848 vector unsigned char);
5849 vector signed int vec_any_eq (vector signed short,
5850 vector unsigned short);
5851 vector signed int vec_any_eq (vector signed short, vector signed short);
5853 vector signed int vec_any_eq (vector unsigned short,
5854 vector signed short);
5855 vector signed int vec_any_eq (vector unsigned short,
5856 vector unsigned short);
5857 vector signed int vec_any_eq (vector signed int, vector unsigned int);
5858 vector signed int vec_any_eq (vector signed int, vector signed int);
5859 vector signed int vec_any_eq (vector unsigned int, vector signed int);
5860 vector signed int vec_any_eq (vector unsigned int, vector unsigned int);
5862 vector signed int vec_any_eq (vector float, vector float);
5864 vector signed int vec_any_ge (vector signed char, vector unsigned char);
5866 vector signed int vec_any_ge (vector unsigned char, vector signed char);
5868 vector signed int vec_any_ge (vector unsigned char,
5869 vector unsigned char);
5870 vector signed int vec_any_ge (vector signed char, vector signed char);
5871 vector signed int vec_any_ge (vector signed short,
5872 vector unsigned short);
5873 vector signed int vec_any_ge (vector unsigned short,
5874 vector signed short);
5875 vector signed int vec_any_ge (vector unsigned short,
5876 vector unsigned short);
5877 vector signed int vec_any_ge (vector signed short, vector signed short);
5879 vector signed int vec_any_ge (vector signed int, vector unsigned int);
5880 vector signed int vec_any_ge (vector unsigned int, vector signed int);
5881 vector signed int vec_any_ge (vector unsigned int, vector unsigned int);
5883 vector signed int vec_any_ge (vector signed int, vector signed int);
5884 vector signed int vec_any_ge (vector float, vector float);
5886 vector signed int vec_any_gt (vector signed char, vector unsigned char);
5888 vector signed int vec_any_gt (vector unsigned char, vector signed char);
5890 vector signed int vec_any_gt (vector unsigned char,
5891 vector unsigned char);
5892 vector signed int vec_any_gt (vector signed char, vector signed char);
5893 vector signed int vec_any_gt (vector signed short,
5894 vector unsigned short);
5895 vector signed int vec_any_gt (vector unsigned short,
5896 vector signed short);
5897 vector signed int vec_any_gt (vector unsigned short,
5898 vector unsigned short);
5899 vector signed int vec_any_gt (vector signed short, vector signed short);
5901 vector signed int vec_any_gt (vector signed int, vector unsigned int);
5902 vector signed int vec_any_gt (vector unsigned int, vector signed int);
5903 vector signed int vec_any_gt (vector unsigned int, vector unsigned int);
5905 vector signed int vec_any_gt (vector signed int, vector signed int);
5906 vector signed int vec_any_gt (vector float, vector float);
5908 vector signed int vec_any_le (vector signed char, vector unsigned char);
5910 vector signed int vec_any_le (vector unsigned char, vector signed char);
5912 vector signed int vec_any_le (vector unsigned char,
5913 vector unsigned char);
5914 vector signed int vec_any_le (vector signed char, vector signed char);
5915 vector signed int vec_any_le (vector signed short,
5916 vector unsigned short);
5917 vector signed int vec_any_le (vector unsigned short,
5918 vector signed short);
5919 vector signed int vec_any_le (vector unsigned short,
5920 vector unsigned short);
5921 vector signed int vec_any_le (vector signed short, vector signed short);
5923 vector signed int vec_any_le (vector signed int, vector unsigned int);
5924 vector signed int vec_any_le (vector unsigned int, vector signed int);
5925 vector signed int vec_any_le (vector unsigned int, vector unsigned int);
5927 vector signed int vec_any_le (vector signed int, vector signed int);
5928 vector signed int vec_any_le (vector float, vector float);
5930 vector signed int vec_any_lt (vector signed char, vector unsigned char);
5932 vector signed int vec_any_lt (vector unsigned char, vector signed char);
5934 vector signed int vec_any_lt (vector unsigned char,
5935 vector unsigned char);
5936 vector signed int vec_any_lt (vector signed char, vector signed char);
5937 vector signed int vec_any_lt (vector signed short,
5938 vector unsigned short);
5939 vector signed int vec_any_lt (vector unsigned short,
5940 vector signed short);
5941 vector signed int vec_any_lt (vector unsigned short,
5942 vector unsigned short);
5943 vector signed int vec_any_lt (vector signed short, vector signed short);
5945 vector signed int vec_any_lt (vector signed int, vector unsigned int);
5946 vector signed int vec_any_lt (vector unsigned int, vector signed int);
5947 vector signed int vec_any_lt (vector unsigned int, vector unsigned int);
5949 vector signed int vec_any_lt (vector signed int, vector signed int);
5950 vector signed int vec_any_lt (vector float, vector float);
5952 vector signed int vec_any_nan (vector float);
5954 vector signed int vec_any_ne (vector signed char, vector unsigned char);
5956 vector signed int vec_any_ne (vector signed char, vector signed char);
5957 vector signed int vec_any_ne (vector unsigned char, vector signed char);
5959 vector signed int vec_any_ne (vector unsigned char,
5960 vector unsigned char);
5961 vector signed int vec_any_ne (vector signed short,
5962 vector unsigned short);
5963 vector signed int vec_any_ne (vector signed short, vector signed short);
5965 vector signed int vec_any_ne (vector unsigned short,
5966 vector signed short);
5967 vector signed int vec_any_ne (vector unsigned short,
5968 vector unsigned short);
5969 vector signed int vec_any_ne (vector signed int, vector unsigned int);
5970 vector signed int vec_any_ne (vector signed int, vector signed int);
5971 vector signed int vec_any_ne (vector unsigned int, vector signed int);
5972 vector signed int vec_any_ne (vector unsigned int, vector unsigned int);
5974 vector signed int vec_any_ne (vector float, vector float);
5976 vector signed int vec_any_nge (vector float, vector float);
5978 vector signed int vec_any_ngt (vector float, vector float);
5980 vector signed int vec_any_nle (vector float, vector float);
5982 vector signed int vec_any_nlt (vector float, vector float);
5984 vector signed int vec_any_numeric (vector float);
5986 vector signed int vec_any_out (vector float, vector float);
5990 @section Pragmas Accepted by GCC
5994 GCC supports several types of pragmas, primarily in order to compile
5995 code originally written for other compilers. Note that in general
5996 we do not recommend the use of pragmas; @xref{Function Attributes},
5997 for further explanation.
6001 * RS/6000 and PowerPC Pragmas::
6008 @subsection ARM Pragmas
6010 The ARM target defines pragmas for controlling the default addition of
6011 @code{long_call} and @code{short_call} attributes to functions.
6012 @xref{Function Attributes}, for information about the effects of these
6017 @cindex pragma, long_calls
6018 Set all subsequent functions to have the @code{long_call} attribute.
6021 @cindex pragma, no_long_calls
6022 Set all subsequent functions to have the @code{short_call} attribute.
6024 @item long_calls_off
6025 @cindex pragma, long_calls_off
6026 Do not affect the @code{long_call} or @code{short_call} attributes of
6027 subsequent functions.
6030 @node RS/6000 and PowerPC Pragmas
6031 @subsection RS/6000 and PowerPC Pragmas
6033 The RS/6000 and PowerPC targets define one pragma for controlling
6034 whether or not the @code{longcall} attribute is added to function
6035 declarations by default. This pragma overrides the @option{-mlongcall}
6036 option, but not the @code{longcall} and @code{shortcall} attributes.
6037 @xref{RS/6000 and PowerPC Options}, for more information about when long
6038 calls are and are not necessary.
6042 @cindex pragma, longcall
6043 Apply the @code{longcall} attribute to all subsequent function
6047 Do not apply the @code{longcall} attribute to subsequent function
6051 @c Describe c4x pragmas here.
6052 @c Describe h8300 pragmas here.
6053 @c Describe i370 pragmas here.
6054 @c Describe i960 pragmas here.
6055 @c Describe sh pragmas here.
6056 @c Describe v850 pragmas here.
6058 @node Darwin Pragmas
6059 @subsection Darwin Pragmas
6061 The following pragmas are available for all architectures running the
6062 Darwin operating system. These are useful for compatibility with other
6066 @item mark @var{tokens}@dots{}
6067 @cindex pragma, mark
6068 This pragma is accepted, but has no effect.
6070 @item options align=@var{alignment}
6071 @cindex pragma, options align
6072 This pragma sets the alignment of fields in structures. The values of
6073 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
6074 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
6075 properly; to restore the previous setting, use @code{reset} for the
6078 @item segment @var{tokens}@dots{}
6079 @cindex pragma, segment
6080 This pragma is accepted, but has no effect.
6082 @item unused (@var{var} [, @var{var}]@dots{})
6083 @cindex pragma, unused
6084 This pragma declares variables to be possibly unused. GCC will not
6085 produce warnings for the listed variables. The effect is similar to
6086 that of the @code{unused} attribute, except that this pragma may appear
6087 anywhere within the variables' scopes.
6090 @node Solaris Pragmas
6091 @subsection Solaris Pragmas
6093 For compatibility with the SunPRO compiler, the following pragma
6097 @item redefine_extname @var{oldname} @var{newname}
6098 @cindex pragma, redefine_extname
6100 This pragma gives the C function @var{oldname} the assembler label
6101 @var{newname}. The pragma must appear before the function declaration.
6102 This pragma is equivalent to the asm labels extension (@pxref{Asm
6103 Labels}). The preprocessor defines @code{__PRAGMA_REDEFINE_EXTNAME}
6104 if the pragma is available.
6108 @subsection Tru64 Pragmas
6110 For compatibility with the Compaq C compiler, the following pragma
6114 @item extern_prefix @var{string}
6115 @cindex pragma, extern_prefix
6117 This pragma renames all subsequent function and variable declarations
6118 such that @var{string} is prepended to the name. This effect may be
6119 terminated by using another @code{extern_prefix} pragma with the
6122 This pragma is similar in intent to to the asm labels extension
6123 (@pxref{Asm Labels}) in that the system programmer wants to change
6124 the assembly-level ABI without changing the source-level API. The
6125 preprocessor defines @code{__EXTERN_PREFIX} if the pragma is available.
6128 @node Unnamed Fields
6129 @section Unnamed struct/union fields within structs/unions.
6133 For compatibility with other compilers, GCC allows you to define
6134 a structure or union that contains, as fields, structures and unions
6135 without names. For example:
6148 In this example, the user would be able to access members of the unnamed
6149 union with code like @samp{foo.b}. Note that only unnamed structs and
6150 unions are allowed, you may not have, for example, an unnamed
6153 You must never create such structures that cause ambiguous field definitions.
6154 For example, this structure:
6165 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
6166 Such constructs are not supported and must be avoided. In the future,
6167 such constructs may be detected and treated as compilation errors.
6170 @section Thread-Local Storage
6171 @cindex Thread-Local Storage
6172 @cindex @acronym{TLS}
6175 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
6176 are allocated such that there is one instance of the variable per extant
6177 thread. The run-time model GCC uses to implement this originates
6178 in the IA-64 processor-specific ABI, but has since been migrated
6179 to other processors as well. It requires significant support from
6180 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
6181 system libraries (@file{libc.so} and @file{libpthread.so}), so it
6182 is not available everywhere.
6184 At the user level, the extension is visible with a new storage
6185 class keyword: @code{__thread}. For example:
6189 extern __thread struct state s;
6190 static __thread char *p;
6193 The @code{__thread} specifier may be used alone, with the @code{extern}
6194 or @code{static} specifiers, but with no other storage class specifier.
6195 When used with @code{extern} or @code{static}, @code{__thread} must appear
6196 immediately after the other storage class specifier.
6198 The @code{__thread} specifier may be applied to any global, file-scoped
6199 static, function-scoped static, or class-scoped static variable. It may
6200 not be applied to block-scoped automatic or class-scoped member variables.
6202 When the address-of operator is applied to a thread-local variable, it is
6203 evaluated at run-time and returns the address of the current thread's
6204 instance of that variable. An address so obtained may be used by any
6205 thread. When a thread terminates, any pointers to thread-local variables
6206 in that thread become invalid.
6208 No static initialization may refer to the address of a thread-local variable.
6210 In C++, a thread-local variable may not be initialized at runtime,
6211 that is, either by a static constructor or a non-constant expression.
6213 See @uref{http://people.redhat.com/drepper/tls.pdf,
6214 ELF Handling For Thread-Local Storage} for a detailed explanation of
6215 the four thread-local storage addressing models, and how the run-time
6216 is expected to function.
6219 * C99 Thread-Local Edits::
6220 * C++98 Thread-Local Edits::
6223 @node C99 Thread-Local Edits
6224 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
6226 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
6227 that document the exact semantics of the language extension.
6231 @cite{5.1.2 Execution environments}
6233 Add new text after paragraph 1
6236 Within either execution environment, a @dfn{thread} is a flow of
6237 control within a program. It is implementation defined whether
6238 or not there may be more than one thread associated with a program.
6239 It is implementation defined how threads beyond the first are
6240 created, the name and type of the function called at thread
6241 startup, and how threads may be terminated. However, objects
6242 with thread storage duration shall be initialized before thread
6247 @cite{6.2.4 Storage durations of objects}
6249 Add new text before paragraph 3
6252 An object whose identifier is declared with the storage-class
6253 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
6254 Its lifetime is the entire execution of the thread, and its
6255 stored value is initialized only once, prior to thread startup.
6259 @cite{6.4.1 Keywords}
6261 Add @code{__thread}.
6264 @cite{6.7.1 Storage-class specifiers}
6266 Add @code{__thread} to the list of storage class specifiers in
6269 Change paragraph 2 to
6272 With the exception of @code{__thread}, at most one storage-class
6273 specifier may be given [@dots{}]. The @code{__thread} specifier may
6274 be used alone, or immediately following @code{extern} or
6278 Add new text after paragraph 6
6281 The declaration of an identifier for a variable that has
6282 block scope that specifies @code{__thread} shall also
6283 specify either @code{extern} or @code{static}.
6285 The @code{__thread} specifier shall be used only with
6290 @node C++98 Thread-Local Edits
6291 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
6293 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
6294 that document the exact semantics of the language extension.
6297 @b{[intro.execution]}
6299 New text after paragraph 4
6302 A @dfn{thread} is a flow of control within the abstract machine.
6303 It is implementation defined whether or not there may be more than
6307 New text after paragraph 7
6310 It is unspecified whether additional action must be taken to
6311 ensure when and whether side effects are visible to other threads.
6317 Add @code{__thread}.
6320 @b{[basic.start.main]}
6322 Add after paragraph 5
6325 The thread that begins execution at the @code{main} function is called
6326 the @dfn{main thread}. It is implementation defined how functions
6327 beginning threads other than the main thread are designated or typed.
6328 A function so designated, as well as the @code{main} function, is called
6329 a @dfn{thread startup function}. It is implementation defined what
6330 happens if a thread startup function returns. It is implementation
6331 defined what happens to other threads when any thread calls @code{exit}.
6335 @b{[basic.start.init]}
6337 Add after paragraph 4
6340 The storage for an object of thread storage duration shall be
6341 staticly initialized before the first statement of the thread startup
6342 function. An object of thread storage duration shall not require
6343 dynamic initialization.
6347 @b{[basic.start.term]}
6349 Add after paragraph 3
6352 An object of thread storage duration shall not require a destructor.
6358 Add ``thread storage duration'' to the list in paragraph 1.
6363 Thread, static, and automatic storage durations are associated with
6364 objects introduced by declarations [@dots{}].
6367 Add @code{__thread} to the list of specifiers in paragraph 3.
6370 @b{[basic.stc.thread]}
6372 New section before @b{[basic.stc.static]}
6375 The keyword @code{__thread} applied to an non-local object gives the
6376 object thread storage duration.
6378 A local variable or class data member declared both @code{static}
6379 and @code{__thread} gives the variable or member thread storage
6384 @b{[basic.stc.static]}
6389 All objects which have neither thread storage duration, dynamic
6390 storage duration nor are local [@dots{}].
6396 Add @code{__thread} to the list in paragraph 1.
6401 With the exception of @code{__thread}, at most one
6402 @var{storage-class-specifier} shall appear in a given
6403 @var{decl-specifier-seq}. The @code{__thread} specifier may
6404 be used alone, or immediately following the @code{extern} or
6405 @code{static} specifiers. [@dots{}]
6408 Add after paragraph 5
6411 The @code{__thread} specifier can be applied only to the names of objects
6412 and to anonymous unions.
6418 Add after paragraph 6
6421 Non-@code{static} members shall not be @code{__thread}.
6425 @node C++ Extensions
6426 @chapter Extensions to the C++ Language
6427 @cindex extensions, C++ language
6428 @cindex C++ language extensions
6430 The GNU compiler provides these extensions to the C++ language (and you
6431 can also use most of the C language extensions in your C++ programs). If you
6432 want to write code that checks whether these features are available, you can
6433 test for the GNU compiler the same way as for C programs: check for a
6434 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
6435 test specifically for GNU C++ (@pxref{Standard Predefined,,Standard
6436 Predefined Macros,cpp.info,The C Preprocessor}).
6439 * Min and Max:: C++ Minimum and maximum operators.
6440 * Volatiles:: What constitutes an access to a volatile object.
6441 * Restricted Pointers:: C99 restricted pointers and references.
6442 * Vague Linkage:: Where G++ puts inlines, vtables and such.
6443 * C++ Interface:: You can use a single C++ header file for both
6444 declarations and definitions.
6445 * Template Instantiation:: Methods for ensuring that exactly one copy of
6446 each needed template instantiation is emitted.
6447 * Bound member functions:: You can extract a function pointer to the
6448 method denoted by a @samp{->*} or @samp{.*} expression.
6449 * C++ Attributes:: Variable, function, and type attributes for C++ only.
6450 * Java Exceptions:: Tweaking exception handling to work with Java.
6451 * Deprecated Features:: Things might disappear from g++.
6452 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
6456 @section Minimum and Maximum Operators in C++
6458 It is very convenient to have operators which return the ``minimum'' or the
6459 ``maximum'' of two arguments. In GNU C++ (but not in GNU C),
6462 @item @var{a} <? @var{b}
6464 @cindex minimum operator
6465 is the @dfn{minimum}, returning the smaller of the numeric values
6466 @var{a} and @var{b};
6468 @item @var{a} >? @var{b}
6470 @cindex maximum operator
6471 is the @dfn{maximum}, returning the larger of the numeric values @var{a}
6475 These operations are not primitive in ordinary C++, since you can
6476 use a macro to return the minimum of two things in C++, as in the
6480 #define MIN(X,Y) ((X) < (Y) ? : (X) : (Y))
6484 You might then use @w{@samp{int min = MIN (i, j);}} to set @var{min} to
6485 the minimum value of variables @var{i} and @var{j}.
6487 However, side effects in @code{X} or @code{Y} may cause unintended
6488 behavior. For example, @code{MIN (i++, j++)} will fail, incrementing
6489 the smaller counter twice. A GNU C extension allows you to write safe
6490 macros that avoid this kind of problem (@pxref{Naming Types,,Naming an
6491 Expression's Type}). However, writing @code{MIN} and @code{MAX} as
6492 macros also forces you to use function-call notation for a
6493 fundamental arithmetic operation. Using GNU C++ extensions, you can
6494 write @w{@samp{int min = i <? j;}} instead.
6496 Since @code{<?} and @code{>?} are built into the compiler, they properly
6497 handle expressions with side-effects; @w{@samp{int min = i++ <? j++;}}
6501 @section When is a Volatile Object Accessed?
6502 @cindex accessing volatiles
6503 @cindex volatile read
6504 @cindex volatile write
6505 @cindex volatile access
6507 Both the C and C++ standard have the concept of volatile objects. These
6508 are normally accessed by pointers and used for accessing hardware. The
6509 standards encourage compilers to refrain from optimizations
6510 concerning accesses to volatile objects that it might perform on
6511 non-volatile objects. The C standard leaves it implementation defined
6512 as to what constitutes a volatile access. The C++ standard omits to
6513 specify this, except to say that C++ should behave in a similar manner
6514 to C with respect to volatiles, where possible. The minimum either
6515 standard specifies is that at a sequence point all previous accesses to
6516 volatile objects have stabilized and no subsequent accesses have
6517 occurred. Thus an implementation is free to reorder and combine
6518 volatile accesses which occur between sequence points, but cannot do so
6519 for accesses across a sequence point. The use of volatiles does not
6520 allow you to violate the restriction on updating objects multiple times
6521 within a sequence point.
6523 In most expressions, it is intuitively obvious what is a read and what is
6524 a write. For instance
6527 volatile int *dst = @var{somevalue};
6528 volatile int *src = @var{someothervalue};
6533 will cause a read of the volatile object pointed to by @var{src} and stores the
6534 value into the volatile object pointed to by @var{dst}. There is no
6535 guarantee that these reads and writes are atomic, especially for objects
6536 larger than @code{int}.
6538 Less obvious expressions are where something which looks like an access
6539 is used in a void context. An example would be,
6542 volatile int *src = @var{somevalue};
6546 With C, such expressions are rvalues, and as rvalues cause a read of
6547 the object, GCC interprets this as a read of the volatile being pointed
6548 to. The C++ standard specifies that such expressions do not undergo
6549 lvalue to rvalue conversion, and that the type of the dereferenced
6550 object may be incomplete. The C++ standard does not specify explicitly
6551 that it is this lvalue to rvalue conversion which is responsible for
6552 causing an access. However, there is reason to believe that it is,
6553 because otherwise certain simple expressions become undefined. However,
6554 because it would surprise most programmers, G++ treats dereferencing a
6555 pointer to volatile object of complete type in a void context as a read
6556 of the object. When the object has incomplete type, G++ issues a
6561 struct T @{int m;@};
6562 volatile S *ptr1 = @var{somevalue};
6563 volatile T *ptr2 = @var{somevalue};
6568 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
6569 causes a read of the object pointed to. If you wish to force an error on
6570 the first case, you must force a conversion to rvalue with, for instance
6571 a static cast, @code{static_cast<S>(*ptr1)}.
6573 When using a reference to volatile, G++ does not treat equivalent
6574 expressions as accesses to volatiles, but instead issues a warning that
6575 no volatile is accessed. The rationale for this is that otherwise it
6576 becomes difficult to determine where volatile access occur, and not
6577 possible to ignore the return value from functions returning volatile
6578 references. Again, if you wish to force a read, cast the reference to
6581 @node Restricted Pointers
6582 @section Restricting Pointer Aliasing
6583 @cindex restricted pointers
6584 @cindex restricted references
6585 @cindex restricted this pointer
6587 As with gcc, g++ understands the C99 feature of restricted pointers,
6588 specified with the @code{__restrict__}, or @code{__restrict} type
6589 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
6590 language flag, @code{restrict} is not a keyword in C++.
6592 In addition to allowing restricted pointers, you can specify restricted
6593 references, which indicate that the reference is not aliased in the local
6597 void fn (int *__restrict__ rptr, int &__restrict__ rref)
6604 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
6605 @var{rref} refers to a (different) unaliased integer.
6607 You may also specify whether a member function's @var{this} pointer is
6608 unaliased by using @code{__restrict__} as a member function qualifier.
6611 void T::fn () __restrict__
6618 Within the body of @code{T::fn}, @var{this} will have the effective
6619 definition @code{T *__restrict__ const this}. Notice that the
6620 interpretation of a @code{__restrict__} member function qualifier is
6621 different to that of @code{const} or @code{volatile} qualifier, in that it
6622 is applied to the pointer rather than the object. This is consistent with
6623 other compilers which implement restricted pointers.
6625 As with all outermost parameter qualifiers, @code{__restrict__} is
6626 ignored in function definition matching. This means you only need to
6627 specify @code{__restrict__} in a function definition, rather than
6628 in a function prototype as well.
6631 @section Vague Linkage
6632 @cindex vague linkage
6634 There are several constructs in C++ which require space in the object
6635 file but are not clearly tied to a single translation unit. We say that
6636 these constructs have ``vague linkage''. Typically such constructs are
6637 emitted wherever they are needed, though sometimes we can be more
6641 @item Inline Functions
6642 Inline functions are typically defined in a header file which can be
6643 included in many different compilations. Hopefully they can usually be
6644 inlined, but sometimes an out-of-line copy is necessary, if the address
6645 of the function is taken or if inlining fails. In general, we emit an
6646 out-of-line copy in all translation units where one is needed. As an
6647 exception, we only emit inline virtual functions with the vtable, since
6648 it will always require a copy.
6650 Local static variables and string constants used in an inline function
6651 are also considered to have vague linkage, since they must be shared
6652 between all inlined and out-of-line instances of the function.
6656 C++ virtual functions are implemented in most compilers using a lookup
6657 table, known as a vtable. The vtable contains pointers to the virtual
6658 functions provided by a class, and each object of the class contains a
6659 pointer to its vtable (or vtables, in some multiple-inheritance
6660 situations). If the class declares any non-inline, non-pure virtual
6661 functions, the first one is chosen as the ``key method'' for the class,
6662 and the vtable is only emitted in the translation unit where the key
6665 @emph{Note:} If the chosen key method is later defined as inline, the
6666 vtable will still be emitted in every translation unit which defines it.
6667 Make sure that any inline virtuals are declared inline in the class
6668 body, even if they are not defined there.
6670 @item type_info objects
6673 C++ requires information about types to be written out in order to
6674 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
6675 For polymorphic classes (classes with virtual functions), the type_info
6676 object is written out along with the vtable so that @samp{dynamic_cast}
6677 can determine the dynamic type of a class object at runtime. For all
6678 other types, we write out the type_info object when it is used: when
6679 applying @samp{typeid} to an expression, throwing an object, or
6680 referring to a type in a catch clause or exception specification.
6682 @item Template Instantiations
6683 Most everything in this section also applies to template instantiations,
6684 but there are other options as well.
6685 @xref{Template Instantiation,,Where's the Template?}.
6689 When used with GNU ld version 2.8 or later on an ELF system such as
6690 Linux/GNU or Solaris 2, or on Microsoft Windows, duplicate copies of
6691 these constructs will be discarded at link time. This is known as
6694 On targets that don't support COMDAT, but do support weak symbols, GCC
6695 will use them. This way one copy will override all the others, but
6696 the unused copies will still take up space in the executable.
6698 For targets which do not support either COMDAT or weak symbols,
6699 most entities with vague linkage will be emitted as local symbols to
6700 avoid duplicate definition errors from the linker. This will not happen
6701 for local statics in inlines, however, as having multiple copies will
6702 almost certainly break things.
6704 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
6705 another way to control placement of these constructs.
6708 @section Declarations and Definitions in One Header
6710 @cindex interface and implementation headers, C++
6711 @cindex C++ interface and implementation headers
6712 C++ object definitions can be quite complex. In principle, your source
6713 code will need two kinds of things for each object that you use across
6714 more than one source file. First, you need an @dfn{interface}
6715 specification, describing its structure with type declarations and
6716 function prototypes. Second, you need the @dfn{implementation} itself.
6717 It can be tedious to maintain a separate interface description in a
6718 header file, in parallel to the actual implementation. It is also
6719 dangerous, since separate interface and implementation definitions may
6720 not remain parallel.
6722 @cindex pragmas, interface and implementation
6723 With GNU C++, you can use a single header file for both purposes.
6726 @emph{Warning:} The mechanism to specify this is in transition. For the
6727 nonce, you must use one of two @code{#pragma} commands; in a future
6728 release of GNU C++, an alternative mechanism will make these
6729 @code{#pragma} commands unnecessary.
6732 The header file contains the full definitions, but is marked with
6733 @samp{#pragma interface} in the source code. This allows the compiler
6734 to use the header file only as an interface specification when ordinary
6735 source files incorporate it with @code{#include}. In the single source
6736 file where the full implementation belongs, you can use either a naming
6737 convention or @samp{#pragma implementation} to indicate this alternate
6738 use of the header file.
6741 @item #pragma interface
6742 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
6743 @kindex #pragma interface
6744 Use this directive in @emph{header files} that define object classes, to save
6745 space in most of the object files that use those classes. Normally,
6746 local copies of certain information (backup copies of inline member
6747 functions, debugging information, and the internal tables that implement
6748 virtual functions) must be kept in each object file that includes class
6749 definitions. You can use this pragma to avoid such duplication. When a
6750 header file containing @samp{#pragma interface} is included in a
6751 compilation, this auxiliary information will not be generated (unless
6752 the main input source file itself uses @samp{#pragma implementation}).
6753 Instead, the object files will contain references to be resolved at link
6756 The second form of this directive is useful for the case where you have
6757 multiple headers with the same name in different directories. If you
6758 use this form, you must specify the same string to @samp{#pragma
6761 @item #pragma implementation
6762 @itemx #pragma implementation "@var{objects}.h"
6763 @kindex #pragma implementation
6764 Use this pragma in a @emph{main input file}, when you want full output from
6765 included header files to be generated (and made globally visible). The
6766 included header file, in turn, should use @samp{#pragma interface}.
6767 Backup copies of inline member functions, debugging information, and the
6768 internal tables used to implement virtual functions are all generated in
6769 implementation files.
6771 @cindex implied @code{#pragma implementation}
6772 @cindex @code{#pragma implementation}, implied
6773 @cindex naming convention, implementation headers
6774 If you use @samp{#pragma implementation} with no argument, it applies to
6775 an include file with the same basename@footnote{A file's @dfn{basename}
6776 was the name stripped of all leading path information and of trailing
6777 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
6778 file. For example, in @file{allclass.cc}, giving just
6779 @samp{#pragma implementation}
6780 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
6782 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
6783 an implementation file whenever you would include it from
6784 @file{allclass.cc} even if you never specified @samp{#pragma
6785 implementation}. This was deemed to be more trouble than it was worth,
6786 however, and disabled.
6788 If you use an explicit @samp{#pragma implementation}, it must appear in
6789 your source file @emph{before} you include the affected header files.
6791 Use the string argument if you want a single implementation file to
6792 include code from multiple header files. (You must also use
6793 @samp{#include} to include the header file; @samp{#pragma
6794 implementation} only specifies how to use the file---it doesn't actually
6797 There is no way to split up the contents of a single header file into
6798 multiple implementation files.
6801 @cindex inlining and C++ pragmas
6802 @cindex C++ pragmas, effect on inlining
6803 @cindex pragmas in C++, effect on inlining
6804 @samp{#pragma implementation} and @samp{#pragma interface} also have an
6805 effect on function inlining.
6807 If you define a class in a header file marked with @samp{#pragma
6808 interface}, the effect on a function defined in that class is similar to
6809 an explicit @code{extern} declaration---the compiler emits no code at
6810 all to define an independent version of the function. Its definition
6811 is used only for inlining with its callers.
6813 @opindex fno-implement-inlines
6814 Conversely, when you include the same header file in a main source file
6815 that declares it as @samp{#pragma implementation}, the compiler emits
6816 code for the function itself; this defines a version of the function
6817 that can be found via pointers (or by callers compiled without
6818 inlining). If all calls to the function can be inlined, you can avoid
6819 emitting the function by compiling with @option{-fno-implement-inlines}.
6820 If any calls were not inlined, you will get linker errors.
6822 @node Template Instantiation
6823 @section Where's the Template?
6825 @cindex template instantiation
6827 C++ templates are the first language feature to require more
6828 intelligence from the environment than one usually finds on a UNIX
6829 system. Somehow the compiler and linker have to make sure that each
6830 template instance occurs exactly once in the executable if it is needed,
6831 and not at all otherwise. There are two basic approaches to this
6832 problem, which I will refer to as the Borland model and the Cfront model.
6836 Borland C++ solved the template instantiation problem by adding the code
6837 equivalent of common blocks to their linker; the compiler emits template
6838 instances in each translation unit that uses them, and the linker
6839 collapses them together. The advantage of this model is that the linker
6840 only has to consider the object files themselves; there is no external
6841 complexity to worry about. This disadvantage is that compilation time
6842 is increased because the template code is being compiled repeatedly.
6843 Code written for this model tends to include definitions of all
6844 templates in the header file, since they must be seen to be
6848 The AT&T C++ translator, Cfront, solved the template instantiation
6849 problem by creating the notion of a template repository, an
6850 automatically maintained place where template instances are stored. A
6851 more modern version of the repository works as follows: As individual
6852 object files are built, the compiler places any template definitions and
6853 instantiations encountered in the repository. At link time, the link
6854 wrapper adds in the objects in the repository and compiles any needed
6855 instances that were not previously emitted. The advantages of this
6856 model are more optimal compilation speed and the ability to use the
6857 system linker; to implement the Borland model a compiler vendor also
6858 needs to replace the linker. The disadvantages are vastly increased
6859 complexity, and thus potential for error; for some code this can be
6860 just as transparent, but in practice it can been very difficult to build
6861 multiple programs in one directory and one program in multiple
6862 directories. Code written for this model tends to separate definitions
6863 of non-inline member templates into a separate file, which should be
6864 compiled separately.
6867 When used with GNU ld version 2.8 or later on an ELF system such as
6868 Linux/GNU or Solaris 2, or on Microsoft Windows, g++ supports the
6869 Borland model. On other systems, g++ implements neither automatic
6872 A future version of g++ will support a hybrid model whereby the compiler
6873 will emit any instantiations for which the template definition is
6874 included in the compile, and store template definitions and
6875 instantiation context information into the object file for the rest.
6876 The link wrapper will extract that information as necessary and invoke
6877 the compiler to produce the remaining instantiations. The linker will
6878 then combine duplicate instantiations.
6880 In the mean time, you have the following options for dealing with
6881 template instantiations:
6886 Compile your template-using code with @option{-frepo}. The compiler will
6887 generate files with the extension @samp{.rpo} listing all of the
6888 template instantiations used in the corresponding object files which
6889 could be instantiated there; the link wrapper, @samp{collect2}, will
6890 then update the @samp{.rpo} files to tell the compiler where to place
6891 those instantiations and rebuild any affected object files. The
6892 link-time overhead is negligible after the first pass, as the compiler
6893 will continue to place the instantiations in the same files.
6895 This is your best option for application code written for the Borland
6896 model, as it will just work. Code written for the Cfront model will
6897 need to be modified so that the template definitions are available at
6898 one or more points of instantiation; usually this is as simple as adding
6899 @code{#include <tmethods.cc>} to the end of each template header.
6901 For library code, if you want the library to provide all of the template
6902 instantiations it needs, just try to link all of its object files
6903 together; the link will fail, but cause the instantiations to be
6904 generated as a side effect. Be warned, however, that this may cause
6905 conflicts if multiple libraries try to provide the same instantiations.
6906 For greater control, use explicit instantiation as described in the next
6910 @opindex fno-implicit-templates
6911 Compile your code with @option{-fno-implicit-templates} to disable the
6912 implicit generation of template instances, and explicitly instantiate
6913 all the ones you use. This approach requires more knowledge of exactly
6914 which instances you need than do the others, but it's less
6915 mysterious and allows greater control. You can scatter the explicit
6916 instantiations throughout your program, perhaps putting them in the
6917 translation units where the instances are used or the translation units
6918 that define the templates themselves; you can put all of the explicit
6919 instantiations you need into one big file; or you can create small files
6926 template class Foo<int>;
6927 template ostream& operator <<
6928 (ostream&, const Foo<int>&);
6931 for each of the instances you need, and create a template instantiation
6934 If you are using Cfront-model code, you can probably get away with not
6935 using @option{-fno-implicit-templates} when compiling files that don't
6936 @samp{#include} the member template definitions.
6938 If you use one big file to do the instantiations, you may want to
6939 compile it without @option{-fno-implicit-templates} so you get all of the
6940 instances required by your explicit instantiations (but not by any
6941 other files) without having to specify them as well.
6943 g++ has extended the template instantiation syntax outlined in the
6944 Working Paper to allow forward declaration of explicit instantiations
6945 (with @code{extern}), instantiation of the compiler support data for a
6946 template class (i.e.@: the vtable) without instantiating any of its
6947 members (with @code{inline}), and instantiation of only the static data
6948 members of a template class, without the support data or member
6949 functions (with (@code{static}):
6952 extern template int max (int, int);
6953 inline template class Foo<int>;
6954 static template class Foo<int>;
6958 Do nothing. Pretend g++ does implement automatic instantiation
6959 management. Code written for the Borland model will work fine, but
6960 each translation unit will contain instances of each of the templates it
6961 uses. In a large program, this can lead to an unacceptable amount of code
6965 @opindex fexternal-templates
6966 Add @samp{#pragma interface} to all files containing template
6967 definitions. For each of these files, add @samp{#pragma implementation
6968 "@var{filename}"} to the top of some @samp{.C} file which
6969 @samp{#include}s it. Then compile everything with
6970 @option{-fexternal-templates}. The templates will then only be expanded
6971 in the translation unit which implements them (i.e.@: has a @samp{#pragma
6972 implementation} line for the file where they live); all other files will
6973 use external references. If you're lucky, everything should work
6974 properly. If you get undefined symbol errors, you need to make sure
6975 that each template instance which is used in the program is used in the
6976 file which implements that template. If you don't have any use for a
6977 particular instance in that file, you can just instantiate it
6978 explicitly, using the syntax from the latest C++ working paper:
6981 template class A<int>;
6982 template ostream& operator << (ostream&, const A<int>&);
6985 This strategy will work with code written for either model. If you are
6986 using code written for the Cfront model, the file containing a class
6987 template and the file containing its member templates should be
6988 implemented in the same translation unit.
6991 @opindex falt-external-templates
6992 A slight variation on this approach is to use the flag
6993 @option{-falt-external-templates} instead. This flag causes template
6994 instances to be emitted in the translation unit that implements the
6995 header where they are first instantiated, rather than the one which
6996 implements the file where the templates are defined. This header must
6997 be the same in all translation units, or things are likely to break.
6999 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
7000 more discussion of these pragmas.
7003 @node Bound member functions
7004 @section Extracting the function pointer from a bound pointer to member function
7007 @cindex pointer to member function
7008 @cindex bound pointer to member function
7010 In C++, pointer to member functions (PMFs) are implemented using a wide
7011 pointer of sorts to handle all the possible call mechanisms; the PMF
7012 needs to store information about how to adjust the @samp{this} pointer,
7013 and if the function pointed to is virtual, where to find the vtable, and
7014 where in the vtable to look for the member function. If you are using
7015 PMFs in an inner loop, you should really reconsider that decision. If
7016 that is not an option, you can extract the pointer to the function that
7017 would be called for a given object/PMF pair and call it directly inside
7018 the inner loop, to save a bit of time.
7020 Note that you will still be paying the penalty for the call through a
7021 function pointer; on most modern architectures, such a call defeats the
7022 branch prediction features of the CPU@. This is also true of normal
7023 virtual function calls.
7025 The syntax for this extension is
7029 extern int (A::*fp)();
7030 typedef int (*fptr)(A *);
7032 fptr p = (fptr)(a.*fp);
7035 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
7036 no object is needed to obtain the address of the function. They can be
7037 converted to function pointers directly:
7040 fptr p1 = (fptr)(&A::foo);
7043 @opindex Wno-pmf-conversions
7044 You must specify @option{-Wno-pmf-conversions} to use this extension.
7046 @node C++ Attributes
7047 @section C++-Specific Variable, Function, and Type Attributes
7049 Some attributes only make sense for C++ programs.
7052 @item init_priority (@var{priority})
7053 @cindex init_priority attribute
7056 In Standard C++, objects defined at namespace scope are guaranteed to be
7057 initialized in an order in strict accordance with that of their definitions
7058 @emph{in a given translation unit}. No guarantee is made for initializations
7059 across translation units. However, GNU C++ allows users to control the
7060 order of initialization of objects defined at namespace scope with the
7061 @code{init_priority} attribute by specifying a relative @var{priority},
7062 a constant integral expression currently bounded between 101 and 65535
7063 inclusive. Lower numbers indicate a higher priority.
7065 In the following example, @code{A} would normally be created before
7066 @code{B}, but the @code{init_priority} attribute has reversed that order:
7069 Some_Class A __attribute__ ((init_priority (2000)));
7070 Some_Class B __attribute__ ((init_priority (543)));
7074 Note that the particular values of @var{priority} do not matter; only their
7077 @item java_interface
7078 @cindex java_interface attribute
7080 This type attribute informs C++ that the class is a Java interface. It may
7081 only be applied to classes declared within an @code{extern "Java"} block.
7082 Calls to methods declared in this interface will be dispatched using GCJ's
7083 interface table mechanism, instead of regular virtual table dispatch.
7087 @node Java Exceptions
7088 @section Java Exceptions
7090 The Java language uses a slightly different exception handling model
7091 from C++. Normally, GNU C++ will automatically detect when you are
7092 writing C++ code that uses Java exceptions, and handle them
7093 appropriately. However, if C++ code only needs to execute destructors
7094 when Java exceptions are thrown through it, GCC will guess incorrectly.
7095 Sample problematic code is:
7098 struct S @{ ~S(); @};
7099 extern void bar(); // is written in Java, and may throw exceptions
7108 The usual effect of an incorrect guess is a link failure, complaining of
7109 a missing routine called @samp{__gxx_personality_v0}.
7111 You can inform the compiler that Java exceptions are to be used in a
7112 translation unit, irrespective of what it might think, by writing
7113 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
7114 @samp{#pragma} must appear before any functions that throw or catch
7115 exceptions, or run destructors when exceptions are thrown through them.
7117 You cannot mix Java and C++ exceptions in the same translation unit. It
7118 is believed to be safe to throw a C++ exception from one file through
7119 another file compiled for the Java exception model, or vice versa, but
7120 there may be bugs in this area.
7122 @node Deprecated Features
7123 @section Deprecated Features
7125 In the past, the GNU C++ compiler was extended to experiment with new
7126 features, at a time when the C++ language was still evolving. Now that
7127 the C++ standard is complete, some of those features are superseded by
7128 superior alternatives. Using the old features might cause a warning in
7129 some cases that the feature will be dropped in the future. In other
7130 cases, the feature might be gone already.
7132 While the list below is not exhaustive, it documents some of the options
7133 that are now deprecated:
7136 @item -fexternal-templates
7137 @itemx -falt-external-templates
7138 These are two of the many ways for g++ to implement template
7139 instantiation. @xref{Template Instantiation}. The C++ standard clearly
7140 defines how template definitions have to be organized across
7141 implementation units. g++ has an implicit instantiation mechanism that
7142 should work just fine for standard-conforming code.
7144 @item -fstrict-prototype
7145 @itemx -fno-strict-prototype
7146 Previously it was possible to use an empty prototype parameter list to
7147 indicate an unspecified number of parameters (like C), rather than no
7148 parameters, as C++ demands. This feature has been removed, except where
7149 it is required for backwards compatibility @xref{Backwards Compatibility}.
7152 The named return value extension has been deprecated, and is now
7155 The use of initializer lists with new expressions has been deprecated,
7156 and is now removed from g++.
7158 Floating and complex non-type template parameters have been deprecated,
7159 and are now removed from g++.
7161 The implicit typename extension has been deprecated and will be removed
7162 from g++ at some point. In some cases g++ determines that a dependant
7163 type such as @code{TPL<T>::X} is a type without needing a
7164 @code{typename} keyword, contrary to the standard.
7166 @node Backwards Compatibility
7167 @section Backwards Compatibility
7168 @cindex Backwards Compatibility
7169 @cindex ARM [Annotated C++ Reference Manual]
7171 Now that there is a definitive ISO standard C++, G++ has a specification
7172 to adhere to. The C++ language evolved over time, and features that
7173 used to be acceptable in previous drafts of the standard, such as the ARM
7174 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
7175 compilation of C++ written to such drafts, G++ contains some backwards
7176 compatibilities. @emph{All such backwards compatibility features are
7177 liable to disappear in future versions of G++.} They should be considered
7178 deprecated @xref{Deprecated Features}.
7182 If a variable is declared at for scope, it used to remain in scope until
7183 the end of the scope which contained the for statement (rather than just
7184 within the for scope). G++ retains this, but issues a warning, if such a
7185 variable is accessed outside the for scope.
7187 @item Implicit C language
7188 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
7189 scope to set the language. On such systems, all header files are
7190 implicitly scoped inside a C language scope. Also, an empty prototype
7191 @code{()} will be treated as an unspecified number of arguments, rather
7192 than no arguments, as C++ demands.