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.
437 @node Statement Exprs
438 @section Statements and Declarations in Expressions
439 @cindex statements inside expressions
440 @cindex declarations inside expressions
441 @cindex expressions containing statements
442 @cindex macros, statements in expressions
444 @c the above section title wrapped and causes an underfull hbox.. i
445 @c changed it from "within" to "in". --mew 4feb93
447 A compound statement enclosed in parentheses may appear as an expression
448 in GNU C@. This allows you to use loops, switches, and local variables
449 within an expression.
451 Recall that a compound statement is a sequence of statements surrounded
452 by braces; in this construct, parentheses go around the braces. For
456 (@{ int y = foo (); int z;
463 is a valid (though slightly more complex than necessary) expression
464 for the absolute value of @code{foo ()}.
466 The last thing in the compound statement should be an expression
467 followed by a semicolon; the value of this subexpression serves as the
468 value of the entire construct. (If you use some other kind of statement
469 last within the braces, the construct has type @code{void}, and thus
470 effectively no value.)
472 This feature is especially useful in making macro definitions ``safe'' (so
473 that they evaluate each operand exactly once). For example, the
474 ``maximum'' function is commonly defined as a macro in standard C as
478 #define max(a,b) ((a) > (b) ? (a) : (b))
482 @cindex side effects, macro argument
483 But this definition computes either @var{a} or @var{b} twice, with bad
484 results if the operand has side effects. In GNU C, if you know the
485 type of the operands (here let's assume @code{int}), you can define
486 the macro safely as follows:
489 #define maxint(a,b) \
490 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
493 Embedded statements are not allowed in constant expressions, such as
494 the value of an enumeration constant, the width of a bit-field, or
495 the initial value of a static variable.
497 If you don't know the type of the operand, you can still do this, but you
498 must use @code{typeof} (@pxref{Typeof}) or type naming (@pxref{Naming
501 Statement expressions are not supported fully in G++, and their fate
502 there is unclear. (It is possible that they will become fully supported
503 at some point, or that they will be deprecated, or that the bugs that
504 are present will continue to exist indefinitely.) Presently, statement
505 expressions do not work well as default arguments.
507 In addition, there are semantic issues with statement-expressions in
508 C++. If you try to use statement-expressions instead of inline
509 functions in C++, you may be surprised at the way object destruction is
510 handled. For example:
513 #define foo(a) (@{int b = (a); b + 3; @})
517 does not work the same way as:
520 inline int foo(int a) @{ int b = a; return b + 3; @}
524 In particular, if the expression passed into @code{foo} involves the
525 creation of temporaries, the destructors for those temporaries will be
526 run earlier in the case of the macro than in the case of the function.
528 These considerations mean that it is probably a bad idea to use
529 statement-expressions of this form in header files that are designed to
530 work with C++. (Note that some versions of the GNU C Library contained
531 header files using statement-expression that lead to precisely this
535 @section Locally Declared Labels
537 @cindex macros, local labels
539 Each statement expression is a scope in which @dfn{local labels} can be
540 declared. A local label is simply an identifier; you can jump to it
541 with an ordinary @code{goto} statement, but only from within the
542 statement expression it belongs to.
544 A local label declaration looks like this:
547 __label__ @var{label};
554 __label__ @var{label1}, @var{label2}, @dots{};
557 Local label declarations must come at the beginning of the statement
558 expression, right after the @samp{(@{}, before any ordinary
561 The label declaration defines the label @emph{name}, but does not define
562 the label itself. You must do this in the usual way, with
563 @code{@var{label}:}, within the statements of the statement expression.
565 The local label feature is useful because statement expressions are
566 often used in macros. If the macro contains nested loops, a @code{goto}
567 can be useful for breaking out of them. However, an ordinary label
568 whose scope is the whole function cannot be used: if the macro can be
569 expanded several times in one function, the label will be multiply
570 defined in that function. A local label avoids this problem. For
574 #define SEARCH(array, target) \
577 typeof (target) _SEARCH_target = (target); \
578 typeof (*(array)) *_SEARCH_array = (array); \
581 for (i = 0; i < max; i++) \
582 for (j = 0; j < max; j++) \
583 if (_SEARCH_array[i][j] == _SEARCH_target) \
584 @{ value = i; goto found; @} \
591 @node Labels as Values
592 @section Labels as Values
593 @cindex labels as values
594 @cindex computed gotos
595 @cindex goto with computed label
596 @cindex address of a label
598 You can get the address of a label defined in the current function
599 (or a containing function) with the unary operator @samp{&&}. The
600 value has type @code{void *}. This value is a constant and can be used
601 wherever a constant of that type is valid. For example:
609 To use these values, you need to be able to jump to one. This is done
610 with the computed goto statement@footnote{The analogous feature in
611 Fortran is called an assigned goto, but that name seems inappropriate in
612 C, where one can do more than simply store label addresses in label
613 variables.}, @code{goto *@var{exp};}. For example,
620 Any expression of type @code{void *} is allowed.
622 One way of using these constants is in initializing a static array that
623 will serve as a jump table:
626 static void *array[] = @{ &&foo, &&bar, &&hack @};
629 Then you can select a label with indexing, like this:
636 Note that this does not check whether the subscript is in bounds---array
637 indexing in C never does that.
639 Such an array of label values serves a purpose much like that of the
640 @code{switch} statement. The @code{switch} statement is cleaner, so
641 use that rather than an array unless the problem does not fit a
642 @code{switch} statement very well.
644 Another use of label values is in an interpreter for threaded code.
645 The labels within the interpreter function can be stored in the
646 threaded code for super-fast dispatching.
648 You may not use this mechanism to jump to code in a different function.
649 If you do that, totally unpredictable things will happen. The best way to
650 avoid this is to store the label address only in automatic variables and
651 never pass it as an argument.
653 An alternate way to write the above example is
656 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
658 goto *(&&foo + array[i]);
662 This is more friendly to code living in shared libraries, as it reduces
663 the number of dynamic relocations that are needed, and by consequence,
664 allows the data to be read-only.
666 @node Nested Functions
667 @section Nested Functions
668 @cindex nested functions
669 @cindex downward funargs
672 A @dfn{nested function} is a function defined inside another function.
673 (Nested functions are not supported for GNU C++.) The nested function's
674 name is local to the block where it is defined. For example, here we
675 define a nested function named @code{square}, and call it twice:
679 foo (double a, double b)
681 double square (double z) @{ return z * z; @}
683 return square (a) + square (b);
688 The nested function can access all the variables of the containing
689 function that are visible at the point of its definition. This is
690 called @dfn{lexical scoping}. For example, here we show a nested
691 function which uses an inherited variable named @code{offset}:
695 bar (int *array, int offset, int size)
697 int access (int *array, int index)
698 @{ return array[index + offset]; @}
701 for (i = 0; i < size; i++)
702 @dots{} access (array, i) @dots{}
707 Nested function definitions are permitted within functions in the places
708 where variable definitions are allowed; that is, in any block, before
709 the first statement in the block.
711 It is possible to call the nested function from outside the scope of its
712 name by storing its address or passing the address to another function:
715 hack (int *array, int size)
717 void store (int index, int value)
718 @{ array[index] = value; @}
720 intermediate (store, size);
724 Here, the function @code{intermediate} receives the address of
725 @code{store} as an argument. If @code{intermediate} calls @code{store},
726 the arguments given to @code{store} are used to store into @code{array}.
727 But this technique works only so long as the containing function
728 (@code{hack}, in this example) does not exit.
730 If you try to call the nested function through its address after the
731 containing function has exited, all hell will break loose. If you try
732 to call it after a containing scope level has exited, and if it refers
733 to some of the variables that are no longer in scope, you may be lucky,
734 but it's not wise to take the risk. If, however, the nested function
735 does not refer to anything that has gone out of scope, you should be
738 GCC implements taking the address of a nested function using a technique
739 called @dfn{trampolines}. A paper describing them is available as
742 @uref{http://people.debian.org/~karlheg/Usenix88-lexic.pdf}.
744 A nested function can jump to a label inherited from a containing
745 function, provided the label was explicitly declared in the containing
746 function (@pxref{Local Labels}). Such a jump returns instantly to the
747 containing function, exiting the nested function which did the
748 @code{goto} and any intermediate functions as well. Here is an example:
752 bar (int *array, int offset, int size)
755 int access (int *array, int index)
759 return array[index + offset];
763 for (i = 0; i < size; i++)
764 @dots{} access (array, i) @dots{}
768 /* @r{Control comes here from @code{access}
769 if it detects an error.} */
776 A nested function always has internal linkage. Declaring one with
777 @code{extern} is erroneous. If you need to declare the nested function
778 before its definition, use @code{auto} (which is otherwise meaningless
779 for function declarations).
782 bar (int *array, int offset, int size)
785 auto int access (int *, int);
787 int access (int *array, int index)
791 return array[index + offset];
797 @node Constructing Calls
798 @section Constructing Function Calls
799 @cindex constructing calls
800 @cindex forwarding calls
802 Using the built-in functions described below, you can record
803 the arguments a function received, and call another function
804 with the same arguments, without knowing the number or types
807 You can also record the return value of that function call,
808 and later return that value, without knowing what data type
809 the function tried to return (as long as your caller expects
812 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
813 This built-in function returns a pointer to data
814 describing how to perform a call with the same arguments as were passed
815 to the current function.
817 The function saves the arg pointer register, structure value address,
818 and all registers that might be used to pass arguments to a function
819 into a block of memory allocated on the stack. Then it returns the
820 address of that block.
823 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
824 This built-in function invokes @var{function}
825 with a copy of the parameters described by @var{arguments}
828 The value of @var{arguments} should be the value returned by
829 @code{__builtin_apply_args}. The argument @var{size} specifies the size
830 of the stack argument data, in bytes.
832 This function returns a pointer to data describing
833 how to return whatever value was returned by @var{function}. The data
834 is saved in a block of memory allocated on the stack.
836 It is not always simple to compute the proper value for @var{size}. The
837 value is used by @code{__builtin_apply} to compute the amount of data
838 that should be pushed on the stack and copied from the incoming argument
842 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
843 This built-in function returns the value described by @var{result} from
844 the containing function. You should specify, for @var{result}, a value
845 returned by @code{__builtin_apply}.
849 @section Naming an Expression's Type
852 You can give a name to the type of an expression using a @code{typedef}
853 declaration with an initializer. Here is how to define @var{name} as a
854 type name for the type of @var{exp}:
857 typedef @var{name} = @var{exp};
860 This is useful in conjunction with the statements-within-expressions
861 feature. Here is how the two together can be used to define a safe
862 ``maximum'' macro that operates on any arithmetic type:
866 (@{typedef _ta = (a), _tb = (b); \
867 _ta _a = (a); _tb _b = (b); \
868 _a > _b ? _a : _b; @})
871 @cindex underscores in variables in macros
872 @cindex @samp{_} in variables in macros
873 @cindex local variables in macros
874 @cindex variables, local, in macros
875 @cindex macros, local variables in
877 The reason for using names that start with underscores for the local
878 variables is to avoid conflicts with variable names that occur within the
879 expressions that are substituted for @code{a} and @code{b}. Eventually we
880 hope to design a new form of declaration syntax that allows you to declare
881 variables whose scopes start only after their initializers; this will be a
882 more reliable way to prevent such conflicts.
885 @section Referring to a Type with @code{typeof}
888 @cindex macros, types of arguments
890 Another way to refer to the type of an expression is with @code{typeof}.
891 The syntax of using of this keyword looks like @code{sizeof}, but the
892 construct acts semantically like a type name defined with @code{typedef}.
894 There are two ways of writing the argument to @code{typeof}: with an
895 expression or with a type. Here is an example with an expression:
902 This assumes that @code{x} is an array of pointers to functions;
903 the type described is that of the values of the functions.
905 Here is an example with a typename as the argument:
912 Here the type described is that of pointers to @code{int}.
914 If you are writing a header file that must work when included in ISO C
915 programs, write @code{__typeof__} instead of @code{typeof}.
916 @xref{Alternate Keywords}.
918 A @code{typeof}-construct can be used anywhere a typedef name could be
919 used. For example, you can use it in a declaration, in a cast, or inside
920 of @code{sizeof} or @code{typeof}.
924 This declares @code{y} with the type of what @code{x} points to.
931 This declares @code{y} as an array of such values.
938 This declares @code{y} as an array of pointers to characters:
941 typeof (typeof (char *)[4]) y;
945 It is equivalent to the following traditional C declaration:
951 To see the meaning of the declaration using @code{typeof}, and why it
952 might be a useful way to write, let's rewrite it with these macros:
955 #define pointer(T) typeof(T *)
956 #define array(T, N) typeof(T [N])
960 Now the declaration can be rewritten this way:
963 array (pointer (char), 4) y;
967 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
968 pointers to @code{char}.
972 @section Generalized Lvalues
973 @cindex compound expressions as lvalues
974 @cindex expressions, compound, as lvalues
975 @cindex conditional expressions as lvalues
976 @cindex expressions, conditional, as lvalues
977 @cindex casts as lvalues
978 @cindex generalized lvalues
979 @cindex lvalues, generalized
980 @cindex extensions, @code{?:}
981 @cindex @code{?:} extensions
982 Compound expressions, conditional expressions and casts are allowed as
983 lvalues provided their operands are lvalues. This means that you can take
984 their addresses or store values into them.
986 Standard C++ allows compound expressions and conditional expressions as
987 lvalues, and permits casts to reference type, so use of this extension
988 is deprecated for C++ code.
990 For example, a compound expression can be assigned, provided the last
991 expression in the sequence is an lvalue. These two expressions are
999 Similarly, the address of the compound expression can be taken. These two
1000 expressions are equivalent:
1007 A conditional expression is a valid lvalue if its type is not void and the
1008 true and false branches are both valid lvalues. For example, these two
1009 expressions are equivalent:
1013 (a ? b = 5 : (c = 5))
1016 A cast is a valid lvalue if its operand is an lvalue. A simple
1017 assignment whose left-hand side is a cast works by converting the
1018 right-hand side first to the specified type, then to the type of the
1019 inner left-hand side expression. After this is stored, the value is
1020 converted back to the specified type to become the value of the
1021 assignment. Thus, if @code{a} has type @code{char *}, the following two
1022 expressions are equivalent:
1026 (int)(a = (char *)(int)5)
1029 An assignment-with-arithmetic operation such as @samp{+=} applied to a cast
1030 performs the arithmetic using the type resulting from the cast, and then
1031 continues as in the previous case. Therefore, these two expressions are
1036 (int)(a = (char *)(int) ((int)a + 5))
1039 You cannot take the address of an lvalue cast, because the use of its
1040 address would not work out coherently. Suppose that @code{&(int)f} were
1041 permitted, where @code{f} has type @code{float}. Then the following
1042 statement would try to store an integer bit-pattern where a floating
1043 point number belongs:
1049 This is quite different from what @code{(int)f = 1} would do---that
1050 would convert 1 to floating point and store it. Rather than cause this
1051 inconsistency, we think it is better to prohibit use of @samp{&} on a cast.
1053 If you really do want an @code{int *} pointer with the address of
1054 @code{f}, you can simply write @code{(int *)&f}.
1057 @section Conditionals with Omitted Operands
1058 @cindex conditional expressions, extensions
1059 @cindex omitted middle-operands
1060 @cindex middle-operands, omitted
1061 @cindex extensions, @code{?:}
1062 @cindex @code{?:} extensions
1064 The middle operand in a conditional expression may be omitted. Then
1065 if the first operand is nonzero, its value is the value of the conditional
1068 Therefore, the expression
1075 has the value of @code{x} if that is nonzero; otherwise, the value of
1078 This example is perfectly equivalent to
1084 @cindex side effect in ?:
1085 @cindex ?: side effect
1087 In this simple case, the ability to omit the middle operand is not
1088 especially useful. When it becomes useful is when the first operand does,
1089 or may (if it is a macro argument), contain a side effect. Then repeating
1090 the operand in the middle would perform the side effect twice. Omitting
1091 the middle operand uses the value already computed without the undesirable
1092 effects of recomputing it.
1095 @section Double-Word Integers
1096 @cindex @code{long long} data types
1097 @cindex double-word arithmetic
1098 @cindex multiprecision arithmetic
1099 @cindex @code{LL} integer suffix
1100 @cindex @code{ULL} integer suffix
1102 ISO C99 supports data types for integers that are at least 64 bits wide,
1103 and as an extension GCC supports them in C89 mode and in C++.
1104 Simply write @code{long long int} for a signed integer, or
1105 @code{unsigned long long int} for an unsigned integer. To make an
1106 integer constant of type @code{long long int}, add the suffix @samp{LL}
1107 to the integer. To make an integer constant of type @code{unsigned long
1108 long int}, add the suffix @samp{ULL} to the integer.
1110 You can use these types in arithmetic like any other integer types.
1111 Addition, subtraction, and bitwise boolean operations on these types
1112 are open-coded on all types of machines. Multiplication is open-coded
1113 if the machine supports fullword-to-doubleword a widening multiply
1114 instruction. Division and shifts are open-coded only on machines that
1115 provide special support. The operations that are not open-coded use
1116 special library routines that come with GCC@.
1118 There may be pitfalls when you use @code{long long} types for function
1119 arguments, unless you declare function prototypes. If a function
1120 expects type @code{int} for its argument, and you pass a value of type
1121 @code{long long int}, confusion will result because the caller and the
1122 subroutine will disagree about the number of bytes for the argument.
1123 Likewise, if the function expects @code{long long int} and you pass
1124 @code{int}. The best way to avoid such problems is to use prototypes.
1127 @section Complex Numbers
1128 @cindex complex numbers
1129 @cindex @code{_Complex} keyword
1130 @cindex @code{__complex__} keyword
1132 ISO C99 supports complex floating data types, and as an extension GCC
1133 supports them in C89 mode and in C++, and supports complex integer data
1134 types which are not part of ISO C99. You can declare complex types
1135 using the keyword @code{_Complex}. As an extension, the older GNU
1136 keyword @code{__complex__} is also supported.
1138 For example, @samp{_Complex double x;} declares @code{x} as a
1139 variable whose real part and imaginary part are both of type
1140 @code{double}. @samp{_Complex short int y;} declares @code{y} to
1141 have real and imaginary parts of type @code{short int}; this is not
1142 likely to be useful, but it shows that the set of complex types is
1145 To write a constant with a complex data type, use the suffix @samp{i} or
1146 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
1147 has type @code{_Complex float} and @code{3i} has type
1148 @code{_Complex int}. Such a constant always has a pure imaginary
1149 value, but you can form any complex value you like by adding one to a
1150 real constant. This is a GNU extension; if you have an ISO C99
1151 conforming C library (such as GNU libc), and want to construct complex
1152 constants of floating type, you should include @code{<complex.h>} and
1153 use the macros @code{I} or @code{_Complex_I} instead.
1155 @cindex @code{__real__} keyword
1156 @cindex @code{__imag__} keyword
1157 To extract the real part of a complex-valued expression @var{exp}, write
1158 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
1159 extract the imaginary part. This is a GNU extension; for values of
1160 floating type, you should use the ISO C99 functions @code{crealf},
1161 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
1162 @code{cimagl}, declared in @code{<complex.h>} and also provided as
1163 built-in functions by GCC@.
1165 @cindex complex conjugation
1166 The operator @samp{~} performs complex conjugation when used on a value
1167 with a complex type. This is a GNU extension; for values of
1168 floating type, you should use the ISO C99 functions @code{conjf},
1169 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
1170 provided as built-in functions by GCC@.
1172 GCC can allocate complex automatic variables in a noncontiguous
1173 fashion; it's even possible for the real part to be in a register while
1174 the imaginary part is on the stack (or vice-versa). None of the
1175 supported debugging info formats has a way to represent noncontiguous
1176 allocation like this, so GCC describes a noncontiguous complex
1177 variable as if it were two separate variables of noncomplex type.
1178 If the variable's actual name is @code{foo}, the two fictitious
1179 variables are named @code{foo$real} and @code{foo$imag}. You can
1180 examine and set these two fictitious variables with your debugger.
1182 A future version of GDB will know how to recognize such pairs and treat
1183 them as a single variable with a complex type.
1189 ISO C99 supports floating-point numbers written not only in the usual
1190 decimal notation, such as @code{1.55e1}, but also numbers such as
1191 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
1192 supports this in C89 mode (except in some cases when strictly
1193 conforming) and in C++. In that format the
1194 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
1195 mandatory. The exponent is a decimal number that indicates the power of
1196 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
1203 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
1204 is the same as @code{1.55e1}.
1206 Unlike for floating-point numbers in the decimal notation the exponent
1207 is always required in the hexadecimal notation. Otherwise the compiler
1208 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
1209 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
1210 extension for floating-point constants of type @code{float}.
1213 @section Arrays of Length Zero
1214 @cindex arrays of length zero
1215 @cindex zero-length arrays
1216 @cindex length-zero arrays
1217 @cindex flexible array members
1219 Zero-length arrays are allowed in GNU C@. They are very useful as the
1220 last element of a structure which is really a header for a variable-length
1229 struct line *thisline = (struct line *)
1230 malloc (sizeof (struct line) + this_length);
1231 thisline->length = this_length;
1234 In ISO C89, you would have to give @code{contents} a length of 1, which
1235 means either you waste space or complicate the argument to @code{malloc}.
1237 In ISO C99, you would use a @dfn{flexible array member}, which is
1238 slightly different in syntax and semantics:
1242 Flexible array members are written as @code{contents[]} without
1246 Flexible array members have incomplete type, and so the @code{sizeof}
1247 operator may not be applied. As a quirk of the original implementation
1248 of zero-length arrays, @code{sizeof} evaluates to zero.
1251 Flexible array members may only appear as the last member of a
1252 @code{struct} that is otherwise non-empty.
1255 GCC versions before 3.0 allowed zero-length arrays to be statically
1256 initialized, as if they were flexible arrays. In addition to those
1257 cases that were useful, it also allowed initializations in situations
1258 that would corrupt later data. Non-empty initialization of zero-length
1259 arrays is now treated like any case where there are more initializer
1260 elements than the array holds, in that a suitable warning about "excess
1261 elements in array" is given, and the excess elements (all of them, in
1262 this case) are ignored.
1264 Instead GCC allows static initialization of flexible array members.
1265 This is equivalent to defining a new structure containing the original
1266 structure followed by an array of sufficient size to contain the data.
1267 I.e.@: in the following, @code{f1} is constructed as if it were declared
1273 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
1276 struct f1 f1; int data[3];
1277 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1281 The convenience of this extension is that @code{f1} has the desired
1282 type, eliminating the need to consistently refer to @code{f2.f1}.
1284 This has symmetry with normal static arrays, in that an array of
1285 unknown size is also written with @code{[]}.
1287 Of course, this extension only makes sense if the extra data comes at
1288 the end of a top-level object, as otherwise we would be overwriting
1289 data at subsequent offsets. To avoid undue complication and confusion
1290 with initialization of deeply nested arrays, we simply disallow any
1291 non-empty initialization except when the structure is the top-level
1292 object. For example:
1295 struct foo @{ int x; int y[]; @};
1296 struct bar @{ struct foo z; @};
1298 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1299 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1300 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1301 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1304 @node Variable Length
1305 @section Arrays of Variable Length
1306 @cindex variable-length arrays
1307 @cindex arrays of variable length
1310 Variable-length automatic arrays are allowed in ISO C99, and as an
1311 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1312 implementation of variable-length arrays does not yet conform in detail
1313 to the ISO C99 standard.) These arrays are
1314 declared like any other automatic arrays, but with a length that is not
1315 a constant expression. The storage is allocated at the point of
1316 declaration and deallocated when the brace-level is exited. For
1321 concat_fopen (char *s1, char *s2, char *mode)
1323 char str[strlen (s1) + strlen (s2) + 1];
1326 return fopen (str, mode);
1330 @cindex scope of a variable length array
1331 @cindex variable-length array scope
1332 @cindex deallocating variable length arrays
1333 Jumping or breaking out of the scope of the array name deallocates the
1334 storage. Jumping into the scope is not allowed; you get an error
1337 @cindex @code{alloca} vs variable-length arrays
1338 You can use the function @code{alloca} to get an effect much like
1339 variable-length arrays. The function @code{alloca} is available in
1340 many other C implementations (but not in all). On the other hand,
1341 variable-length arrays are more elegant.
1343 There are other differences between these two methods. Space allocated
1344 with @code{alloca} exists until the containing @emph{function} returns.
1345 The space for a variable-length array is deallocated as soon as the array
1346 name's scope ends. (If you use both variable-length arrays and
1347 @code{alloca} in the same function, deallocation of a variable-length array
1348 will also deallocate anything more recently allocated with @code{alloca}.)
1350 You can also use variable-length arrays as arguments to functions:
1354 tester (int len, char data[len][len])
1360 The length of an array is computed once when the storage is allocated
1361 and is remembered for the scope of the array in case you access it with
1364 If you want to pass the array first and the length afterward, you can
1365 use a forward declaration in the parameter list---another GNU extension.
1369 tester (int len; char data[len][len], int len)
1375 @cindex parameter forward declaration
1376 The @samp{int len} before the semicolon is a @dfn{parameter forward
1377 declaration}, and it serves the purpose of making the name @code{len}
1378 known when the declaration of @code{data} is parsed.
1380 You can write any number of such parameter forward declarations in the
1381 parameter list. They can be separated by commas or semicolons, but the
1382 last one must end with a semicolon, which is followed by the ``real''
1383 parameter declarations. Each forward declaration must match a ``real''
1384 declaration in parameter name and data type. ISO C99 does not support
1385 parameter forward declarations.
1387 @node Variadic Macros
1388 @section Macros with a Variable Number of Arguments.
1389 @cindex variable number of arguments
1390 @cindex macro with variable arguments
1391 @cindex rest argument (in macro)
1392 @cindex variadic macros
1394 In the ISO C standard of 1999, a macro can be declared to accept a
1395 variable number of arguments much as a function can. The syntax for
1396 defining the macro is similar to that of a function. Here is an
1400 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1403 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1404 such a macro, it represents the zero or more tokens until the closing
1405 parenthesis that ends the invocation, including any commas. This set of
1406 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1407 wherever it appears. See the CPP manual for more information.
1409 GCC has long supported variadic macros, and used a different syntax that
1410 allowed you to give a name to the variable arguments just like any other
1411 argument. Here is an example:
1414 #define debug(format, args...) fprintf (stderr, format, args)
1417 This is in all ways equivalent to the ISO C example above, but arguably
1418 more readable and descriptive.
1420 GNU CPP has two further variadic macro extensions, and permits them to
1421 be used with either of the above forms of macro definition.
1423 In standard C, you are not allowed to leave the variable argument out
1424 entirely; but you are allowed to pass an empty argument. For example,
1425 this invocation is invalid in ISO C, because there is no comma after
1432 GNU CPP permits you to completely omit the variable arguments in this
1433 way. In the above examples, the compiler would complain, though since
1434 the expansion of the macro still has the extra comma after the format
1437 To help solve this problem, CPP behaves specially for variable arguments
1438 used with the token paste operator, @samp{##}. If instead you write
1441 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1444 and if the variable arguments are omitted or empty, the @samp{##}
1445 operator causes the preprocessor to remove the comma before it. If you
1446 do provide some variable arguments in your macro invocation, GNU CPP
1447 does not complain about the paste operation and instead places the
1448 variable arguments after the comma. Just like any other pasted macro
1449 argument, these arguments are not macro expanded.
1451 @node Escaped Newlines
1452 @section Slightly Looser Rules for Escaped Newlines
1453 @cindex escaped newlines
1454 @cindex newlines (escaped)
1456 Recently, the non-traditional preprocessor has relaxed its treatment of
1457 escaped newlines. Previously, the newline had to immediately follow a
1458 backslash. The current implementation allows whitespace in the form of
1459 spaces, horizontal and vertical tabs, and form feeds between the
1460 backslash and the subsequent newline. The preprocessor issues a
1461 warning, but treats it as a valid escaped newline and combines the two
1462 lines to form a single logical line. This works within comments and
1463 tokens, including multi-line strings, as well as between tokens.
1464 Comments are @emph{not} treated as whitespace for the purposes of this
1465 relaxation, since they have not yet been replaced with spaces.
1467 @node Multi-line Strings
1468 @section String Literals with Embedded Newlines
1469 @cindex multi-line string literals
1471 As an extension, GNU CPP permits string literals to cross multiple lines
1472 without escaping the embedded newlines. Each embedded newline is
1473 replaced with a single @samp{\n} character in the resulting string
1474 literal, regardless of what form the newline took originally.
1476 CPP currently allows such strings in directives as well (other than the
1477 @samp{#include} family). This is deprecated and will eventually be
1481 @section Non-Lvalue Arrays May Have Subscripts
1482 @cindex subscripting
1483 @cindex arrays, non-lvalue
1485 @cindex subscripting and function values
1486 In ISO C99, arrays that are not lvalues still decay to pointers, and
1487 may be subscripted, although they may not be modified or used after
1488 the next sequence point and the unary @samp{&} operator may not be
1489 applied to them. As an extension, GCC allows such arrays to be
1490 subscripted in C89 mode, though otherwise they do not decay to
1491 pointers outside C99 mode. For example,
1492 this is valid in GNU C though not valid in C89:
1496 struct foo @{int a[4];@};
1502 return f().a[index];
1508 @section Arithmetic on @code{void}- and Function-Pointers
1509 @cindex void pointers, arithmetic
1510 @cindex void, size of pointer to
1511 @cindex function pointers, arithmetic
1512 @cindex function, size of pointer to
1514 In GNU C, addition and subtraction operations are supported on pointers to
1515 @code{void} and on pointers to functions. This is done by treating the
1516 size of a @code{void} or of a function as 1.
1518 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1519 and on function types, and returns 1.
1521 @opindex Wpointer-arith
1522 The option @option{-Wpointer-arith} requests a warning if these extensions
1526 @section Non-Constant Initializers
1527 @cindex initializers, non-constant
1528 @cindex non-constant initializers
1530 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1531 automatic variable are not required to be constant expressions in GNU C@.
1532 Here is an example of an initializer with run-time varying elements:
1535 foo (float f, float g)
1537 float beat_freqs[2] = @{ f-g, f+g @};
1542 @node Compound Literals
1543 @section Compound Literals
1544 @cindex constructor expressions
1545 @cindex initializations in expressions
1546 @cindex structures, constructor expression
1547 @cindex expressions, constructor
1548 @cindex compound literals
1549 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1551 ISO C99 supports compound literals. A compound literal looks like
1552 a cast containing an initializer. Its value is an object of the
1553 type specified in the cast, containing the elements specified in
1554 the initializer; it is an lvalue. As an extension, GCC supports
1555 compound literals in C89 mode and in C++.
1557 Usually, the specified type is a structure. Assume that
1558 @code{struct foo} and @code{structure} are declared as shown:
1561 struct foo @{int a; char b[2];@} structure;
1565 Here is an example of constructing a @code{struct foo} with a compound literal:
1568 structure = ((struct foo) @{x + y, 'a', 0@});
1572 This is equivalent to writing the following:
1576 struct foo temp = @{x + y, 'a', 0@};
1581 You can also construct an array. If all the elements of the compound literal
1582 are (made up of) simple constant expressions, suitable for use in
1583 initializers of objects of static storage duration, then the compound
1584 literal can be coerced to a pointer to its first element and used in
1585 such an initializer, as shown here:
1588 char **foo = (char *[]) @{ "x", "y", "z" @};
1591 Compound literals for scalar types and union types are is
1592 also allowed, but then the compound literal is equivalent
1595 As a GNU extension, GCC allows initialization of objects with static storage
1596 duration by compound literals (which is not possible in ISO C99, because
1597 the initializer is not a constant).
1598 It is handled as if the object was initialized only with the bracket
1599 enclosed list if compound literal's and object types match.
1600 The initializer list of the compound literal must be constant.
1601 If the object being initialized has array type of unknown size, the size is
1602 determined by compound literal size.
1605 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1606 static int y[] = (int []) @{1, 2, 3@};
1607 static int z[] = (int [3]) @{1@};
1611 The above lines are equivalent to the following:
1613 static struct foo x = @{1, 'a', 'b'@};
1614 static int y[] = @{1, 2, 3@};
1615 static int z[] = @{1, 0, 0@};
1618 @node Designated Inits
1619 @section Designated Initializers
1620 @cindex initializers with labeled elements
1621 @cindex labeled elements in initializers
1622 @cindex case labels in initializers
1623 @cindex designated initializers
1625 Standard C89 requires the elements of an initializer to appear in a fixed
1626 order, the same as the order of the elements in the array or structure
1629 In ISO C99 you can give the elements in any order, specifying the array
1630 indices or structure field names they apply to, and GNU C allows this as
1631 an extension in C89 mode as well. This extension is not
1632 implemented in GNU C++.
1634 To specify an array index, write
1635 @samp{[@var{index}] =} before the element value. For example,
1638 int a[6] = @{ [4] = 29, [2] = 15 @};
1645 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1649 The index values must be constant expressions, even if the array being
1650 initialized is automatic.
1652 An alternative syntax for this which has been obsolete since GCC 2.5 but
1653 GCC still accepts is to write @samp{[@var{index}]} before the element
1654 value, with no @samp{=}.
1656 To initialize a range of elements to the same value, write
1657 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1658 extension. For example,
1661 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1665 If the value in it has side-effects, the side-effects will happen only once,
1666 not for each initialized field by the range initializer.
1669 Note that the length of the array is the highest value specified
1672 In a structure initializer, specify the name of a field to initialize
1673 with @samp{.@var{fieldname} =} before the element value. For example,
1674 given the following structure,
1677 struct point @{ int x, y; @};
1681 the following initialization
1684 struct point p = @{ .y = yvalue, .x = xvalue @};
1691 struct point p = @{ xvalue, yvalue @};
1694 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1695 @samp{@var{fieldname}:}, as shown here:
1698 struct point p = @{ y: yvalue, x: xvalue @};
1702 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1703 @dfn{designator}. You can also use a designator (or the obsolete colon
1704 syntax) when initializing a union, to specify which element of the union
1705 should be used. For example,
1708 union foo @{ int i; double d; @};
1710 union foo f = @{ .d = 4 @};
1714 will convert 4 to a @code{double} to store it in the union using
1715 the second element. By contrast, casting 4 to type @code{union foo}
1716 would store it into the union as the integer @code{i}, since it is
1717 an integer. (@xref{Cast to Union}.)
1719 You can combine this technique of naming elements with ordinary C
1720 initialization of successive elements. Each initializer element that
1721 does not have a designator applies to the next consecutive element of the
1722 array or structure. For example,
1725 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1732 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1735 Labeling the elements of an array initializer is especially useful
1736 when the indices are characters or belong to an @code{enum} type.
1741 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1742 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1745 @cindex designator lists
1746 You can also write a series of @samp{.@var{fieldname}} and
1747 @samp{[@var{index}]} designators before an @samp{=} to specify a
1748 nested subobject to initialize; the list is taken relative to the
1749 subobject corresponding to the closest surrounding brace pair. For
1750 example, with the @samp{struct point} declaration above:
1753 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1757 If the same field is initialized multiple times, it will have value from
1758 the last initialization. If any such overridden initialization has
1759 side-effect, it is unspecified whether the side-effect happens or not.
1760 Currently, gcc will discard them and issue a warning.
1763 @section Case Ranges
1765 @cindex ranges in case statements
1767 You can specify a range of consecutive values in a single @code{case} label,
1771 case @var{low} ... @var{high}:
1775 This has the same effect as the proper number of individual @code{case}
1776 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1778 This feature is especially useful for ranges of ASCII character codes:
1784 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1785 it may be parsed wrong when you use it with integer values. For example,
1800 @section Cast to a Union Type
1801 @cindex cast to a union
1802 @cindex union, casting to a
1804 A cast to union type is similar to other casts, except that the type
1805 specified is a union type. You can specify the type either with
1806 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1807 a constructor though, not a cast, and hence does not yield an lvalue like
1808 normal casts. (@xref{Compound Literals}.)
1810 The types that may be cast to the union type are those of the members
1811 of the union. Thus, given the following union and variables:
1814 union foo @{ int i; double d; @};
1820 both @code{x} and @code{y} can be cast to type @code{union foo}.
1822 Using the cast as the right-hand side of an assignment to a variable of
1823 union type is equivalent to storing in a member of the union:
1828 u = (union foo) x @equiv{} u.i = x
1829 u = (union foo) y @equiv{} u.d = y
1832 You can also use the union cast as a function argument:
1835 void hack (union foo);
1837 hack ((union foo) x);
1840 @node Mixed Declarations
1841 @section Mixed Declarations and Code
1842 @cindex mixed declarations and code
1843 @cindex declarations, mixed with code
1844 @cindex code, mixed with declarations
1846 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1847 within compound statements. As an extension, GCC also allows this in
1848 C89 mode. For example, you could do:
1857 Each identifier is visible from where it is declared until the end of
1858 the enclosing block.
1860 @node Function Attributes
1861 @section Declaring Attributes of Functions
1862 @cindex function attributes
1863 @cindex declaring attributes of functions
1864 @cindex functions that never return
1865 @cindex functions that have no side effects
1866 @cindex functions in arbitrary sections
1867 @cindex functions that behave like malloc
1868 @cindex @code{volatile} applied to function
1869 @cindex @code{const} applied to function
1870 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1871 @cindex functions that are passed arguments in registers on the 386
1872 @cindex functions that pop the argument stack on the 386
1873 @cindex functions that do not pop the argument stack on the 386
1875 In GNU C, you declare certain things about functions called in your program
1876 which help the compiler optimize function calls and check your code more
1879 The keyword @code{__attribute__} allows you to specify special
1880 attributes when making a declaration. This keyword is followed by an
1881 attribute specification inside double parentheses. The following
1882 attributes are currently defined for functions on all targets:
1883 @code{noreturn}, @code{noinline}, @code{always_inline},
1884 @code{pure}, @code{const},
1885 @code{format}, @code{format_arg}, @code{no_instrument_function},
1886 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1887 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc}, and
1888 @code{alias}. Several other attributes are defined for functions on
1889 particular target systems. Other attributes, including @code{section}
1890 are supported for variables declarations (@pxref{Variable Attributes})
1891 and for types (@pxref{Type Attributes}).
1893 You may also specify attributes with @samp{__} preceding and following
1894 each keyword. This allows you to use them in header files without
1895 being concerned about a possible macro of the same name. For example,
1896 you may use @code{__noreturn__} instead of @code{noreturn}.
1898 @xref{Attribute Syntax}, for details of the exact syntax for using
1902 @cindex @code{noreturn} function attribute
1904 A few standard library functions, such as @code{abort} and @code{exit},
1905 cannot return. GCC knows this automatically. Some programs define
1906 their own functions that never return. You can declare them
1907 @code{noreturn} to tell the compiler this fact. For example,
1911 void fatal () __attribute__ ((noreturn));
1916 @dots{} /* @r{Print error message.} */ @dots{}
1922 The @code{noreturn} keyword tells the compiler to assume that
1923 @code{fatal} cannot return. It can then optimize without regard to what
1924 would happen if @code{fatal} ever did return. This makes slightly
1925 better code. More importantly, it helps avoid spurious warnings of
1926 uninitialized variables.
1928 Do not assume that registers saved by the calling function are
1929 restored before calling the @code{noreturn} function.
1931 It does not make sense for a @code{noreturn} function to have a return
1932 type other than @code{void}.
1934 The attribute @code{noreturn} is not implemented in GCC versions
1935 earlier than 2.5. An alternative way to declare that a function does
1936 not return, which works in the current version and in some older
1937 versions, is as follows:
1940 typedef void voidfn ();
1942 volatile voidfn fatal;
1945 @cindex @code{noinline} function attribute
1947 This function attribute prevents a function from being considered for
1950 @cindex @code{always_inline} function attribute
1952 Generally, functions are not inlined unless optimization is specified.
1953 For functions declared inline, this attribute inlines the function even
1954 if no optimization level was specified.
1956 @cindex @code{pure} function attribute
1958 Many functions have no effects except the return value and their
1959 return value depends only on the parameters and/or global variables.
1960 Such a function can be subject
1961 to common subexpression elimination and loop optimization just as an
1962 arithmetic operator would be. These functions should be declared
1963 with the attribute @code{pure}. For example,
1966 int square (int) __attribute__ ((pure));
1970 says that the hypothetical function @code{square} is safe to call
1971 fewer times than the program says.
1973 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
1974 Interesting non-pure functions are functions with infinite loops or those
1975 depending on volatile memory or other system resource, that may change between
1976 two consecutive calls (such as @code{feof} in a multithreading environment).
1978 The attribute @code{pure} is not implemented in GCC versions earlier
1980 @cindex @code{const} function attribute
1982 Many functions do not examine any values except their arguments, and
1983 have no effects except the return value. Basically this is just slightly
1984 more strict class than the @code{pure} attribute above, since function is not
1985 allowed to read global memory.
1987 @cindex pointer arguments
1988 Note that a function that has pointer arguments and examines the data
1989 pointed to must @emph{not} be declared @code{const}. Likewise, a
1990 function that calls a non-@code{const} function usually must not be
1991 @code{const}. It does not make sense for a @code{const} function to
1994 The attribute @code{const} is not implemented in GCC versions earlier
1995 than 2.5. An alternative way to declare that a function has no side
1996 effects, which works in the current version and in some older versions,
2000 typedef int intfn ();
2002 extern const intfn square;
2005 This approach does not work in GNU C++ from 2.6.0 on, since the language
2006 specifies that the @samp{const} must be attached to the return value.
2009 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2010 @cindex @code{format} function attribute
2012 The @code{format} attribute specifies that a function takes @code{printf},
2013 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
2014 should be type-checked against a format string. For example, the
2019 my_printf (void *my_object, const char *my_format, ...)
2020 __attribute__ ((format (printf, 2, 3)));
2024 causes the compiler to check the arguments in calls to @code{my_printf}
2025 for consistency with the @code{printf} style format string argument
2028 The parameter @var{archetype} determines how the format string is
2029 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
2030 or @code{strfmon}. (You can also use @code{__printf__},
2031 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
2032 parameter @var{string-index} specifies which argument is the format
2033 string argument (starting from 1), while @var{first-to-check} is the
2034 number of the first argument to check against the format string. For
2035 functions where the arguments are not available to be checked (such as
2036 @code{vprintf}), specify the third parameter as zero. In this case the
2037 compiler only checks the format string for consistency. For
2038 @code{strftime} formats, the third parameter is required to be zero.
2040 In the example above, the format string (@code{my_format}) is the second
2041 argument of the function @code{my_print}, and the arguments to check
2042 start with the third argument, so the correct parameters for the format
2043 attribute are 2 and 3.
2045 @opindex ffreestanding
2046 The @code{format} attribute allows you to identify your own functions
2047 which take format strings as arguments, so that GCC can check the
2048 calls to these functions for errors. The compiler always (unless
2049 @option{-ffreestanding} is used) checks formats
2050 for the standard library functions @code{printf}, @code{fprintf},
2051 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2052 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2053 warnings are requested (using @option{-Wformat}), so there is no need to
2054 modify the header file @file{stdio.h}. In C99 mode, the functions
2055 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2056 @code{vsscanf} are also checked. Except in strictly conforming C
2057 standard modes, the X/Open function @code{strfmon} is also checked as
2058 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2059 @xref{C Dialect Options,,Options Controlling C Dialect}.
2061 @item format_arg (@var{string-index})
2062 @cindex @code{format_arg} function attribute
2063 @opindex Wformat-nonliteral
2064 The @code{format_arg} attribute specifies that a function takes a format
2065 string for a @code{printf}, @code{scanf}, @code{strftime} or
2066 @code{strfmon} style function and modifies it (for example, to translate
2067 it into another language), so the result can be passed to a
2068 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2069 function (with the remaining arguments to the format function the same
2070 as they would have been for the unmodified string). For example, the
2075 my_dgettext (char *my_domain, const char *my_format)
2076 __attribute__ ((format_arg (2)));
2080 causes the compiler to check the arguments in calls to a @code{printf},
2081 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2082 format string argument is a call to the @code{my_dgettext} function, for
2083 consistency with the format string argument @code{my_format}. If the
2084 @code{format_arg} attribute had not been specified, all the compiler
2085 could tell in such calls to format functions would be that the format
2086 string argument is not constant; this would generate a warning when
2087 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2088 without the attribute.
2090 The parameter @var{string-index} specifies which argument is the format
2091 string argument (starting from 1).
2093 The @code{format-arg} attribute allows you to identify your own
2094 functions which modify format strings, so that GCC can check the
2095 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2096 type function whose operands are a call to one of your own function.
2097 The compiler always treats @code{gettext}, @code{dgettext}, and
2098 @code{dcgettext} in this manner except when strict ISO C support is
2099 requested by @option{-ansi} or an appropriate @option{-std} option, or
2100 @option{-ffreestanding} is used. @xref{C Dialect Options,,Options
2101 Controlling C Dialect}.
2103 @item no_instrument_function
2104 @cindex @code{no_instrument_function} function attribute
2105 @opindex finstrument-functions
2106 If @option{-finstrument-functions} is given, profiling function calls will
2107 be generated at entry and exit of most user-compiled functions.
2108 Functions with this attribute will not be so instrumented.
2110 @item section ("@var{section-name}")
2111 @cindex @code{section} function attribute
2112 Normally, the compiler places the code it generates in the @code{text} section.
2113 Sometimes, however, you need additional sections, or you need certain
2114 particular functions to appear in special sections. The @code{section}
2115 attribute specifies that a function lives in a particular section.
2116 For example, the declaration:
2119 extern void foobar (void) __attribute__ ((section ("bar")));
2123 puts the function @code{foobar} in the @code{bar} section.
2125 Some file formats do not support arbitrary sections so the @code{section}
2126 attribute is not available on all platforms.
2127 If you need to map the entire contents of a module to a particular
2128 section, consider using the facilities of the linker instead.
2132 @cindex @code{constructor} function attribute
2133 @cindex @code{destructor} function attribute
2134 The @code{constructor} attribute causes the function to be called
2135 automatically before execution enters @code{main ()}. Similarly, the
2136 @code{destructor} attribute causes the function to be called
2137 automatically after @code{main ()} has completed or @code{exit ()} has
2138 been called. Functions with these attributes are useful for
2139 initializing data that will be used implicitly during the execution of
2142 These attributes are not currently implemented for Objective-C@.
2144 @cindex @code{unused} attribute.
2146 This attribute, attached to a function, means that the function is meant
2147 to be possibly unused. GCC will not produce a warning for this
2148 function. GNU C++ does not currently support this attribute as
2149 definitions without parameters are valid in C++.
2151 @cindex @code{used} attribute.
2153 This attribute, attached to a function, means that code must be emitted
2154 for the function even if it appears that the function is not referenced.
2155 This is useful, for example, when the function is referenced only in
2158 @cindex @code{deprecated} attribute.
2160 The @code{deprecated} attribute results in a warning if the function
2161 is used anywhere in the source file. This is useful when identifying
2162 functions that are expected to be removed in a future version of a
2163 program. The warning also includes the location of the declaration
2164 of the deprecated function, to enable users to easily find further
2165 information about why the function is deprecated, or what they should
2166 do instead. Note that the warnings only occurs for uses:
2169 int old_fn () __attribute__ ((deprecated));
2171 int (*fn_ptr)() = old_fn;
2174 results in a warning on line 3 but not line 2.
2176 The @code{deprecated} attribute can also be used for variables and
2177 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2180 @cindex @code{weak} attribute
2181 The @code{weak} attribute causes the declaration to be emitted as a weak
2182 symbol rather than a global. This is primarily useful in defining
2183 library functions which can be overridden in user code, though it can
2184 also be used with non-function declarations. Weak symbols are supported
2185 for ELF targets, and also for a.out targets when using the GNU assembler
2189 @cindex @code{malloc} attribute
2190 The @code{malloc} attribute is used to tell the compiler that a function
2191 may be treated as if it were the malloc function. The compiler assumes
2192 that calls to malloc result in a pointers that cannot alias anything.
2193 This will often improve optimization.
2195 @item alias ("@var{target}")
2196 @cindex @code{alias} attribute
2197 The @code{alias} attribute causes the declaration to be emitted as an
2198 alias for another symbol, which must be specified. For instance,
2201 void __f () @{ /* do something */; @}
2202 void f () __attribute__ ((weak, alias ("__f")));
2205 declares @samp{f} to be a weak alias for @samp{__f}. In C++, the
2206 mangled name for the target must be used.
2208 Not all target machines support this attribute.
2210 @item regparm (@var{number})
2211 @cindex functions that are passed arguments in registers on the 386
2212 On the Intel 386, the @code{regparm} attribute causes the compiler to
2213 pass up to @var{number} integer arguments in registers EAX,
2214 EDX, and ECX instead of on the stack. Functions that take a
2215 variable number of arguments will continue to be passed all of their
2216 arguments on the stack.
2219 @cindex functions that pop the argument stack on the 386
2220 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2221 assume that the called function will pop off the stack space used to
2222 pass arguments, unless it takes a variable number of arguments.
2224 The PowerPC compiler for Windows NT currently ignores the @code{stdcall}
2228 @cindex functions that do pop the argument stack on the 386
2230 On the Intel 386, the @code{cdecl} attribute causes the compiler to
2231 assume that the calling function will pop off the stack space used to
2232 pass arguments. This is
2233 useful to override the effects of the @option{-mrtd} switch.
2235 The PowerPC compiler for Windows NT currently ignores the @code{cdecl}
2239 @cindex functions called via pointer on the RS/6000 and PowerPC
2240 On the RS/6000 and PowerPC, the @code{longcall} attribute causes the
2241 compiler to always call the function via a pointer, so that functions
2242 which reside further than 64 megabytes (67,108,864 bytes) from the
2243 current location can be called.
2245 @item long_call/short_call
2246 @cindex indirect calls on ARM
2247 This attribute allows to specify how to call a particular function on
2248 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2249 command line switch and @code{#pragma long_calls} settings. The
2250 @code{long_call} attribute causes the compiler to always call the
2251 function by first loading its address into a register and then using the
2252 contents of that register. The @code{short_call} attribute always places
2253 the offset to the function from the call site into the @samp{BL}
2254 instruction directly.
2257 @cindex functions which are imported from a dll on PowerPC Windows NT
2258 On the PowerPC running Windows NT, the @code{dllimport} attribute causes
2259 the compiler to call the function via a global pointer to the function
2260 pointer that is set up by the Windows NT dll library. The pointer name
2261 is formed by combining @code{__imp_} and the function name.
2264 @cindex functions which are exported from a dll on PowerPC Windows NT
2265 On the PowerPC running Windows NT, the @code{dllexport} attribute causes
2266 the compiler to provide a global pointer to the function pointer, so
2267 that it can be called with the @code{dllimport} attribute. The pointer
2268 name is formed by combining @code{__imp_} and the function name.
2270 @item exception (@var{except-func} [, @var{except-arg}])
2271 @cindex functions which specify exception handling on PowerPC Windows NT
2272 On the PowerPC running Windows NT, the @code{exception} attribute causes
2273 the compiler to modify the structured exception table entry it emits for
2274 the declared function. The string or identifier @var{except-func} is
2275 placed in the third entry of the structured exception table. It
2276 represents a function, which is called by the exception handling
2277 mechanism if an exception occurs. If it was specified, the string or
2278 identifier @var{except-arg} is placed in the fourth entry of the
2279 structured exception table.
2281 @item function_vector
2282 @cindex calling functions through the function vector on the H8/300 processors
2283 Use this attribute on the H8/300 and H8/300H to indicate that the specified
2284 function should be called through the function vector. Calling a
2285 function through the function vector will reduce code size, however;
2286 the function vector has a limited size (maximum 128 entries on the H8/300
2287 and 64 entries on the H8/300H) and shares space with the interrupt vector.
2289 You must use GAS and GLD from GNU binutils version 2.7 or later for
2290 this attribute to work correctly.
2293 @cindex interrupt handler functions
2294 Use this attribute on the ARM, AVR, M32R/D and Xstormy16 ports to indicate
2295 that the specified function is an interrupt handler. The compiler will
2296 generate function entry and exit sequences suitable for use in an
2297 interrupt handler when this attribute is present.
2299 Note, interrupt handlers for the H8/300, H8/300H and SH processors can
2300 be specified via the @code{interrupt_handler} attribute.
2302 Note, on the AVR interrupts will be enabled inside the function.
2304 Note, for the ARM you can specify the kind of interrupt to be handled by
2305 adding an optional parameter to the interrupt attribute like this:
2308 void f () __attribute__ ((interrupt ("IRQ")));
2311 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2313 @item interrupt_handler
2314 @cindex interrupt handler functions on the H8/300 and SH processors
2315 Use this attribute on the H8/300, H8/300H and SH to indicate that the
2316 specified function is an interrupt handler. The compiler will generate
2317 function entry and exit sequences suitable for use in an interrupt
2318 handler when this attribute is present.
2321 Use this attribute on the SH to indicate an @code{interrupt_handler}
2322 function should switch to an alternate stack. It expects a string
2323 argument that names a global variable holding the address of the
2328 void f () __attribute__ ((interrupt_handler,
2329 sp_switch ("alt_stack")));
2333 Use this attribute on the SH for an @code{interrupt_handle} to return using
2334 @code{trapa} instead of @code{rte}. This attribute expects an integer
2335 argument specifying the trap number to be used.
2338 @cindex eight bit data on the H8/300 and H8/300H
2339 Use this attribute on the H8/300 and H8/300H to indicate that the specified
2340 variable should be placed into the eight bit data section.
2341 The compiler will generate more efficient code for certain operations
2342 on data in the eight bit data area. Note the eight bit data area is limited to
2345 You must use GAS and GLD from GNU binutils version 2.7 or later for
2346 this attribute to work correctly.
2349 @cindex tiny data section on the H8/300H
2350 Use this attribute on the H8/300H to indicate that the specified
2351 variable should be placed into the tiny data section.
2352 The compiler will generate more efficient code for loads and stores
2353 on data in the tiny data section. Note the tiny data area is limited to
2354 slightly under 32kbytes of data.
2357 @cindex signal handler functions on the AVR processors
2358 Use this attribute on the AVR to indicate that the specified
2359 function is an signal handler. The compiler will generate function
2360 entry and exit sequences suitable for use in an signal handler when this
2361 attribute is present. Interrupts will be disabled inside function.
2364 @cindex function without a prologue/epilogue code
2365 Use this attribute on the ARM or AVR ports to indicate that the specified
2366 function do not need prologue/epilogue sequences generated by the
2367 compiler. It is up to the programmer to provide these sequences.
2369 @item model (@var{model-name})
2370 @cindex function addressability on the M32R/D
2371 Use this attribute on the M32R/D to set the addressability of an object,
2372 and the code generated for a function.
2373 The identifier @var{model-name} is one of @code{small}, @code{medium},
2374 or @code{large}, representing each of the code models.
2376 Small model objects live in the lower 16MB of memory (so that their
2377 addresses can be loaded with the @code{ld24} instruction), and are
2378 callable with the @code{bl} instruction.
2380 Medium model objects may live anywhere in the 32-bit address space (the
2381 compiler will generate @code{seth/add3} instructions to load their addresses),
2382 and are callable with the @code{bl} instruction.
2384 Large model objects may live anywhere in the 32-bit address space (the
2385 compiler will generate @code{seth/add3} instructions to load their addresses),
2386 and may not be reachable with the @code{bl} instruction (the compiler will
2387 generate the much slower @code{seth/add3/jl} instruction sequence).
2391 You can specify multiple attributes in a declaration by separating them
2392 by commas within the double parentheses or by immediately following an
2393 attribute declaration with another attribute declaration.
2395 @cindex @code{#pragma}, reason for not using
2396 @cindex pragma, reason for not using
2397 Some people object to the @code{__attribute__} feature, suggesting that
2398 ISO C's @code{#pragma} should be used instead. At the time
2399 @code{__attribute__} was designed, there were two reasons for not doing
2404 It is impossible to generate @code{#pragma} commands from a macro.
2407 There is no telling what the same @code{#pragma} might mean in another
2411 These two reasons applied to almost any application that might have been
2412 proposed for @code{#pragma}. It was basically a mistake to use
2413 @code{#pragma} for @emph{anything}.
2415 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2416 to be generated from macros. In addition, a @code{#pragma GCC}
2417 namespace is now in use for GCC-specific pragmas. However, it has been
2418 found convenient to use @code{__attribute__} to achieve a natural
2419 attachment of attributes to their corresponding declarations, whereas
2420 @code{#pragma GCC} is of use for constructs that do not naturally form
2421 part of the grammar. @xref{Other Directives,,Miscellaneous
2422 Preprocessing Directives, cpp, The C Preprocessor}.
2424 @node Attribute Syntax
2425 @section Attribute Syntax
2426 @cindex attribute syntax
2428 This section describes the syntax with which @code{__attribute__} may be
2429 used, and the constructs to which attribute specifiers bind, for the C
2430 language. Some details may vary for C++ and Objective-C@. Because of
2431 infelicities in the grammar for attributes, some forms described here
2432 may not be successfully parsed in all cases.
2434 There are some problems with the semantics of attributes in C++. For
2435 example, there are no manglings for attributes, although they may affect
2436 code generation, so problems may arise when attributed types are used in
2437 conjunction with templates or overloading. Similarly, @code{typeid}
2438 does not distinguish between types with different attributes. Support
2439 for attributes in C++ may be restricted in future to attributes on
2440 declarations only, but not on nested declarators.
2442 @xref{Function Attributes}, for details of the semantics of attributes
2443 applying to functions. @xref{Variable Attributes}, for details of the
2444 semantics of attributes applying to variables. @xref{Type Attributes},
2445 for details of the semantics of attributes applying to structure, union
2446 and enumerated types.
2448 An @dfn{attribute specifier} is of the form
2449 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2450 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2451 each attribute is one of the following:
2455 Empty. Empty attributes are ignored.
2458 A word (which may be an identifier such as @code{unused}, or a reserved
2459 word such as @code{const}).
2462 A word, followed by, in parentheses, parameters for the attribute.
2463 These parameters take one of the following forms:
2467 An identifier. For example, @code{mode} attributes use this form.
2470 An identifier followed by a comma and a non-empty comma-separated list
2471 of expressions. For example, @code{format} attributes use this form.
2474 A possibly empty comma-separated list of expressions. For example,
2475 @code{format_arg} attributes use this form with the list being a single
2476 integer constant expression, and @code{alias} attributes use this form
2477 with the list being a single string constant.
2481 An @dfn{attribute specifier list} is a sequence of one or more attribute
2482 specifiers, not separated by any other tokens.
2484 An attribute specifier list may appear after the colon following a
2485 label, other than a @code{case} or @code{default} label. The only
2486 attribute it makes sense to use after a label is @code{unused}. This
2487 feature is intended for code generated by programs which contains labels
2488 that may be unused but which is compiled with @option{-Wall}. It would
2489 not normally be appropriate to use in it human-written code, though it
2490 could be useful in cases where the code that jumps to the label is
2491 contained within an @code{#ifdef} conditional.
2493 An attribute specifier list may appear as part of a @code{struct},
2494 @code{union} or @code{enum} specifier. It may go either immediately
2495 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2496 the closing brace. It is ignored if the content of the structure, union
2497 or enumerated type is not defined in the specifier in which the
2498 attribute specifier list is used---that is, in usages such as
2499 @code{struct __attribute__((foo)) bar} with no following opening brace.
2500 Where attribute specifiers follow the closing brace, they are considered
2501 to relate to the structure, union or enumerated type defined, not to any
2502 enclosing declaration the type specifier appears in, and the type
2503 defined is not complete until after the attribute specifiers.
2504 @c Otherwise, there would be the following problems: a shift/reduce
2505 @c conflict between attributes binding the struct/union/enum and
2506 @c binding to the list of specifiers/qualifiers; and "aligned"
2507 @c attributes could use sizeof for the structure, but the size could be
2508 @c changed later by "packed" attributes.
2510 Otherwise, an attribute specifier appears as part of a declaration,
2511 counting declarations of unnamed parameters and type names, and relates
2512 to that declaration (which may be nested in another declaration, for
2513 example in the case of a parameter declaration), or to a particular declarator
2514 within a declaration. Where an
2515 attribute specifier is applied to a parameter declared as a function or
2516 an array, it should apply to the function or array rather than the
2517 pointer to which the parameter is implicitly converted, but this is not
2518 yet correctly implemented.
2520 Any list of specifiers and qualifiers at the start of a declaration may
2521 contain attribute specifiers, whether or not such a list may in that
2522 context contain storage class specifiers. (Some attributes, however,
2523 are essentially in the nature of storage class specifiers, and only make
2524 sense where storage class specifiers may be used; for example,
2525 @code{section}.) There is one necessary limitation to this syntax: the
2526 first old-style parameter declaration in a function definition cannot
2527 begin with an attribute specifier, because such an attribute applies to
2528 the function instead by syntax described below (which, however, is not
2529 yet implemented in this case). In some other cases, attribute
2530 specifiers are permitted by this grammar but not yet supported by the
2531 compiler. All attribute specifiers in this place relate to the
2532 declaration as a whole. In the obsolescent usage where a type of
2533 @code{int} is implied by the absence of type specifiers, such a list of
2534 specifiers and qualifiers may be an attribute specifier list with no
2535 other specifiers or qualifiers.
2537 An attribute specifier list may appear immediately before a declarator
2538 (other than the first) in a comma-separated list of declarators in a
2539 declaration of more than one identifier using a single list of
2540 specifiers and qualifiers. Such attribute specifiers apply
2541 only to the identifier before whose declarator they appear. For
2545 __attribute__((noreturn)) void d0 (void),
2546 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2551 the @code{noreturn} attribute applies to all the functions
2552 declared; the @code{format} attribute only applies to @code{d1}.
2554 An attribute specifier list may appear immediately before the comma,
2555 @code{=} or semicolon terminating the declaration of an identifier other
2556 than a function definition. At present, such attribute specifiers apply
2557 to the declared object or function, but in future they may attach to the
2558 outermost adjacent declarator. In simple cases there is no difference,
2559 but, for example, in
2562 void (****f)(void) __attribute__((noreturn));
2566 at present the @code{noreturn} attribute applies to @code{f}, which
2567 causes a warning since @code{f} is not a function, but in future it may
2568 apply to the function @code{****f}. The precise semantics of what
2569 attributes in such cases will apply to are not yet specified. Where an
2570 assembler name for an object or function is specified (@pxref{Asm
2571 Labels}), at present the attribute must follow the @code{asm}
2572 specification; in future, attributes before the @code{asm} specification
2573 may apply to the adjacent declarator, and those after it to the declared
2576 An attribute specifier list may, in future, be permitted to appear after
2577 the declarator in a function definition (before any old-style parameter
2578 declarations or the function body).
2580 Attribute specifiers may be mixed with type qualifiers appearing inside
2581 the @code{[]} of a parameter array declarator, in the C99 construct by
2582 which such qualifiers are applied to the pointer to which the array is
2583 implicitly converted. Such attribute specifiers apply to the pointer,
2584 not to the array, but at present this is not implemented and they are
2587 An attribute specifier list may appear at the start of a nested
2588 declarator. At present, there are some limitations in this usage: the
2589 attributes correctly apply to the declarator, but for most individual
2590 attributes the semantics this implies are not implemented.
2591 When attribute specifiers follow the @code{*} of a pointer
2592 declarator, they may be mixed with any type qualifiers present.
2593 The following describes the formal semantics of this syntax. It will make the
2594 most sense if you are familiar with the formal specification of
2595 declarators in the ISO C standard.
2597 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2598 D1}, where @code{T} contains declaration specifiers that specify a type
2599 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2600 contains an identifier @var{ident}. The type specified for @var{ident}
2601 for derived declarators whose type does not include an attribute
2602 specifier is as in the ISO C standard.
2604 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2605 and the declaration @code{T D} specifies the type
2606 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2607 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2608 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2610 If @code{D1} has the form @code{*
2611 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2612 declaration @code{T D} specifies the type
2613 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2614 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2615 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2621 void (__attribute__((noreturn)) ****f) (void);
2625 specifies the type ``pointer to pointer to pointer to pointer to
2626 non-returning function returning @code{void}''. As another example,
2629 char *__attribute__((aligned(8))) *f;
2633 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2634 Note again that this does not work with most attributes; for example,
2635 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2636 is not yet supported.
2638 For compatibility with existing code written for compiler versions that
2639 did not implement attributes on nested declarators, some laxity is
2640 allowed in the placing of attributes. If an attribute that only applies
2641 to types is applied to a declaration, it will be treated as applying to
2642 the type of that declaration. If an attribute that only applies to
2643 declarations is applied to the type of a declaration, it will be treated
2644 as applying to that declaration; and, for compatibility with code
2645 placing the attributes immediately before the identifier declared, such
2646 an attribute applied to a function return type will be treated as
2647 applying to the function type, and such an attribute applied to an array
2648 element type will be treated as applying to the array type. If an
2649 attribute that only applies to function types is applied to a
2650 pointer-to-function type, it will be treated as applying to the pointer
2651 target type; if such an attribute is applied to a function return type
2652 that is not a pointer-to-function type, it will be treated as applying
2653 to the function type.
2655 @node Function Prototypes
2656 @section Prototypes and Old-Style Function Definitions
2657 @cindex function prototype declarations
2658 @cindex old-style function definitions
2659 @cindex promotion of formal parameters
2661 GNU C extends ISO C to allow a function prototype to override a later
2662 old-style non-prototype definition. Consider the following example:
2665 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2672 /* @r{Prototype function declaration.} */
2673 int isroot P((uid_t));
2675 /* @r{Old-style function definition.} */
2677 isroot (x) /* ??? lossage here ??? */
2684 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2685 not allow this example, because subword arguments in old-style
2686 non-prototype definitions are promoted. Therefore in this example the
2687 function definition's argument is really an @code{int}, which does not
2688 match the prototype argument type of @code{short}.
2690 This restriction of ISO C makes it hard to write code that is portable
2691 to traditional C compilers, because the programmer does not know
2692 whether the @code{uid_t} type is @code{short}, @code{int}, or
2693 @code{long}. Therefore, in cases like these GNU C allows a prototype
2694 to override a later old-style definition. More precisely, in GNU C, a
2695 function prototype argument type overrides the argument type specified
2696 by a later old-style definition if the former type is the same as the
2697 latter type before promotion. Thus in GNU C the above example is
2698 equivalent to the following:
2711 GNU C++ does not support old-style function definitions, so this
2712 extension is irrelevant.
2715 @section C++ Style Comments
2717 @cindex C++ comments
2718 @cindex comments, C++ style
2720 In GNU C, you may use C++ style comments, which start with @samp{//} and
2721 continue until the end of the line. Many other C implementations allow
2722 such comments, and they are likely to be in a future C standard.
2723 However, C++ style comments are not recognized if you specify
2724 @w{@option{-ansi}}, a @option{-std} option specifying a version of ISO C
2725 before C99, or @w{@option{-traditional}}, since they are incompatible
2726 with traditional constructs like @code{dividend//*comment*/divisor}.
2729 @section Dollar Signs in Identifier Names
2731 @cindex dollar signs in identifier names
2732 @cindex identifier names, dollar signs in
2734 In GNU C, you may normally use dollar signs in identifier names.
2735 This is because many traditional C implementations allow such identifiers.
2736 However, dollar signs in identifiers are not supported on a few target
2737 machines, typically because the target assembler does not allow them.
2739 @node Character Escapes
2740 @section The Character @key{ESC} in Constants
2742 You can use the sequence @samp{\e} in a string or character constant to
2743 stand for the ASCII character @key{ESC}.
2746 @section Inquiring on Alignment of Types or Variables
2748 @cindex type alignment
2749 @cindex variable alignment
2751 The keyword @code{__alignof__} allows you to inquire about how an object
2752 is aligned, or the minimum alignment usually required by a type. Its
2753 syntax is just like @code{sizeof}.
2755 For example, if the target machine requires a @code{double} value to be
2756 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2757 This is true on many RISC machines. On more traditional machine
2758 designs, @code{__alignof__ (double)} is 4 or even 2.
2760 Some machines never actually require alignment; they allow reference to any
2761 data type even at an odd addresses. For these machines, @code{__alignof__}
2762 reports the @emph{recommended} alignment of a type.
2764 If the operand of @code{__alignof__} is an lvalue rather than a type,
2765 its value is the required alignment for its type, taking into account
2766 any minimum alignment specified with GCC's @code{__attribute__}
2767 extension (@pxref{Variable Attributes}). For example, after this
2771 struct foo @{ int x; char y; @} foo1;
2775 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2776 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2778 It is an error to ask for the alignment of an incomplete type.
2780 @node Variable Attributes
2781 @section Specifying Attributes of Variables
2782 @cindex attribute of variables
2783 @cindex variable attributes
2785 The keyword @code{__attribute__} allows you to specify special
2786 attributes of variables or structure fields. This keyword is followed
2787 by an attribute specification inside double parentheses. Ten
2788 attributes are currently defined for variables: @code{aligned},
2789 @code{mode}, @code{nocommon}, @code{packed}, @code{section},
2790 @code{transparent_union}, @code{unused}, @code{deprecated},
2791 @code{vector_size}, and @code{weak}. Some other attributes are defined
2792 for variables on particular target systems. Other attributes are
2793 available for functions (@pxref{Function Attributes}) and for types
2794 (@pxref{Type Attributes}). Other front ends might define more
2795 attributes (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2797 You may also specify attributes with @samp{__} preceding and following
2798 each keyword. This allows you to use them in header files without
2799 being concerned about a possible macro of the same name. For example,
2800 you may use @code{__aligned__} instead of @code{aligned}.
2802 @xref{Attribute Syntax}, for details of the exact syntax for using
2806 @cindex @code{aligned} attribute
2807 @item aligned (@var{alignment})
2808 This attribute specifies a minimum alignment for the variable or
2809 structure field, measured in bytes. For example, the declaration:
2812 int x __attribute__ ((aligned (16))) = 0;
2816 causes the compiler to allocate the global variable @code{x} on a
2817 16-byte boundary. On a 68040, this could be used in conjunction with
2818 an @code{asm} expression to access the @code{move16} instruction which
2819 requires 16-byte aligned operands.
2821 You can also specify the alignment of structure fields. For example, to
2822 create a double-word aligned @code{int} pair, you could write:
2825 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2829 This is an alternative to creating a union with a @code{double} member
2830 that forces the union to be double-word aligned.
2832 As in the preceding examples, you can explicitly specify the alignment
2833 (in bytes) that you wish the compiler to use for a given variable or
2834 structure field. Alternatively, you can leave out the alignment factor
2835 and just ask the compiler to align a variable or field to the maximum
2836 useful alignment for the target machine you are compiling for. For
2837 example, you could write:
2840 short array[3] __attribute__ ((aligned));
2843 Whenever you leave out the alignment factor in an @code{aligned} attribute
2844 specification, the compiler automatically sets the alignment for the declared
2845 variable or field to the largest alignment which is ever used for any data
2846 type on the target machine you are compiling for. Doing this can often make
2847 copy operations more efficient, because the compiler can use whatever
2848 instructions copy the biggest chunks of memory when performing copies to
2849 or from the variables or fields that you have aligned this way.
2851 The @code{aligned} attribute can only increase the alignment; but you
2852 can decrease it by specifying @code{packed} as well. See below.
2854 Note that the effectiveness of @code{aligned} attributes may be limited
2855 by inherent limitations in your linker. On many systems, the linker is
2856 only able to arrange for variables to be aligned up to a certain maximum
2857 alignment. (For some linkers, the maximum supported alignment may
2858 be very very small.) If your linker is only able to align variables
2859 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
2860 in an @code{__attribute__} will still only provide you with 8 byte
2861 alignment. See your linker documentation for further information.
2863 @item mode (@var{mode})
2864 @cindex @code{mode} attribute
2865 This attribute specifies the data type for the declaration---whichever
2866 type corresponds to the mode @var{mode}. This in effect lets you
2867 request an integer or floating point type according to its width.
2869 You may also specify a mode of @samp{byte} or @samp{__byte__} to
2870 indicate the mode corresponding to a one-byte integer, @samp{word} or
2871 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
2872 or @samp{__pointer__} for the mode used to represent pointers.
2875 @cindex @code{nocommon} attribute
2877 This attribute specifies requests GCC not to place a variable
2878 ``common'' but instead to allocate space for it directly. If you
2879 specify the @option{-fno-common} flag, GCC will do this for all
2882 Specifying the @code{nocommon} attribute for a variable provides an
2883 initialization of zeros. A variable may only be initialized in one
2887 @cindex @code{packed} attribute
2888 The @code{packed} attribute specifies that a variable or structure field
2889 should have the smallest possible alignment---one byte for a variable,
2890 and one bit for a field, unless you specify a larger value with the
2891 @code{aligned} attribute.
2893 Here is a structure in which the field @code{x} is packed, so that it
2894 immediately follows @code{a}:
2900 int x[2] __attribute__ ((packed));
2904 @item section ("@var{section-name}")
2905 @cindex @code{section} variable attribute
2906 Normally, the compiler places the objects it generates in sections like
2907 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
2908 or you need certain particular variables to appear in special sections,
2909 for example to map to special hardware. The @code{section}
2910 attribute specifies that a variable (or function) lives in a particular
2911 section. For example, this small program uses several specific section names:
2914 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
2915 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
2916 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
2917 int init_data __attribute__ ((section ("INITDATA"))) = 0;
2921 /* Initialize stack pointer */
2922 init_sp (stack + sizeof (stack));
2924 /* Initialize initialized data */
2925 memcpy (&init_data, &data, &edata - &data);
2927 /* Turn on the serial ports */
2934 Use the @code{section} attribute with an @emph{initialized} definition
2935 of a @emph{global} variable, as shown in the example. GCC issues
2936 a warning and otherwise ignores the @code{section} attribute in
2937 uninitialized variable declarations.
2939 You may only use the @code{section} attribute with a fully initialized
2940 global definition because of the way linkers work. The linker requires
2941 each object be defined once, with the exception that uninitialized
2942 variables tentatively go in the @code{common} (or @code{bss}) section
2943 and can be multiply ``defined''. You can force a variable to be
2944 initialized with the @option{-fno-common} flag or the @code{nocommon}
2947 Some file formats do not support arbitrary sections so the @code{section}
2948 attribute is not available on all platforms.
2949 If you need to map the entire contents of a module to a particular
2950 section, consider using the facilities of the linker instead.
2953 @cindex @code{shared} variable attribute
2954 On Windows NT, in addition to putting variable definitions in a named
2955 section, the section can also be shared among all running copies of an
2956 executable or DLL@. For example, this small program defines shared data
2957 by putting it in a named section @code{shared} and marking the section
2961 int foo __attribute__((section ("shared"), shared)) = 0;
2966 /* Read and write foo. All running
2967 copies see the same value. */
2973 You may only use the @code{shared} attribute along with @code{section}
2974 attribute with a fully initialized global definition because of the way
2975 linkers work. See @code{section} attribute for more information.
2977 The @code{shared} attribute is only available on Windows NT@.
2979 @item transparent_union
2980 This attribute, attached to a function parameter which is a union, means
2981 that the corresponding argument may have the type of any union member,
2982 but the argument is passed as if its type were that of the first union
2983 member. For more details see @xref{Type Attributes}. You can also use
2984 this attribute on a @code{typedef} for a union data type; then it
2985 applies to all function parameters with that type.
2988 This attribute, attached to a variable, means that the variable is meant
2989 to be possibly unused. GCC will not produce a warning for this
2993 The @code{deprecated} attribute results in a warning if the variable
2994 is used anywhere in the source file. This is useful when identifying
2995 variables that are expected to be removed in a future version of a
2996 program. The warning also includes the location of the declaration
2997 of the deprecated variable, to enable users to easily find further
2998 information about why the variable is deprecated, or what they should
2999 do instead. Note that the warnings only occurs for uses:
3002 extern int old_var __attribute__ ((deprecated));
3004 int new_fn () @{ return old_var; @}
3007 results in a warning on line 3 but not line 2.
3009 The @code{deprecated} attribute can also be used for functions and
3010 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3012 @item vector_size (@var{bytes})
3013 This attribute specifies the vector size for the variable, measured in
3014 bytes. For example, the declaration:
3017 int foo __attribute__ ((vector_size (16)));
3021 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3022 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3023 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3025 This attribute is only applicable to integral and float scalars,
3026 although arrays, pointers, and function return values are allowed in
3027 conjunction with this construct.
3029 Aggregates with this attribute are invalid, even if they are of the same
3030 size as a corresponding scalar. For example, the declaration:
3033 struct S @{ int a; @};
3034 struct S __attribute__ ((vector_size (16))) foo;
3038 is invalid even if the size of the structure is the same as the size of
3042 The @code{weak} attribute is described in @xref{Function Attributes}.
3044 @item model (@var{model-name})
3045 @cindex variable addressability on the M32R/D
3046 Use this attribute on the M32R/D to set the addressability of an object.
3047 The identifier @var{model-name} is one of @code{small}, @code{medium},
3048 or @code{large}, representing each of the code models.
3050 Small model objects live in the lower 16MB of memory (so that their
3051 addresses can be loaded with the @code{ld24} instruction).
3053 Medium and large model objects may live anywhere in the 32-bit address space
3054 (the compiler will generate @code{seth/add3} instructions to load their
3059 To specify multiple attributes, separate them by commas within the
3060 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3063 @node Type Attributes
3064 @section Specifying Attributes of Types
3065 @cindex attribute of types
3066 @cindex type attributes
3068 The keyword @code{__attribute__} allows you to specify special
3069 attributes of @code{struct} and @code{union} types when you define such
3070 types. This keyword is followed by an attribute specification inside
3071 double parentheses. Five attributes are currently defined for types:
3072 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3073 and @code{deprecated}. Other attributes are defined for functions
3074 (@pxref{Function Attributes}) and for variables (@pxref{Variable Attributes}).
3076 You may also specify any one of these attributes with @samp{__}
3077 preceding and following its keyword. This allows you to use these
3078 attributes in header files without being concerned about a possible
3079 macro of the same name. For example, you may use @code{__aligned__}
3080 instead of @code{aligned}.
3082 You may specify the @code{aligned} and @code{transparent_union}
3083 attributes either in a @code{typedef} declaration or just past the
3084 closing curly brace of a complete enum, struct or union type
3085 @emph{definition} and the @code{packed} attribute only past the closing
3086 brace of a definition.
3088 You may also specify attributes between the enum, struct or union
3089 tag and the name of the type rather than after the closing brace.
3091 @xref{Attribute Syntax}, for details of the exact syntax for using
3095 @cindex @code{aligned} attribute
3096 @item aligned (@var{alignment})
3097 This attribute specifies a minimum alignment (in bytes) for variables
3098 of the specified type. For example, the declarations:
3101 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3102 typedef int more_aligned_int __attribute__ ((aligned (8)));
3106 force the compiler to insure (as far as it can) that each variable whose
3107 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3108 aligned @emph{at least} on a 8-byte boundary. On a Sparc, having all
3109 variables of type @code{struct S} aligned to 8-byte boundaries allows
3110 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3111 store) instructions when copying one variable of type @code{struct S} to
3112 another, thus improving run-time efficiency.
3114 Note that the alignment of any given @code{struct} or @code{union} type
3115 is required by the ISO C standard to be at least a perfect multiple of
3116 the lowest common multiple of the alignments of all of the members of
3117 the @code{struct} or @code{union} in question. This means that you @emph{can}
3118 effectively adjust the alignment of a @code{struct} or @code{union}
3119 type by attaching an @code{aligned} attribute to any one of the members
3120 of such a type, but the notation illustrated in the example above is a
3121 more obvious, intuitive, and readable way to request the compiler to
3122 adjust the alignment of an entire @code{struct} or @code{union} type.
3124 As in the preceding example, you can explicitly specify the alignment
3125 (in bytes) that you wish the compiler to use for a given @code{struct}
3126 or @code{union} type. Alternatively, you can leave out the alignment factor
3127 and just ask the compiler to align a type to the maximum
3128 useful alignment for the target machine you are compiling for. For
3129 example, you could write:
3132 struct S @{ short f[3]; @} __attribute__ ((aligned));
3135 Whenever you leave out the alignment factor in an @code{aligned}
3136 attribute specification, the compiler automatically sets the alignment
3137 for the type to the largest alignment which is ever used for any data
3138 type on the target machine you are compiling for. Doing this can often
3139 make copy operations more efficient, because the compiler can use
3140 whatever instructions copy the biggest chunks of memory when performing
3141 copies to or from the variables which have types that you have aligned
3144 In the example above, if the size of each @code{short} is 2 bytes, then
3145 the size of the entire @code{struct S} type is 6 bytes. The smallest
3146 power of two which is greater than or equal to that is 8, so the
3147 compiler sets the alignment for the entire @code{struct S} type to 8
3150 Note that although you can ask the compiler to select a time-efficient
3151 alignment for a given type and then declare only individual stand-alone
3152 objects of that type, the compiler's ability to select a time-efficient
3153 alignment is primarily useful only when you plan to create arrays of
3154 variables having the relevant (efficiently aligned) type. If you
3155 declare or use arrays of variables of an efficiently-aligned type, then
3156 it is likely that your program will also be doing pointer arithmetic (or
3157 subscripting, which amounts to the same thing) on pointers to the
3158 relevant type, and the code that the compiler generates for these
3159 pointer arithmetic operations will often be more efficient for
3160 efficiently-aligned types than for other types.
3162 The @code{aligned} attribute can only increase the alignment; but you
3163 can decrease it by specifying @code{packed} as well. See below.
3165 Note that the effectiveness of @code{aligned} attributes may be limited
3166 by inherent limitations in your linker. On many systems, the linker is
3167 only able to arrange for variables to be aligned up to a certain maximum
3168 alignment. (For some linkers, the maximum supported alignment may
3169 be very very small.) If your linker is only able to align variables
3170 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3171 in an @code{__attribute__} will still only provide you with 8 byte
3172 alignment. See your linker documentation for further information.
3175 This attribute, attached to an @code{enum}, @code{struct}, or
3176 @code{union} type definition, specified that the minimum required memory
3177 be used to represent the type.
3179 @opindex fshort-enums
3180 Specifying this attribute for @code{struct} and @code{union} types is
3181 equivalent to specifying the @code{packed} attribute on each of the
3182 structure or union members. Specifying the @option{-fshort-enums}
3183 flag on the line is equivalent to specifying the @code{packed}
3184 attribute on all @code{enum} definitions.
3186 You may only specify this attribute after a closing curly brace on an
3187 @code{enum} definition, not in a @code{typedef} declaration, unless that
3188 declaration also contains the definition of the @code{enum}.
3190 @item transparent_union
3191 This attribute, attached to a @code{union} type definition, indicates
3192 that any function parameter having that union type causes calls to that
3193 function to be treated in a special way.
3195 First, the argument corresponding to a transparent union type can be of
3196 any type in the union; no cast is required. Also, if the union contains
3197 a pointer type, the corresponding argument can be a null pointer
3198 constant or a void pointer expression; and if the union contains a void
3199 pointer type, the corresponding argument can be any pointer expression.
3200 If the union member type is a pointer, qualifiers like @code{const} on
3201 the referenced type must be respected, just as with normal pointer
3204 Second, the argument is passed to the function using the calling
3205 conventions of first member of the transparent union, not the calling
3206 conventions of the union itself. All members of the union must have the
3207 same machine representation; this is necessary for this argument passing
3210 Transparent unions are designed for library functions that have multiple
3211 interfaces for compatibility reasons. For example, suppose the
3212 @code{wait} function must accept either a value of type @code{int *} to
3213 comply with Posix, or a value of type @code{union wait *} to comply with
3214 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3215 @code{wait} would accept both kinds of arguments, but it would also
3216 accept any other pointer type and this would make argument type checking
3217 less useful. Instead, @code{<sys/wait.h>} might define the interface
3225 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3227 pid_t wait (wait_status_ptr_t);
3230 This interface allows either @code{int *} or @code{union wait *}
3231 arguments to be passed, using the @code{int *} calling convention.
3232 The program can call @code{wait} with arguments of either type:
3235 int w1 () @{ int w; return wait (&w); @}
3236 int w2 () @{ union wait w; return wait (&w); @}
3239 With this interface, @code{wait}'s implementation might look like this:
3242 pid_t wait (wait_status_ptr_t p)
3244 return waitpid (-1, p.__ip, 0);
3249 When attached to a type (including a @code{union} or a @code{struct}),
3250 this attribute means that variables of that type are meant to appear
3251 possibly unused. GCC will not produce a warning for any variables of
3252 that type, even if the variable appears to do nothing. This is often
3253 the case with lock or thread classes, which are usually defined and then
3254 not referenced, but contain constructors and destructors that have
3255 nontrivial bookkeeping functions.
3258 The @code{deprecated} attribute results in a warning if the type
3259 is used anywhere in the source file. This is useful when identifying
3260 types that are expected to be removed in a future version of a program.
3261 If possible, the warning also includes the location of the declaration
3262 of the deprecated type, to enable users to easily find further
3263 information about why the type is deprecated, or what they should do
3264 instead. Note that the warnings only occur for uses and then only
3265 if the type is being applied to an identifier that itself is not being
3266 declared as deprecated.
3269 typedef int T1 __attribute__ ((deprecated));
3273 typedef T1 T3 __attribute__ ((deprecated));
3274 T3 z __attribute__ ((deprecated));
3277 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3278 warning is issued for line 4 because T2 is not explicitly
3279 deprecated. Line 5 has no warning because T3 is explicitly
3280 deprecated. Similarly for line 6.
3282 The @code{deprecated} attribute can also be used for functions and
3283 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3287 To specify multiple attributes, separate them by commas within the
3288 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3292 @section An Inline Function is As Fast As a Macro
3293 @cindex inline functions
3294 @cindex integrating function code
3296 @cindex macros, inline alternative
3298 By declaring a function @code{inline}, you can direct GCC to
3299 integrate that function's code into the code for its callers. This
3300 makes execution faster by eliminating the function-call overhead; in
3301 addition, if any of the actual argument values are constant, their known
3302 values may permit simplifications at compile time so that not all of the
3303 inline function's code needs to be included. The effect on code size is
3304 less predictable; object code may be larger or smaller with function
3305 inlining, depending on the particular case. Inlining of functions is an
3306 optimization and it really ``works'' only in optimizing compilation. If
3307 you don't use @option{-O}, no function is really inline.
3309 Inline functions are included in the ISO C99 standard, but there are
3310 currently substantial differences between what GCC implements and what
3311 the ISO C99 standard requires.
3313 To declare a function inline, use the @code{inline} keyword in its
3314 declaration, like this:
3324 (If you are writing a header file to be included in ISO C programs, write
3325 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3326 You can also make all ``simple enough'' functions inline with the option
3327 @option{-finline-functions}.
3330 Note that certain usages in a function definition can make it unsuitable
3331 for inline substitution. Among these usages are: use of varargs, use of
3332 alloca, use of variable sized data types (@pxref{Variable Length}),
3333 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3334 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3335 will warn when a function marked @code{inline} could not be substituted,
3336 and will give the reason for the failure.
3338 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3339 does not affect the linkage of the function.
3341 @cindex automatic @code{inline} for C++ member fns
3342 @cindex @code{inline} automatic for C++ member fns
3343 @cindex member fns, automatically @code{inline}
3344 @cindex C++ member fns, automatically @code{inline}
3345 @opindex fno-default-inline
3346 GCC automatically inlines member functions defined within the class
3347 body of C++ programs even if they are not explicitly declared
3348 @code{inline}. (You can override this with @option{-fno-default-inline};
3349 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3351 @cindex inline functions, omission of
3352 @opindex fkeep-inline-functions
3353 When a function is both inline and @code{static}, if all calls to the
3354 function are integrated into the caller, and the function's address is
3355 never used, then the function's own assembler code is never referenced.
3356 In this case, GCC does not actually output assembler code for the
3357 function, unless you specify the option @option{-fkeep-inline-functions}.
3358 Some calls cannot be integrated for various reasons (in particular,
3359 calls that precede the function's definition cannot be integrated, and
3360 neither can recursive calls within the definition). If there is a
3361 nonintegrated call, then the function is compiled to assembler code as
3362 usual. The function must also be compiled as usual if the program
3363 refers to its address, because that can't be inlined.
3365 @cindex non-static inline function
3366 When an inline function is not @code{static}, then the compiler must assume
3367 that there may be calls from other source files; since a global symbol can
3368 be defined only once in any program, the function must not be defined in
3369 the other source files, so the calls therein cannot be integrated.
3370 Therefore, a non-@code{static} inline function is always compiled on its
3371 own in the usual fashion.
3373 If you specify both @code{inline} and @code{extern} in the function
3374 definition, then the definition is used only for inlining. In no case
3375 is the function compiled on its own, not even if you refer to its
3376 address explicitly. Such an address becomes an external reference, as
3377 if you had only declared the function, and had not defined it.
3379 This combination of @code{inline} and @code{extern} has almost the
3380 effect of a macro. The way to use it is to put a function definition in
3381 a header file with these keywords, and put another copy of the
3382 definition (lacking @code{inline} and @code{extern}) in a library file.
3383 The definition in the header file will cause most calls to the function
3384 to be inlined. If any uses of the function remain, they will refer to
3385 the single copy in the library.
3387 For future compatibility with when GCC implements ISO C99 semantics for
3388 inline functions, it is best to use @code{static inline} only. (The
3389 existing semantics will remain available when @option{-std=gnu89} is
3390 specified, but eventually the default will be @option{-std=gnu99} and
3391 that will implement the C99 semantics, though it does not do so yet.)
3393 GCC does not inline any functions when not optimizing unless you specify
3394 the @samp{always_inline} attribute for the function, like this:
3398 inline void foo (const char) __attribute__((always_inline));
3402 @section Assembler Instructions with C Expression Operands
3403 @cindex extended @code{asm}
3404 @cindex @code{asm} expressions
3405 @cindex assembler instructions
3408 In an assembler instruction using @code{asm}, you can specify the
3409 operands of the instruction using C expressions. This means you need not
3410 guess which registers or memory locations will contain the data you want
3413 You must specify an assembler instruction template much like what
3414 appears in a machine description, plus an operand constraint string for
3417 For example, here is how to use the 68881's @code{fsinx} instruction:
3420 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3424 Here @code{angle} is the C expression for the input operand while
3425 @code{result} is that of the output operand. Each has @samp{"f"} as its
3426 operand constraint, saying that a floating point register is required.
3427 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3428 output operands' constraints must use @samp{=}. The constraints use the
3429 same language used in the machine description (@pxref{Constraints}).
3431 Each operand is described by an operand-constraint string followed by
3432 the C expression in parentheses. A colon separates the assembler
3433 template from the first output operand and another separates the last
3434 output operand from the first input, if any. Commas separate the
3435 operands within each group. The total number of operands is currently
3436 limited to 30; this limitation may be lifted in some future version of
3439 If there are no output operands but there are input operands, you must
3440 place two consecutive colons surrounding the place where the output
3443 As of GCC version 3.1, it is also possible to specify input and output
3444 operands using symbolic names which can be referenced within the
3445 assembler code. These names are specified inside square brackets
3446 preceding the constraint string, and can be referenced inside the
3447 assembler code using @code{%[@var{name}]} instead of a percentage sign
3448 followed by the operand number. Using named operands the above example
3452 asm ("fsinx %[angle],%[output]"
3453 : [output] "=f" (result)
3454 : [angle] "f" (angle));
3458 Note that the symbolic operand names have no relation whatsoever to
3459 other C identifiers. You may use any name you like, even those of
3460 existing C symbols, but must ensure that no two operands within the same
3461 assembler construct use the same symbolic name.
3463 Output operand expressions must be lvalues; the compiler can check this.
3464 The input operands need not be lvalues. The compiler cannot check
3465 whether the operands have data types that are reasonable for the
3466 instruction being executed. It does not parse the assembler instruction
3467 template and does not know what it means or even whether it is valid
3468 assembler input. The extended @code{asm} feature is most often used for
3469 machine instructions the compiler itself does not know exist. If
3470 the output expression cannot be directly addressed (for example, it is a
3471 bit-field), your constraint must allow a register. In that case, GCC
3472 will use the register as the output of the @code{asm}, and then store
3473 that register into the output.
3475 The ordinary output operands must be write-only; GCC will assume that
3476 the values in these operands before the instruction are dead and need
3477 not be generated. Extended asm supports input-output or read-write
3478 operands. Use the constraint character @samp{+} to indicate such an
3479 operand and list it with the output operands.
3481 When the constraints for the read-write operand (or the operand in which
3482 only some of the bits are to be changed) allows a register, you may, as
3483 an alternative, logically split its function into two separate operands,
3484 one input operand and one write-only output operand. The connection
3485 between them is expressed by constraints which say they need to be in
3486 the same location when the instruction executes. You can use the same C
3487 expression for both operands, or different expressions. For example,
3488 here we write the (fictitious) @samp{combine} instruction with
3489 @code{bar} as its read-only source operand and @code{foo} as its
3490 read-write destination:
3493 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3497 The constraint @samp{"0"} for operand 1 says that it must occupy the
3498 same location as operand 0. A number in constraint is allowed only in
3499 an input operand and it must refer to an output operand.
3501 Only a number in the constraint can guarantee that one operand will be in
3502 the same place as another. The mere fact that @code{foo} is the value
3503 of both operands is not enough to guarantee that they will be in the
3504 same place in the generated assembler code. The following would not
3508 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3511 Various optimizations or reloading could cause operands 0 and 1 to be in
3512 different registers; GCC knows no reason not to do so. For example, the
3513 compiler might find a copy of the value of @code{foo} in one register and
3514 use it for operand 1, but generate the output operand 0 in a different
3515 register (copying it afterward to @code{foo}'s own address). Of course,
3516 since the register for operand 1 is not even mentioned in the assembler
3517 code, the result will not work, but GCC can't tell that.
3519 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3520 the operand number for a matching constraint. For example:
3523 asm ("cmoveq %1,%2,%[result]"
3524 : [result] "=r"(result)
3525 : "r" (test), "r"(new), "[result]"(old));
3528 Some instructions clobber specific hard registers. To describe this,
3529 write a third colon after the input operands, followed by the names of
3530 the clobbered hard registers (given as strings). Here is a realistic
3531 example for the VAX:
3534 asm volatile ("movc3 %0,%1,%2"
3536 : "g" (from), "g" (to), "g" (count)
3537 : "r0", "r1", "r2", "r3", "r4", "r5");
3540 You may not write a clobber description in a way that overlaps with an
3541 input or output operand. For example, you may not have an operand
3542 describing a register class with one member if you mention that register
3543 in the clobber list. There is no way for you to specify that an input
3544 operand is modified without also specifying it as an output
3545 operand. Note that if all the output operands you specify are for this
3546 purpose (and hence unused), you will then also need to specify
3547 @code{volatile} for the @code{asm} construct, as described below, to
3548 prevent GCC from deleting the @code{asm} statement as unused.
3550 If you refer to a particular hardware register from the assembler code,
3551 you will probably have to list the register after the third colon to
3552 tell the compiler the register's value is modified. In some assemblers,
3553 the register names begin with @samp{%}; to produce one @samp{%} in the
3554 assembler code, you must write @samp{%%} in the input.
3556 If your assembler instruction can alter the condition code register, add
3557 @samp{cc} to the list of clobbered registers. GCC on some machines
3558 represents the condition codes as a specific hardware register;
3559 @samp{cc} serves to name this register. On other machines, the
3560 condition code is handled differently, and specifying @samp{cc} has no
3561 effect. But it is valid no matter what the machine.
3563 If your assembler instruction modifies memory in an unpredictable
3564 fashion, add @samp{memory} to the list of clobbered registers. This
3565 will cause GCC to not keep memory values cached in registers across
3566 the assembler instruction. You will also want to add the
3567 @code{volatile} keyword if the memory affected is not listed in the
3568 inputs or outputs of the @code{asm}, as the @samp{memory} clobber does
3569 not count as a side-effect of the @code{asm}.
3571 You can put multiple assembler instructions together in a single
3572 @code{asm} template, separated by the characters normally used in assembly
3573 code for the system. A combination that works in most places is a newline
3574 to break the line, plus a tab character to move to the instruction field
3575 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
3576 assembler allows semicolons as a line-breaking character. Note that some
3577 assembler dialects use semicolons to start a comment.
3578 The input operands are guaranteed not to use any of the clobbered
3579 registers, and neither will the output operands' addresses, so you can
3580 read and write the clobbered registers as many times as you like. Here
3581 is an example of multiple instructions in a template; it assumes the
3582 subroutine @code{_foo} accepts arguments in registers 9 and 10:
3585 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
3587 : "g" (from), "g" (to)
3591 Unless an output operand has the @samp{&} constraint modifier, GCC
3592 may allocate it in the same register as an unrelated input operand, on
3593 the assumption the inputs are consumed before the outputs are produced.
3594 This assumption may be false if the assembler code actually consists of
3595 more than one instruction. In such a case, use @samp{&} for each output
3596 operand that may not overlap an input. @xref{Modifiers}.
3598 If you want to test the condition code produced by an assembler
3599 instruction, you must include a branch and a label in the @code{asm}
3600 construct, as follows:
3603 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
3609 This assumes your assembler supports local labels, as the GNU assembler
3610 and most Unix assemblers do.
3612 Speaking of labels, jumps from one @code{asm} to another are not
3613 supported. The compiler's optimizers do not know about these jumps, and
3614 therefore they cannot take account of them when deciding how to
3617 @cindex macros containing @code{asm}
3618 Usually the most convenient way to use these @code{asm} instructions is to
3619 encapsulate them in macros that look like functions. For example,
3623 (@{ double __value, __arg = (x); \
3624 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
3629 Here the variable @code{__arg} is used to make sure that the instruction
3630 operates on a proper @code{double} value, and to accept only those
3631 arguments @code{x} which can convert automatically to a @code{double}.
3633 Another way to make sure the instruction operates on the correct data
3634 type is to use a cast in the @code{asm}. This is different from using a
3635 variable @code{__arg} in that it converts more different types. For
3636 example, if the desired type were @code{int}, casting the argument to
3637 @code{int} would accept a pointer with no complaint, while assigning the
3638 argument to an @code{int} variable named @code{__arg} would warn about
3639 using a pointer unless the caller explicitly casts it.
3641 If an @code{asm} has output operands, GCC assumes for optimization
3642 purposes the instruction has no side effects except to change the output
3643 operands. This does not mean instructions with a side effect cannot be
3644 used, but you must be careful, because the compiler may eliminate them
3645 if the output operands aren't used, or move them out of loops, or
3646 replace two with one if they constitute a common subexpression. Also,
3647 if your instruction does have a side effect on a variable that otherwise
3648 appears not to change, the old value of the variable may be reused later
3649 if it happens to be found in a register.
3651 You can prevent an @code{asm} instruction from being deleted, moved
3652 significantly, or combined, by writing the keyword @code{volatile} after
3653 the @code{asm}. For example:
3656 #define get_and_set_priority(new) \
3658 asm volatile ("get_and_set_priority %0, %1" \
3659 : "=g" (__old) : "g" (new)); \
3664 If you write an @code{asm} instruction with no outputs, GCC will know
3665 the instruction has side-effects and will not delete the instruction or
3666 move it outside of loops.
3668 The @code{volatile} keyword indicates that the instruction has
3669 important side-effects. GCC will not delete a volatile @code{asm} if
3670 it is reachable. (The instruction can still be deleted if GCC can
3671 prove that control-flow will never reach the location of the
3672 instruction.) In addition, GCC will not reschedule instructions
3673 across a volatile @code{asm} instruction. For example:
3676 *(volatile int *)addr = foo;
3677 asm volatile ("eieio" : : );
3681 Assume @code{addr} contains the address of a memory mapped device
3682 register. The PowerPC @code{eieio} instruction (Enforce In-order
3683 Execution of I/O) tells the CPU to make sure that the store to that
3684 device register happens before it issues any other I/O@.
3686 Note that even a volatile @code{asm} instruction can be moved in ways
3687 that appear insignificant to the compiler, such as across jump
3688 instructions. You can't expect a sequence of volatile @code{asm}
3689 instructions to remain perfectly consecutive. If you want consecutive
3690 output, use a single @code{asm}. Also, GCC will perform some
3691 optimizations across a volatile @code{asm} instruction; GCC does not
3692 ``forget everything'' when it encounters a volatile @code{asm}
3693 instruction the way some other compilers do.
3695 An @code{asm} instruction without any operands or clobbers (an ``old
3696 style'' @code{asm}) will be treated identically to a volatile
3697 @code{asm} instruction.
3699 It is a natural idea to look for a way to give access to the condition
3700 code left by the assembler instruction. However, when we attempted to
3701 implement this, we found no way to make it work reliably. The problem
3702 is that output operands might need reloading, which would result in
3703 additional following ``store'' instructions. On most machines, these
3704 instructions would alter the condition code before there was time to
3705 test it. This problem doesn't arise for ordinary ``test'' and
3706 ``compare'' instructions because they don't have any output operands.
3708 For reasons similar to those described above, it is not possible to give
3709 an assembler instruction access to the condition code left by previous
3712 If you are writing a header file that should be includable in ISO C
3713 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
3716 @subsection i386 floating point asm operands
3718 There are several rules on the usage of stack-like regs in
3719 asm_operands insns. These rules apply only to the operands that are
3724 Given a set of input regs that die in an asm_operands, it is
3725 necessary to know which are implicitly popped by the asm, and
3726 which must be explicitly popped by gcc.
3728 An input reg that is implicitly popped by the asm must be
3729 explicitly clobbered, unless it is constrained to match an
3733 For any input reg that is implicitly popped by an asm, it is
3734 necessary to know how to adjust the stack to compensate for the pop.
3735 If any non-popped input is closer to the top of the reg-stack than
3736 the implicitly popped reg, it would not be possible to know what the
3737 stack looked like---it's not clear how the rest of the stack ``slides
3740 All implicitly popped input regs must be closer to the top of
3741 the reg-stack than any input that is not implicitly popped.
3743 It is possible that if an input dies in an insn, reload might
3744 use the input reg for an output reload. Consider this example:
3747 asm ("foo" : "=t" (a) : "f" (b));
3750 This asm says that input B is not popped by the asm, and that
3751 the asm pushes a result onto the reg-stack, i.e., the stack is one
3752 deeper after the asm than it was before. But, it is possible that
3753 reload will think that it can use the same reg for both the input and
3754 the output, if input B dies in this insn.
3756 If any input operand uses the @code{f} constraint, all output reg
3757 constraints must use the @code{&} earlyclobber.
3759 The asm above would be written as
3762 asm ("foo" : "=&t" (a) : "f" (b));
3766 Some operands need to be in particular places on the stack. All
3767 output operands fall in this category---there is no other way to
3768 know which regs the outputs appear in unless the user indicates
3769 this in the constraints.
3771 Output operands must specifically indicate which reg an output
3772 appears in after an asm. @code{=f} is not allowed: the operand
3773 constraints must select a class with a single reg.
3776 Output operands may not be ``inserted'' between existing stack regs.
3777 Since no 387 opcode uses a read/write operand, all output operands
3778 are dead before the asm_operands, and are pushed by the asm_operands.
3779 It makes no sense to push anywhere but the top of the reg-stack.
3781 Output operands must start at the top of the reg-stack: output
3782 operands may not ``skip'' a reg.
3785 Some asm statements may need extra stack space for internal
3786 calculations. This can be guaranteed by clobbering stack registers
3787 unrelated to the inputs and outputs.
3791 Here are a couple of reasonable asms to want to write. This asm
3792 takes one input, which is internally popped, and produces two outputs.
3795 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
3798 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
3799 and replaces them with one output. The user must code the @code{st(1)}
3800 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
3803 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
3809 @section Controlling Names Used in Assembler Code
3810 @cindex assembler names for identifiers
3811 @cindex names used in assembler code
3812 @cindex identifiers, names in assembler code
3814 You can specify the name to be used in the assembler code for a C
3815 function or variable by writing the @code{asm} (or @code{__asm__})
3816 keyword after the declarator as follows:
3819 int foo asm ("myfoo") = 2;
3823 This specifies that the name to be used for the variable @code{foo} in
3824 the assembler code should be @samp{myfoo} rather than the usual
3827 On systems where an underscore is normally prepended to the name of a C
3828 function or variable, this feature allows you to define names for the
3829 linker that do not start with an underscore.
3831 It does not make sense to use this feature with a non-static local
3832 variable since such variables do not have assembler names. If you are
3833 trying to put the variable in a particular register, see @ref{Explicit
3834 Reg Vars}. GCC presently accepts such code with a warning, but will
3835 probably be changed to issue an error, rather than a warning, in the
3838 You cannot use @code{asm} in this way in a function @emph{definition}; but
3839 you can get the same effect by writing a declaration for the function
3840 before its definition and putting @code{asm} there, like this:
3843 extern func () asm ("FUNC");
3850 It is up to you to make sure that the assembler names you choose do not
3851 conflict with any other assembler symbols. Also, you must not use a
3852 register name; that would produce completely invalid assembler code. GCC
3853 does not as yet have the ability to store static variables in registers.
3854 Perhaps that will be added.
3856 @node Explicit Reg Vars
3857 @section Variables in Specified Registers
3858 @cindex explicit register variables
3859 @cindex variables in specified registers
3860 @cindex specified registers
3861 @cindex registers, global allocation
3863 GNU C allows you to put a few global variables into specified hardware
3864 registers. You can also specify the register in which an ordinary
3865 register variable should be allocated.
3869 Global register variables reserve registers throughout the program.
3870 This may be useful in programs such as programming language
3871 interpreters which have a couple of global variables that are accessed
3875 Local register variables in specific registers do not reserve the
3876 registers. The compiler's data flow analysis is capable of determining
3877 where the specified registers contain live values, and where they are
3878 available for other uses. Stores into local register variables may be deleted
3879 when they appear to be dead according to dataflow analysis. References
3880 to local register variables may be deleted or moved or simplified.
3882 These local variables are sometimes convenient for use with the extended
3883 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
3884 output of the assembler instruction directly into a particular register.
3885 (This will work provided the register you specify fits the constraints
3886 specified for that operand in the @code{asm}.)
3894 @node Global Reg Vars
3895 @subsection Defining Global Register Variables
3896 @cindex global register variables
3897 @cindex registers, global variables in
3899 You can define a global register variable in GNU C like this:
3902 register int *foo asm ("a5");
3906 Here @code{a5} is the name of the register which should be used. Choose a
3907 register which is normally saved and restored by function calls on your
3908 machine, so that library routines will not clobber it.
3910 Naturally the register name is cpu-dependent, so you would need to
3911 conditionalize your program according to cpu type. The register
3912 @code{a5} would be a good choice on a 68000 for a variable of pointer
3913 type. On machines with register windows, be sure to choose a ``global''
3914 register that is not affected magically by the function call mechanism.
3916 In addition, operating systems on one type of cpu may differ in how they
3917 name the registers; then you would need additional conditionals. For
3918 example, some 68000 operating systems call this register @code{%a5}.
3920 Eventually there may be a way of asking the compiler to choose a register
3921 automatically, but first we need to figure out how it should choose and
3922 how to enable you to guide the choice. No solution is evident.
3924 Defining a global register variable in a certain register reserves that
3925 register entirely for this use, at least within the current compilation.
3926 The register will not be allocated for any other purpose in the functions
3927 in the current compilation. The register will not be saved and restored by
3928 these functions. Stores into this register are never deleted even if they
3929 would appear to be dead, but references may be deleted or moved or
3932 It is not safe to access the global register variables from signal
3933 handlers, or from more than one thread of control, because the system
3934 library routines may temporarily use the register for other things (unless
3935 you recompile them specially for the task at hand).
3937 @cindex @code{qsort}, and global register variables
3938 It is not safe for one function that uses a global register variable to
3939 call another such function @code{foo} by way of a third function
3940 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
3941 different source file in which the variable wasn't declared). This is
3942 because @code{lose} might save the register and put some other value there.
3943 For example, you can't expect a global register variable to be available in
3944 the comparison-function that you pass to @code{qsort}, since @code{qsort}
3945 might have put something else in that register. (If you are prepared to
3946 recompile @code{qsort} with the same global register variable, you can
3947 solve this problem.)
3949 If you want to recompile @code{qsort} or other source files which do not
3950 actually use your global register variable, so that they will not use that
3951 register for any other purpose, then it suffices to specify the compiler
3952 option @option{-ffixed-@var{reg}}. You need not actually add a global
3953 register declaration to their source code.
3955 A function which can alter the value of a global register variable cannot
3956 safely be called from a function compiled without this variable, because it
3957 could clobber the value the caller expects to find there on return.
3958 Therefore, the function which is the entry point into the part of the
3959 program that uses the global register variable must explicitly save and
3960 restore the value which belongs to its caller.
3962 @cindex register variable after @code{longjmp}
3963 @cindex global register after @code{longjmp}
3964 @cindex value after @code{longjmp}
3967 On most machines, @code{longjmp} will restore to each global register
3968 variable the value it had at the time of the @code{setjmp}. On some
3969 machines, however, @code{longjmp} will not change the value of global
3970 register variables. To be portable, the function that called @code{setjmp}
3971 should make other arrangements to save the values of the global register
3972 variables, and to restore them in a @code{longjmp}. This way, the same
3973 thing will happen regardless of what @code{longjmp} does.
3975 All global register variable declarations must precede all function
3976 definitions. If such a declaration could appear after function
3977 definitions, the declaration would be too late to prevent the register from
3978 being used for other purposes in the preceding functions.
3980 Global register variables may not have initial values, because an
3981 executable file has no means to supply initial contents for a register.
3983 On the Sparc, there are reports that g3 @dots{} g7 are suitable
3984 registers, but certain library functions, such as @code{getwd}, as well
3985 as the subroutines for division and remainder, modify g3 and g4. g1 and
3986 g2 are local temporaries.
3988 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
3989 Of course, it will not do to use more than a few of those.
3991 @node Local Reg Vars
3992 @subsection Specifying Registers for Local Variables
3993 @cindex local variables, specifying registers
3994 @cindex specifying registers for local variables
3995 @cindex registers for local variables
3997 You can define a local register variable with a specified register
4001 register int *foo asm ("a5");
4005 Here @code{a5} is the name of the register which should be used. Note
4006 that this is the same syntax used for defining global register
4007 variables, but for a local variable it would appear within a function.
4009 Naturally the register name is cpu-dependent, but this is not a
4010 problem, since specific registers are most often useful with explicit
4011 assembler instructions (@pxref{Extended Asm}). Both of these things
4012 generally require that you conditionalize your program according to
4015 In addition, operating systems on one type of cpu may differ in how they
4016 name the registers; then you would need additional conditionals. For
4017 example, some 68000 operating systems call this register @code{%a5}.
4019 Defining such a register variable does not reserve the register; it
4020 remains available for other uses in places where flow control determines
4021 the variable's value is not live. However, these registers are made
4022 unavailable for use in the reload pass; excessive use of this feature
4023 leaves the compiler too few available registers to compile certain
4026 This option does not guarantee that GCC will generate code that has
4027 this variable in the register you specify at all times. You may not
4028 code an explicit reference to this register in an @code{asm} statement
4029 and assume it will always refer to this variable.
4031 Stores into local register variables may be deleted when they appear to be dead
4032 according to dataflow analysis. References to local register variables may
4033 be deleted or moved or simplified.
4035 @node Alternate Keywords
4036 @section Alternate Keywords
4037 @cindex alternate keywords
4038 @cindex keywords, alternate
4040 The option @option{-traditional} disables certain keywords;
4041 @option{-ansi} and the various @option{-std} options disable certain
4042 others. This causes trouble when you want to use GNU C extensions, or
4043 ISO C features, in a general-purpose header file that should be usable
4044 by all programs, including ISO C programs and traditional ones. The
4045 keywords @code{asm}, @code{typeof} and @code{inline} cannot be used
4046 since they won't work in a program compiled with @option{-ansi}
4047 (although @code{inline} can be used in a program compiled with
4048 @option{-std=c99}), while the keywords @code{const}, @code{volatile},
4049 @code{signed}, @code{typeof} and @code{inline} won't work in a program
4050 compiled with @option{-traditional}. The ISO C99 keyword
4051 @code{restrict} is only available when @option{-std=gnu99} (which will
4052 eventually be the default) or @option{-std=c99} (or the equivalent
4053 @option{-std=iso9899:1999}) is used.
4055 The way to solve these problems is to put @samp{__} at the beginning and
4056 end of each problematical keyword. For example, use @code{__asm__}
4057 instead of @code{asm}, @code{__const__} instead of @code{const}, and
4058 @code{__inline__} instead of @code{inline}.
4060 Other C compilers won't accept these alternative keywords; if you want to
4061 compile with another compiler, you can define the alternate keywords as
4062 macros to replace them with the customary keywords. It looks like this:
4070 @findex __extension__
4072 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4074 prevent such warnings within one expression by writing
4075 @code{__extension__} before the expression. @code{__extension__} has no
4076 effect aside from this.
4078 @node Incomplete Enums
4079 @section Incomplete @code{enum} Types
4081 You can define an @code{enum} tag without specifying its possible values.
4082 This results in an incomplete type, much like what you get if you write
4083 @code{struct foo} without describing the elements. A later declaration
4084 which does specify the possible values completes the type.
4086 You can't allocate variables or storage using the type while it is
4087 incomplete. However, you can work with pointers to that type.
4089 This extension may not be very useful, but it makes the handling of
4090 @code{enum} more consistent with the way @code{struct} and @code{union}
4093 This extension is not supported by GNU C++.
4095 @node Function Names
4096 @section Function Names as Strings
4097 @cindex @code{__FUNCTION__} identifier
4098 @cindex @code{__PRETTY_FUNCTION__} identifier
4099 @cindex @code{__func__} identifier
4101 GCC predefines two magic identifiers to hold the name of the current
4102 function. The identifier @code{__FUNCTION__} holds the name of the function
4103 as it appears in the source. The identifier @code{__PRETTY_FUNCTION__}
4104 holds the name of the function pretty printed in a language specific
4107 These names are always the same in a C function, but in a C++ function
4108 they may be different. For example, this program:
4112 extern int printf (char *, ...);
4119 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4120 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4138 __PRETTY_FUNCTION__ = int a::sub (int)
4141 The compiler automagically replaces the identifiers with a string
4142 literal containing the appropriate name. Thus, they are neither
4143 preprocessor macros, like @code{__FILE__} and @code{__LINE__}, nor
4144 variables. This means that they catenate with other string literals, and
4145 that they can be used to initialize char arrays. For example
4148 char here[] = "Function " __FUNCTION__ " in " __FILE__;
4151 On the other hand, @samp{#ifdef __FUNCTION__} does not have any special
4152 meaning inside a function, since the preprocessor does not do anything
4153 special with the identifier @code{__FUNCTION__}.
4155 Note that these semantics are deprecated, and that GCC 3.2 will handle
4156 @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__} the same way as
4157 @code{__func__}. @code{__func__} is defined by the ISO standard C99:
4160 The identifier @code{__func__} is implicitly declared by the translator
4161 as if, immediately following the opening brace of each function
4162 definition, the declaration
4165 static const char __func__[] = "function-name";
4168 appeared, where function-name is the name of the lexically-enclosing
4169 function. This name is the unadorned name of the function.
4172 By this definition, @code{__func__} is a variable, not a string literal.
4173 In particular, @code{__func__} does not catenate with other string
4176 In @code{C++}, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__} are
4177 variables, declared in the same way as @code{__func__}.
4179 @node Return Address
4180 @section Getting the Return or Frame Address of a Function
4182 These functions may be used to get information about the callers of a
4185 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4186 This function returns the return address of the current function, or of
4187 one of its callers. The @var{level} argument is number of frames to
4188 scan up the call stack. A value of @code{0} yields the return address
4189 of the current function, a value of @code{1} yields the return address
4190 of the caller of the current function, and so forth.
4192 The @var{level} argument must be a constant integer.
4194 On some machines it may be impossible to determine the return address of
4195 any function other than the current one; in such cases, or when the top
4196 of the stack has been reached, this function will return @code{0} or a
4197 random value. In addition, @code{__builtin_frame_address} may be used
4198 to determine if the top of the stack has been reached.
4200 This function should only be used with a nonzero argument for debugging
4204 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4205 This function is similar to @code{__builtin_return_address}, but it
4206 returns the address of the function frame rather than the return address
4207 of the function. Calling @code{__builtin_frame_address} with a value of
4208 @code{0} yields the frame address of the current function, a value of
4209 @code{1} yields the frame address of the caller of the current function,
4212 The frame is the area on the stack which holds local variables and saved
4213 registers. The frame address is normally the address of the first word
4214 pushed on to the stack by the function. However, the exact definition
4215 depends upon the processor and the calling convention. If the processor
4216 has a dedicated frame pointer register, and the function has a frame,
4217 then @code{__builtin_frame_address} will return the value of the frame
4220 On some machines it may be impossible to determine the frame address of
4221 any function other than the current one; in such cases, or when the top
4222 of the stack has been reached, this function will return @code{0} if
4223 the first frame pointer is properly initialized by the startup code.
4225 This function should only be used with a nonzero argument for debugging
4229 @node Vector Extensions
4230 @section Using vector instructions through built-in functions
4232 On some targets, the instruction set contains SIMD vector instructions that
4233 operate on multiple values contained in one large register at the same time.
4234 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4237 The first step in using these extensions is to provide the necessary data
4238 types. This should be done using an appropriate @code{typedef}:
4241 typedef int v4si __attribute__ ((mode(V4SI)));
4244 The base type @code{int} is effectively ignored by the compiler, the
4245 actual properties of the new type @code{v4si} are defined by the
4246 @code{__attribute__}. It defines the machine mode to be used; for vector
4247 types these have the form @code{V@var{n}@var{B}}; @var{n} should be the
4248 number of elements in the vector, and @var{B} should be the base mode of the
4249 individual elements. The following can be used as base modes:
4253 An integer that is as wide as the smallest addressable unit, usually 8 bits.
4255 An integer, twice as wide as a QI mode integer, usually 16 bits.
4257 An integer, four times as wide as a QI mode integer, usually 32 bits.
4259 An integer, eight times as wide as a QI mode integer, usually 64 bits.
4261 A floating point value, as wide as a SI mode integer, usually 32 bits.
4263 A floating point value, as wide as a DI mode integer, usually 64 bits.
4266 Not all base types or combinations are always valid; which modes can be used
4267 is determined by the target machine. For example, if targetting the i386 MMX
4268 extensions, only @code{V8QI}, @code{V4HI} and @code{V2SI} are allowed modes.
4270 There are no @code{V1xx} vector modes - they would be identical to the
4271 corresponding base mode.
4273 There is no distinction between signed and unsigned vector modes. This
4274 distinction is made by the operations that perform on the vectors, not
4277 The types defined in this manner are somewhat special, they cannot be
4278 used with most normal C operations (i.e., a vector addition can @emph{not}
4279 be represented by a normal addition of two vector type variables). You
4280 can declare only variables and use them in function calls and returns, as
4281 well as in assignments and some casts. It is possible to cast from one
4282 vector type to another, provided they are of the same size (in fact, you
4283 can also cast vectors to and from other datatypes of the same size).
4285 A port that supports vector operations provides a set of built-in functions
4286 that can be used to operate on vectors. For example, a function to add two
4287 vectors and multiply the result by a third could look like this:
4290 v4si f (v4si a, v4si b, v4si c)
4292 v4si tmp = __builtin_addv4si (a, b);
4293 return __builtin_mulv4si (tmp, c);
4298 @node Other Builtins
4299 @section Other built-in functions provided by GCC
4300 @cindex built-in functions
4301 @findex __builtin_isgreater
4302 @findex __builtin_isgreaterequal
4303 @findex __builtin_isless
4304 @findex __builtin_islessequal
4305 @findex __builtin_islessgreater
4306 @findex __builtin_isunordered
4332 @findex fprintf_unlocked
4334 @findex fputs_unlocked
4343 @findex printf_unlocked
4365 GCC provides a large number of built-in functions other than the ones
4366 mentioned above. Some of these are for internal use in the processing
4367 of exceptions or variable-length argument lists and will not be
4368 documented here because they may change from time to time; we do not
4369 recommend general use of these functions.
4371 The remaining functions are provided for optimization purposes.
4373 @opindex fno-builtin
4374 GCC includes built-in versions of many of the functions in the standard
4375 C library. The versions prefixed with @code{__builtin_} will always be
4376 treated as having the same meaning as the C library function even if you
4377 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
4378 Many of these functions are only optimized in certain cases; if they are
4379 not optimized in a particular case, a call to the library function will
4384 The functions @code{abort}, @code{exit}, @code{_Exit} and @code{_exit}
4385 are recognized and presumed not to return, but otherwise are not built
4386 in. @code{_exit} is not recognized in strict ISO C mode (@option{-ansi},
4387 @option{-std=c89} or @option{-std=c99}). @code{_Exit} is not recognized in
4388 strict C89 mode (@option{-ansi} or @option{-std=c89}).
4390 Outside strict ISO C mode, the functions @code{alloca}, @code{bcmp},
4391 @code{bzero}, @code{index}, @code{rindex}, @code{ffs}, @code{fputs_unlocked},
4392 @code{printf_unlocked} and @code{fprintf_unlocked} may be handled as
4393 built-in functions. All these functions have corresponding versions
4394 prefixed with @code{__builtin_}, which may be used even in strict C89
4397 The ISO C99 functions @code{conj}, @code{conjf}, @code{conjl},
4398 @code{creal}, @code{crealf}, @code{creall}, @code{cimag}, @code{cimagf},
4399 @code{cimagl}, @code{llabs} and @code{imaxabs} are handled as built-in
4400 functions except in strict ISO C89 mode. There are also built-in
4401 versions of the ISO C99 functions @code{cosf}, @code{cosl},
4402 @code{fabsf}, @code{fabsl}, @code{sinf}, @code{sinl}, @code{sqrtf}, and
4403 @code{sqrtl}, that are recognized in any mode since ISO C89 reserves
4404 these names for the purpose to which ISO C99 puts them. All these
4405 functions have corresponding versions prefixed with @code{__builtin_}.
4407 The ISO C89 functions @code{abs}, @code{cos}, @code{fabs},
4408 @code{fprintf}, @code{fputs}, @code{labs}, @code{memcmp}, @code{memcpy},
4409 @code{memset}, @code{printf}, @code{sin}, @code{sqrt}, @code{strcat},
4410 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
4411 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
4412 @code{strpbrk}, @code{strrchr}, @code{strspn}, and @code{strstr} are all
4413 recognized as built-in functions unless @option{-fno-builtin} is
4414 specified (or @option{-fno-builtin-@var{function}} is specified for an
4415 individual function). All of these functions have corresponding
4416 versions prefixed with @code{__builtin_}.
4418 GCC provides built-in versions of the ISO C99 floating point comparison
4419 macros that avoid raising exceptions for unordered operands. They have
4420 the same names as the standard macros ( @code{isgreater},
4421 @code{isgreaterequal}, @code{isless}, @code{islessequal},
4422 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
4423 prefixed. We intend for a library implementor to be able to simply
4424 @code{#define} each standard macro to its built-in equivalent.
4426 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
4428 You can use the built-in function @code{__builtin_types_compatible_p} to
4429 determine whether two types are the same.
4431 This built-in function returns 1 if the unqualified versions of the
4432 types @var{type1} and @var{type2} (which are types, not expressions) are
4433 compatible, 0 otherwise. The result of this built-in function can be
4434 used in integer constant expressions.
4436 This built-in function ignores top level qualifiers (e.g., @code{const},
4437 @code{volatile}). For example, @code{int} is equivalent to @code{const
4440 The type @code{int[]} and @code{int[5]} are compatible. On the other
4441 hand, @code{int} and @code{char *} are not compatible, even if the size
4442 of their types, on the particular architecture are the same. Also, the
4443 amount of pointer indirection is taken into account when determining
4444 similarity. Consequently, @code{short *} is not similar to
4445 @code{short **}. Furthermore, two types that are typedefed are
4446 considered compatible if their underlying types are compatible.
4448 An @code{enum} type is considered to be compatible with another
4449 @code{enum} type. For example, @code{enum @{foo, bar@}} is similar to
4450 @code{enum @{hot, dog@}}.
4452 You would typically use this function in code whose execution varies
4453 depending on the arguments' types. For example:
4459 if (__builtin_types_compatible_p (typeof (x), long double)) \
4460 tmp = foo_long_double (tmp); \
4461 else if (__builtin_types_compatible_p (typeof (x), double)) \
4462 tmp = foo_double (tmp); \
4463 else if (__builtin_types_compatible_p (typeof (x), float)) \
4464 tmp = foo_float (tmp); \
4471 @emph{Note:} This construct is only available for C.
4475 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
4477 You can use the built-in function @code{__builtin_choose_expr} to
4478 evaluate code depending on the value of a constant expression. This
4479 built-in function returns @var{exp1} if @var{const_exp}, which is a
4480 constant expression that must be able to be determined at compile time,
4481 is nonzero. Otherwise it returns 0.
4483 This built-in function is analogous to the @samp{? :} operator in C,
4484 except that the expression returned has its type unaltered by promotion
4485 rules. Also, the built-in function does not evaluate the expression
4486 that was not chosen. For example, if @var{const_exp} evaluates to true,
4487 @var{exp2} is not evaluated even if it has side-effects.
4489 This built-in function can return an lvalue if the chosen argument is an
4492 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
4493 type. Similarly, if @var{exp2} is returned, its return type is the same
4500 __builtin_choose_expr (__builtin_types_compatible_p (typeof (x), double), \
4502 __builtin_choose_expr (__builtin_types_compatible_p (typeof (x), float), \
4504 /* @r{The void expression results in a compile-time error} \
4505 @r{when assigning the result to something.} */ \
4509 @emph{Note:} This construct is only available for C. Furthermore, the
4510 unused expression (@var{exp1} or @var{exp2} depending on the value of
4511 @var{const_exp}) may still generate syntax errors. This may change in
4516 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
4517 You can use the built-in function @code{__builtin_constant_p} to
4518 determine if a value is known to be constant at compile-time and hence
4519 that GCC can perform constant-folding on expressions involving that
4520 value. The argument of the function is the value to test. The function
4521 returns the integer 1 if the argument is known to be a compile-time
4522 constant and 0 if it is not known to be a compile-time constant. A
4523 return of 0 does not indicate that the value is @emph{not} a constant,
4524 but merely that GCC cannot prove it is a constant with the specified
4525 value of the @option{-O} option.
4527 You would typically use this function in an embedded application where
4528 memory was a critical resource. If you have some complex calculation,
4529 you may want it to be folded if it involves constants, but need to call
4530 a function if it does not. For example:
4533 #define Scale_Value(X) \
4534 (__builtin_constant_p (X) \
4535 ? ((X) * SCALE + OFFSET) : Scale (X))
4538 You may use this built-in function in either a macro or an inline
4539 function. However, if you use it in an inlined function and pass an
4540 argument of the function as the argument to the built-in, GCC will
4541 never return 1 when you call the inline function with a string constant
4542 or compound literal (@pxref{Compound Literals}) and will not return 1
4543 when you pass a constant numeric value to the inline function unless you
4544 specify the @option{-O} option.
4546 You may also use @code{__builtin_constant_p} in initializers for static
4547 data. For instance, you can write
4550 static const int table[] = @{
4551 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
4557 This is an acceptable initializer even if @var{EXPRESSION} is not a
4558 constant expression. GCC must be more conservative about evaluating the
4559 built-in in this case, because it has no opportunity to perform
4562 Previous versions of GCC did not accept this built-in in data
4563 initializers. The earliest version where it is completely safe is
4567 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
4568 @opindex fprofile-arcs
4569 You may use @code{__builtin_expect} to provide the compiler with
4570 branch prediction information. In general, you should prefer to
4571 use actual profile feedback for this (@option{-fprofile-arcs}), as
4572 programmers are notoriously bad at predicting how their programs
4573 actually perform. However, there are applications in which this
4574 data is hard to collect.
4576 The return value is the value of @var{exp}, which should be an
4577 integral expression. The value of @var{c} must be a compile-time
4578 constant. The semantics of the built-in are that it is expected
4579 that @var{exp} == @var{c}. For example:
4582 if (__builtin_expect (x, 0))
4587 would indicate that we do not expect to call @code{foo}, since
4588 we expect @code{x} to be zero. Since you are limited to integral
4589 expressions for @var{exp}, you should use constructions such as
4592 if (__builtin_expect (ptr != NULL, 1))
4597 when testing pointer or floating-point values.
4600 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
4601 This function is used to minimize cache-miss latency by moving data into
4602 a cache before it is accessed.
4603 You can insert calls to @code{__builtin_prefetch} into code for which
4604 you know addresses of data in memory that is likely to be accessed soon.
4605 If the target supports them, data prefetch instructions will be generated.
4606 If the prefetch is done early enough before the access then the data will
4607 be in the cache by the time it is accessed.
4609 The value of @var{addr} is the address of the memory to prefetch.
4610 There are two optional arguments, @var{rw} and @var{locality}.
4611 The value of @var{rw} is a compile-time constant one or zero; one
4612 means that the prefetch is preparing for a write to the memory address
4613 and zero, the default, means that the prefetch is preparing for a read.
4614 The value @var{locality} must be a compile-time constant integer between
4615 zero and three. A value of zero means that the data has no temporal
4616 locality, so it need not be left in the cache after the access. A value
4617 of three means that the data has a high degree of temporal locality and
4618 should be left in all levels of cache possible. Values of one and two
4619 mean, respectively, a low or moderate degree of temporal locality. The
4623 for (i = 0; i < n; i++)
4626 __builtin_prefetch (&a[i+j], 1, 1);
4627 __builtin_prefetch (&b[i+j], 0, 1);
4632 Data prefetch does not generate faults if @var{addr} is invalid, but
4633 the address expression itself must be valid. For example, a prefetch
4634 of @code{p->next} will not fault if @code{p->next} is not a valid
4635 address, but evaluation will fault if @code{p} is not a valid address.
4637 If the target does not support data prefetch, the address expression
4638 is evaluated if it includes side effects but no other code is generated
4639 and GCC does not issue a warning.
4642 @node Target Builtins
4643 @section Built-in Functions Specific to Particular Target Machines
4645 On some target machines, GCC supports many built-in functions specific
4646 to those machines. Generally these generate calls to specific machine
4647 instructions, but allow the compiler to schedule those calls.
4650 * X86 Built-in Functions::
4651 * PowerPC AltiVec Built-in Functions::
4654 @node X86 Built-in Functions
4655 @subsection X86 Built-in Functions
4657 These built-in functions are available for the i386 and x86-64 family
4658 of computers, depending on the command-line switches used.
4660 The following machine modes are available for use with MMX built-in functions
4661 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
4662 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
4663 vector of eight 8-bit integers. Some of the built-in functions operate on
4664 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
4666 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
4667 of two 32-bit floating point values.
4669 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
4670 floating point values. Some instructions use a vector of four 32-bit
4671 integers, these use @code{V4SI}. Finally, some instructions operate on an
4672 entire vector register, interpreting it as a 128-bit integer, these use mode
4675 The following built-in functions are made available by @option{-mmmx}.
4676 All of them generate the machine instruction that is part of the name.
4679 v8qi __builtin_ia32_paddb (v8qi, v8qi)
4680 v4hi __builtin_ia32_paddw (v4hi, v4hi)
4681 v2si __builtin_ia32_paddd (v2si, v2si)
4682 v8qi __builtin_ia32_psubb (v8qi, v8qi)
4683 v4hi __builtin_ia32_psubw (v4hi, v4hi)
4684 v2si __builtin_ia32_psubd (v2si, v2si)
4685 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
4686 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
4687 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
4688 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
4689 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
4690 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
4691 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
4692 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
4693 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
4694 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
4695 di __builtin_ia32_pand (di, di)
4696 di __builtin_ia32_pandn (di,di)
4697 di __builtin_ia32_por (di, di)
4698 di __builtin_ia32_pxor (di, di)
4699 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
4700 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
4701 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
4702 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
4703 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
4704 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
4705 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
4706 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
4707 v2si __builtin_ia32_punpckhdq (v2si, v2si)
4708 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
4709 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
4710 v2si __builtin_ia32_punpckldq (v2si, v2si)
4711 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
4712 v4hi __builtin_ia32_packssdw (v2si, v2si)
4713 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
4716 The following built-in functions are made available either with
4717 @option{-msse}, or with a combination of @option{-m3dnow} and
4718 @option{-march=athlon}. All of them generate the machine
4719 instruction that is part of the name.
4722 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
4723 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
4724 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
4725 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
4726 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
4727 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
4728 v8qi __builtin_ia32_pminub (v8qi, v8qi)
4729 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
4730 int __builtin_ia32_pextrw (v4hi, int)
4731 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
4732 int __builtin_ia32_pmovmskb (v8qi)
4733 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
4734 void __builtin_ia32_movntq (di *, di)
4735 void __builtin_ia32_sfence (void)
4738 The following built-in functions are available when @option{-msse} is used.
4739 All of them generate the machine instruction that is part of the name.
4742 int __builtin_ia32_comieq (v4sf, v4sf)
4743 int __builtin_ia32_comineq (v4sf, v4sf)
4744 int __builtin_ia32_comilt (v4sf, v4sf)
4745 int __builtin_ia32_comile (v4sf, v4sf)
4746 int __builtin_ia32_comigt (v4sf, v4sf)
4747 int __builtin_ia32_comige (v4sf, v4sf)
4748 int __builtin_ia32_ucomieq (v4sf, v4sf)
4749 int __builtin_ia32_ucomineq (v4sf, v4sf)
4750 int __builtin_ia32_ucomilt (v4sf, v4sf)
4751 int __builtin_ia32_ucomile (v4sf, v4sf)
4752 int __builtin_ia32_ucomigt (v4sf, v4sf)
4753 int __builtin_ia32_ucomige (v4sf, v4sf)
4754 v4sf __builtin_ia32_addps (v4sf, v4sf)
4755 v4sf __builtin_ia32_subps (v4sf, v4sf)
4756 v4sf __builtin_ia32_mulps (v4sf, v4sf)
4757 v4sf __builtin_ia32_divps (v4sf, v4sf)
4758 v4sf __builtin_ia32_addss (v4sf, v4sf)
4759 v4sf __builtin_ia32_subss (v4sf, v4sf)
4760 v4sf __builtin_ia32_mulss (v4sf, v4sf)
4761 v4sf __builtin_ia32_divss (v4sf, v4sf)
4762 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
4763 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
4764 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
4765 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
4766 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
4767 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
4768 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
4769 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
4770 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
4771 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
4772 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
4773 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
4774 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
4775 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
4776 v4si __builtin_ia32_cmpless (v4sf, v4sf)
4777 v4si __builtin_ia32_cmpgtss (v4sf, v4sf)
4778 v4si __builtin_ia32_cmpgess (v4sf, v4sf)
4779 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
4780 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
4781 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
4782 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
4783 v4si __builtin_ia32_cmpngtss (v4sf, v4sf)
4784 v4si __builtin_ia32_cmpngess (v4sf, v4sf)
4785 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
4786 v4sf __builtin_ia32_maxps (v4sf, v4sf)
4787 v4sf __builtin_ia32_maxss (v4sf, v4sf)
4788 v4sf __builtin_ia32_minps (v4sf, v4sf)
4789 v4sf __builtin_ia32_minss (v4sf, v4sf)
4790 v4sf __builtin_ia32_andps (v4sf, v4sf)
4791 v4sf __builtin_ia32_andnps (v4sf, v4sf)
4792 v4sf __builtin_ia32_orps (v4sf, v4sf)
4793 v4sf __builtin_ia32_xorps (v4sf, v4sf)
4794 v4sf __builtin_ia32_movss (v4sf, v4sf)
4795 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
4796 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
4797 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
4798 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
4799 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
4800 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
4801 v2si __builtin_ia32_cvtps2pi (v4sf)
4802 int __builtin_ia32_cvtss2si (v4sf)
4803 v2si __builtin_ia32_cvttps2pi (v4sf)
4804 int __builtin_ia32_cvttss2si (v4sf)
4805 v4sf __builtin_ia32_rcpps (v4sf)
4806 v4sf __builtin_ia32_rsqrtps (v4sf)
4807 v4sf __builtin_ia32_sqrtps (v4sf)
4808 v4sf __builtin_ia32_rcpss (v4sf)
4809 v4sf __builtin_ia32_rsqrtss (v4sf)
4810 v4sf __builtin_ia32_sqrtss (v4sf)
4811 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
4812 void __builtin_ia32_movntps (float *, v4sf)
4813 int __builtin_ia32_movmskps (v4sf)
4816 The following built-in functions are available when @option{-msse} is used.
4819 @item v4sf __builtin_ia32_loadaps (float *)
4820 Generates the @code{movaps} machine instruction as a load from memory.
4821 @item void __builtin_ia32_storeaps (float *, v4sf)
4822 Generates the @code{movaps} machine instruction as a store to memory.
4823 @item v4sf __builtin_ia32_loadups (float *)
4824 Generates the @code{movups} machine instruction as a load from memory.
4825 @item void __builtin_ia32_storeups (float *, v4sf)
4826 Generates the @code{movups} machine instruction as a store to memory.
4827 @item v4sf __builtin_ia32_loadsss (float *)
4828 Generates the @code{movss} machine instruction as a load from memory.
4829 @item void __builtin_ia32_storess (float *, v4sf)
4830 Generates the @code{movss} machine instruction as a store to memory.
4831 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
4832 Generates the @code{movhps} machine instruction as a load from memory.
4833 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
4834 Generates the @code{movlps} machine instruction as a load from memory
4835 @item void __builtin_ia32_storehps (v4sf, v2si *)
4836 Generates the @code{movhps} machine instruction as a store to memory.
4837 @item void __builtin_ia32_storelps (v4sf, v2si *)
4838 Generates the @code{movlps} machine instruction as a store to memory.
4841 The following built-in functions are available when @option{-m3dnow} is used.
4842 All of them generate the machine instruction that is part of the name.
4845 void __builtin_ia32_femms (void)
4846 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
4847 v2si __builtin_ia32_pf2id (v2sf)
4848 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
4849 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
4850 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
4851 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
4852 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
4853 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
4854 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
4855 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
4856 v2sf __builtin_ia32_pfrcp (v2sf)
4857 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
4858 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
4859 v2sf __builtin_ia32_pfrsqrt (v2sf)
4860 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
4861 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
4862 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
4863 v2sf __builtin_ia32_pi2fd (v2si)
4864 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
4867 The following built-in functions are available when both @option{-m3dnow}
4868 and @option{-march=athlon} are used. All of them generate the machine
4869 instruction that is part of the name.
4872 v2si __builtin_ia32_pf2iw (v2sf)
4873 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
4874 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
4875 v2sf __builtin_ia32_pi2fw (v2si)
4876 v2sf __builtin_ia32_pswapdsf (v2sf)
4877 v2si __builtin_ia32_pswapdsi (v2si)
4880 @node PowerPC AltiVec Built-in Functions
4881 @subsection PowerPC AltiVec Built-in Functions
4883 These built-in functions are available for the PowerPC family
4884 of computers, depending on the command-line switches used.
4886 The following machine modes are available for use with AltiVec built-in
4887 functions (@pxref{Vector Extensions}): @code{V4SI} for a vector of four
4888 32-bit integers, @code{V4SF} for a vector of four 32-bit floating point
4889 numbers, @code{V8HI} for a vector of eight 16-bit integers, and
4890 @code{V16QI} for a vector of sixteen 8-bit integers.
4892 The following functions are made available by including
4893 @code{<altivec.h>} and using @option{-maltivec} and
4894 @option{-mabi=altivec}. The functions implement the functionality
4895 described in Motorola's AltiVec Programming Interface Manual.
4897 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
4898 Internally, GCC uses built-in functions to achieve the functionality in
4899 the aforementioned header file, but they are not supported and are
4900 subject to change without notice.
4903 vector signed char vec_abs (vector signed char, vector signed char);
4904 vector signed short vec_abs (vector signed short, vector signed short);
4905 vector signed int vec_abs (vector signed int, vector signed int);
4906 vector signed float vec_abs (vector signed float, vector signed float);
4908 vector signed char vec_abss (vector signed char, vector signed char);
4909 vector signed short vec_abss (vector signed short, vector signed short);
4911 vector signed char vec_add (vector signed char, vector signed char);
4912 vector unsigned char vec_add (vector signed char, vector unsigned char);
4914 vector unsigned char vec_add (vector unsigned char, vector signed char);
4916 vector unsigned char vec_add (vector unsigned char,
4917 vector unsigned char);
4918 vector signed short vec_add (vector signed short, vector signed short);
4919 vector unsigned short vec_add (vector signed short,
4920 vector unsigned short);
4921 vector unsigned short vec_add (vector unsigned short,
4922 vector signed short);
4923 vector unsigned short vec_add (vector unsigned short,
4924 vector unsigned short);
4925 vector signed int vec_add (vector signed int, vector signed int);
4926 vector unsigned int vec_add (vector signed int, vector unsigned int);
4927 vector unsigned int vec_add (vector unsigned int, vector signed int);
4928 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
4929 vector float vec_add (vector float, vector float);
4931 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
4933 vector unsigned char vec_adds (vector signed char,
4934 vector unsigned char);
4935 vector unsigned char vec_adds (vector unsigned char,
4936 vector signed char);
4937 vector unsigned char vec_adds (vector unsigned char,
4938 vector unsigned char);
4939 vector signed char vec_adds (vector signed char, vector signed char);
4940 vector unsigned short vec_adds (vector signed short,
4941 vector unsigned short);
4942 vector unsigned short vec_adds (vector unsigned short,
4943 vector signed short);
4944 vector unsigned short vec_adds (vector unsigned short,
4945 vector unsigned short);
4946 vector signed short vec_adds (vector signed short, vector signed short);
4948 vector unsigned int vec_adds (vector signed int, vector unsigned int);
4949 vector unsigned int vec_adds (vector unsigned int, vector signed int);
4950 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
4952 vector signed int vec_adds (vector signed int, vector signed int);
4954 vector float vec_and (vector float, vector float);
4955 vector float vec_and (vector float, vector signed int);
4956 vector float vec_and (vector signed int, vector float);
4957 vector signed int vec_and (vector signed int, vector signed int);
4958 vector unsigned int vec_and (vector signed int, vector unsigned int);
4959 vector unsigned int vec_and (vector unsigned int, vector signed int);
4960 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
4961 vector signed short vec_and (vector signed short, vector signed short);
4962 vector unsigned short vec_and (vector signed short,
4963 vector unsigned short);
4964 vector unsigned short vec_and (vector unsigned short,
4965 vector signed short);
4966 vector unsigned short vec_and (vector unsigned short,
4967 vector unsigned short);
4968 vector signed char vec_and (vector signed char, vector signed char);
4969 vector unsigned char vec_and (vector signed char, vector unsigned char);
4971 vector unsigned char vec_and (vector unsigned char, vector signed char);
4973 vector unsigned char vec_and (vector unsigned char,
4974 vector unsigned char);
4976 vector float vec_andc (vector float, vector float);
4977 vector float vec_andc (vector float, vector signed int);
4978 vector float vec_andc (vector signed int, vector float);
4979 vector signed int vec_andc (vector signed int, vector signed int);
4980 vector unsigned int vec_andc (vector signed int, vector unsigned int);
4981 vector unsigned int vec_andc (vector unsigned int, vector signed int);
4982 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
4984 vector signed short vec_andc (vector signed short, vector signed short);
4986 vector unsigned short vec_andc (vector signed short,
4987 vector unsigned short);
4988 vector unsigned short vec_andc (vector unsigned short,
4989 vector signed short);
4990 vector unsigned short vec_andc (vector unsigned short,
4991 vector unsigned short);
4992 vector signed char vec_andc (vector signed char, vector signed char);
4993 vector unsigned char vec_andc (vector signed char,
4994 vector unsigned char);
4995 vector unsigned char vec_andc (vector unsigned char,
4996 vector signed char);
4997 vector unsigned char vec_andc (vector unsigned char,
4998 vector unsigned char);
5000 vector unsigned char vec_avg (vector unsigned char,
5001 vector unsigned char);
5002 vector signed char vec_avg (vector signed char, vector signed char);
5003 vector unsigned short vec_avg (vector unsigned short,
5004 vector unsigned short);
5005 vector signed short vec_avg (vector signed short, vector signed short);
5006 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
5007 vector signed int vec_avg (vector signed int, vector signed int);
5009 vector float vec_ceil (vector float);
5011 vector signed int vec_cmpb (vector float, vector float);
5013 vector signed char vec_cmpeq (vector signed char, vector signed char);
5014 vector signed char vec_cmpeq (vector unsigned char,
5015 vector unsigned char);
5016 vector signed short vec_cmpeq (vector signed short,
5017 vector signed short);
5018 vector signed short vec_cmpeq (vector unsigned short,
5019 vector unsigned short);
5020 vector signed int vec_cmpeq (vector signed int, vector signed int);
5021 vector signed int vec_cmpeq (vector unsigned int, vector unsigned int);
5022 vector signed int vec_cmpeq (vector float, vector float);
5024 vector signed int vec_cmpge (vector float, vector float);
5026 vector signed char vec_cmpgt (vector unsigned char,
5027 vector unsigned char);
5028 vector signed char vec_cmpgt (vector signed char, vector signed char);
5029 vector signed short vec_cmpgt (vector unsigned short,
5030 vector unsigned short);
5031 vector signed short vec_cmpgt (vector signed short,
5032 vector signed short);
5033 vector signed int vec_cmpgt (vector unsigned int, vector unsigned int);
5034 vector signed int vec_cmpgt (vector signed int, vector signed int);
5035 vector signed int vec_cmpgt (vector float, vector float);
5037 vector signed int vec_cmple (vector float, vector float);
5039 vector signed char vec_cmplt (vector unsigned char,
5040 vector unsigned char);
5041 vector signed char vec_cmplt (vector signed char, vector signed char);
5042 vector signed short vec_cmplt (vector unsigned short,
5043 vector unsigned short);
5044 vector signed short vec_cmplt (vector signed short,
5045 vector signed short);
5046 vector signed int vec_cmplt (vector unsigned int, vector unsigned int);
5047 vector signed int vec_cmplt (vector signed int, vector signed int);
5048 vector signed int vec_cmplt (vector float, vector float);
5050 vector float vec_ctf (vector unsigned int, const char);
5051 vector float vec_ctf (vector signed int, const char);
5053 vector signed int vec_cts (vector float, const char);
5055 vector unsigned int vec_ctu (vector float, const char);
5057 void vec_dss (const char);
5059 void vec_dssall (void);
5061 void vec_dst (void *, int, const char);
5063 void vec_dstst (void *, int, const char);
5065 void vec_dststt (void *, int, const char);
5067 void vec_dstt (void *, int, const char);
5069 vector float vec_expte (vector float, vector float);
5071 vector float vec_floor (vector float, vector float);
5073 vector float vec_ld (int, vector float *);
5074 vector float vec_ld (int, float *):
5075 vector signed int vec_ld (int, int *);
5076 vector signed int vec_ld (int, vector signed int *);
5077 vector unsigned int vec_ld (int, vector unsigned int *);
5078 vector unsigned int vec_ld (int, unsigned int *);
5079 vector signed short vec_ld (int, short *, vector signed short *);
5080 vector unsigned short vec_ld (int, unsigned short *,
5081 vector unsigned short *);
5082 vector signed char vec_ld (int, signed char *);
5083 vector signed char vec_ld (int, vector signed char *);
5084 vector unsigned char vec_ld (int, unsigned char *);
5085 vector unsigned char vec_ld (int, vector unsigned char *);
5087 vector signed char vec_lde (int, signed char *);
5088 vector unsigned char vec_lde (int, unsigned char *);
5089 vector signed short vec_lde (int, short *);
5090 vector unsigned short vec_lde (int, unsigned short *);
5091 vector float vec_lde (int, float *);
5092 vector signed int vec_lde (int, int *);
5093 vector unsigned int vec_lde (int, unsigned int *);
5095 void float vec_ldl (int, float *);
5096 void float vec_ldl (int, vector float *);
5097 void signed int vec_ldl (int, vector signed int *);
5098 void signed int vec_ldl (int, int *);
5099 void unsigned int vec_ldl (int, unsigned int *);
5100 void unsigned int vec_ldl (int, vector unsigned int *);
5101 void signed short vec_ldl (int, vector signed short *);
5102 void signed short vec_ldl (int, short *);
5103 void unsigned short vec_ldl (int, vector unsigned short *);
5104 void unsigned short vec_ldl (int, unsigned short *);
5105 void signed char vec_ldl (int, vector signed char *);
5106 void signed char vec_ldl (int, signed char *);
5107 void unsigned char vec_ldl (int, vector unsigned char *);
5108 void unsigned char vec_ldl (int, unsigned char *);
5110 vector float vec_loge (vector float);
5112 vector unsigned char vec_lvsl (int, void *, int *);
5114 vector unsigned char vec_lvsr (int, void *, int *);
5116 vector float vec_madd (vector float, vector float, vector float);
5118 vector signed short vec_madds (vector signed short, vector signed short,
5119 vector signed short);
5121 vector unsigned char vec_max (vector signed char, vector unsigned char);
5123 vector unsigned char vec_max (vector unsigned char, vector signed char);
5125 vector unsigned char vec_max (vector unsigned char,
5126 vector unsigned char);
5127 vector signed char vec_max (vector signed char, vector signed char);
5128 vector unsigned short vec_max (vector signed short,
5129 vector unsigned short);
5130 vector unsigned short vec_max (vector unsigned short,
5131 vector signed short);
5132 vector unsigned short vec_max (vector unsigned short,
5133 vector unsigned short);
5134 vector signed short vec_max (vector signed short, vector signed short);
5135 vector unsigned int vec_max (vector signed int, vector unsigned int);
5136 vector unsigned int vec_max (vector unsigned int, vector signed int);
5137 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
5138 vector signed int vec_max (vector signed int, vector signed int);
5139 vector float vec_max (vector float, vector float);
5141 vector signed char vec_mergeh (vector signed char, vector signed char);
5142 vector unsigned char vec_mergeh (vector unsigned char,
5143 vector unsigned char);
5144 vector signed short vec_mergeh (vector signed short,
5145 vector signed short);
5146 vector unsigned short vec_mergeh (vector unsigned short,
5147 vector unsigned short);
5148 vector float vec_mergeh (vector float, vector float);
5149 vector signed int vec_mergeh (vector signed int, vector signed int);
5150 vector unsigned int vec_mergeh (vector unsigned int,
5151 vector unsigned int);
5153 vector signed char vec_mergel (vector signed char, vector signed char);
5154 vector unsigned char vec_mergel (vector unsigned char,
5155 vector unsigned char);
5156 vector signed short vec_mergel (vector signed short,
5157 vector signed short);
5158 vector unsigned short vec_mergel (vector unsigned short,
5159 vector unsigned short);
5160 vector float vec_mergel (vector float, vector float);
5161 vector signed int vec_mergel (vector signed int, vector signed int);
5162 vector unsigned int vec_mergel (vector unsigned int,
5163 vector unsigned int);
5165 vector unsigned short vec_mfvscr (void);
5167 vector unsigned char vec_min (vector signed char, vector unsigned char);
5169 vector unsigned char vec_min (vector unsigned char, vector signed char);
5171 vector unsigned char vec_min (vector unsigned char,
5172 vector unsigned char);
5173 vector signed char vec_min (vector signed char, vector signed char);
5174 vector unsigned short vec_min (vector signed short,
5175 vector unsigned short);
5176 vector unsigned short vec_min (vector unsigned short,
5177 vector signed short);
5178 vector unsigned short vec_min (vector unsigned short,
5179 vector unsigned short);
5180 vector signed short vec_min (vector signed short, vector signed short);
5181 vector unsigned int vec_min (vector signed int, vector unsigned int);
5182 vector unsigned int vec_min (vector unsigned int, vector signed int);
5183 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
5184 vector signed int vec_min (vector signed int, vector signed int);
5185 vector float vec_min (vector float, vector float);
5187 vector signed short vec_mladd (vector signed short, vector signed short,
5188 vector signed short);
5189 vector signed short vec_mladd (vector signed short,
5190 vector unsigned short,
5191 vector unsigned short);
5192 vector signed short vec_mladd (vector unsigned short,
5193 vector signed short,
5194 vector signed short);
5195 vector unsigned short vec_mladd (vector unsigned short,
5196 vector unsigned short,
5197 vector unsigned short);
5199 vector signed short vec_mradds (vector signed short,
5200 vector signed short,
5201 vector signed short);
5203 vector unsigned int vec_msum (vector unsigned char,
5204 vector unsigned char,
5205 vector unsigned int);
5206 vector signed int vec_msum (vector signed char, vector unsigned char,
5208 vector unsigned int vec_msum (vector unsigned short,
5209 vector unsigned short,
5210 vector unsigned int);
5211 vector signed int vec_msum (vector signed short, vector signed short,
5214 vector unsigned int vec_msums (vector unsigned short,
5215 vector unsigned short,
5216 vector unsigned int);
5217 vector signed int vec_msums (vector signed short, vector signed short,
5220 void vec_mtvscr (vector signed int);
5221 void vec_mtvscr (vector unsigned int);
5222 void vec_mtvscr (vector signed short);
5223 void vec_mtvscr (vector unsigned short);
5224 void vec_mtvscr (vector signed char);
5225 void vec_mtvscr (vector unsigned char);
5227 vector unsigned short vec_mule (vector unsigned char,
5228 vector unsigned char);
5229 vector signed short vec_mule (vector signed char, vector signed char);
5230 vector unsigned int vec_mule (vector unsigned short,
5231 vector unsigned short);
5232 vector signed int vec_mule (vector signed short, vector signed short);
5234 vector unsigned short vec_mulo (vector unsigned char,
5235 vector unsigned char);
5236 vector signed short vec_mulo (vector signed char, vector signed char);
5237 vector unsigned int vec_mulo (vector unsigned short,
5238 vector unsigned short);
5239 vector signed int vec_mulo (vector signed short, vector signed short);
5241 vector float vec_nmsub (vector float, vector float, vector float);
5243 vector float vec_nor (vector float, vector float);
5244 vector signed int vec_nor (vector signed int, vector signed int);
5245 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
5246 vector signed short vec_nor (vector signed short, vector signed short);
5247 vector unsigned short vec_nor (vector unsigned short,
5248 vector unsigned short);
5249 vector signed char vec_nor (vector signed char, vector signed char);
5250 vector unsigned char vec_nor (vector unsigned char,
5251 vector unsigned char);
5253 vector float vec_or (vector float, vector float);
5254 vector float vec_or (vector float, vector signed int);
5255 vector float vec_or (vector signed int, vector float);
5256 vector signed int vec_or (vector signed int, vector signed int);
5257 vector unsigned int vec_or (vector signed int, vector unsigned int);
5258 vector unsigned int vec_or (vector unsigned int, vector signed int);
5259 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
5260 vector signed short vec_or (vector signed short, vector signed short);
5261 vector unsigned short vec_or (vector signed short,
5262 vector unsigned short);
5263 vector unsigned short vec_or (vector unsigned short,
5264 vector signed short);
5265 vector unsigned short vec_or (vector unsigned short,
5266 vector unsigned short);
5267 vector signed char vec_or (vector signed char, vector signed char);
5268 vector unsigned char vec_or (vector signed char, vector unsigned char);
5269 vector unsigned char vec_or (vector unsigned char, vector signed char);
5270 vector unsigned char vec_or (vector unsigned char,
5271 vector unsigned char);
5273 vector signed char vec_pack (vector signed short, vector signed short);
5274 vector unsigned char vec_pack (vector unsigned short,
5275 vector unsigned short);
5276 vector signed short vec_pack (vector signed int, vector signed int);
5277 vector unsigned short vec_pack (vector unsigned int,
5278 vector unsigned int);
5280 vector signed short vec_packpx (vector unsigned int,
5281 vector unsigned int);
5283 vector unsigned char vec_packs (vector unsigned short,
5284 vector unsigned short);
5285 vector signed char vec_packs (vector signed short, vector signed short);
5287 vector unsigned short vec_packs (vector unsigned int,
5288 vector unsigned int);
5289 vector signed short vec_packs (vector signed int, vector signed int);
5291 vector unsigned char vec_packsu (vector unsigned short,
5292 vector unsigned short);
5293 vector unsigned char vec_packsu (vector signed short,
5294 vector signed short);
5295 vector unsigned short vec_packsu (vector unsigned int,
5296 vector unsigned int);
5297 vector unsigned short vec_packsu (vector signed int, vector signed int);
5299 vector float vec_perm (vector float, vector float,
5300 vector unsigned char);
5301 vector signed int vec_perm (vector signed int, vector signed int,
5302 vector unsigned char);
5303 vector unsigned int vec_perm (vector unsigned int, vector unsigned int,
5304 vector unsigned char);
5305 vector signed short vec_perm (vector signed short, vector signed short,
5306 vector unsigned char);
5307 vector unsigned short vec_perm (vector unsigned short,
5308 vector unsigned short,
5309 vector unsigned char);
5310 vector signed char vec_perm (vector signed char, vector signed char,
5311 vector unsigned char);
5312 vector unsigned char vec_perm (vector unsigned char,
5313 vector unsigned char,
5314 vector unsigned char);
5316 vector float vec_re (vector float);
5318 vector signed char vec_rl (vector signed char, vector unsigned char);
5319 vector unsigned char vec_rl (vector unsigned char,
5320 vector unsigned char);
5321 vector signed short vec_rl (vector signed short, vector unsigned short);
5323 vector unsigned short vec_rl (vector unsigned short,
5324 vector unsigned short);
5325 vector signed int vec_rl (vector signed int, vector unsigned int);
5326 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
5328 vector float vec_round (vector float);
5330 vector float vec_rsqrte (vector float);
5332 vector float vec_sel (vector float, vector float, vector signed int);
5333 vector float vec_sel (vector float, vector float, vector unsigned int);
5334 vector signed int vec_sel (vector signed int, vector signed int,
5336 vector signed int vec_sel (vector signed int, vector signed int,
5337 vector unsigned int);
5338 vector unsigned int vec_sel (vector unsigned int, vector unsigned int,
5340 vector unsigned int vec_sel (vector unsigned int, vector unsigned int,
5341 vector unsigned int);
5342 vector signed short vec_sel (vector signed short, vector signed short,
5343 vector signed short);
5344 vector signed short vec_sel (vector signed short, vector signed short,
5345 vector unsigned short);
5346 vector unsigned short vec_sel (vector unsigned short,
5347 vector unsigned short,
5348 vector signed short);
5349 vector unsigned short vec_sel (vector unsigned short,
5350 vector unsigned short,
5351 vector unsigned short);
5352 vector signed char vec_sel (vector signed char, vector signed char,
5353 vector signed char);
5354 vector signed char vec_sel (vector signed char, vector signed char,
5355 vector unsigned char);
5356 vector unsigned char vec_sel (vector unsigned char,
5357 vector unsigned char,
5358 vector signed char);
5359 vector unsigned char vec_sel (vector unsigned char,
5360 vector unsigned char,
5361 vector unsigned char);
5363 vector signed char vec_sl (vector signed char, vector unsigned char);
5364 vector unsigned char vec_sl (vector unsigned char,
5365 vector unsigned char);
5366 vector signed short vec_sl (vector signed short, vector unsigned short);
5368 vector unsigned short vec_sl (vector unsigned short,
5369 vector unsigned short);
5370 vector signed int vec_sl (vector signed int, vector unsigned int);
5371 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
5373 vector float vec_sld (vector float, vector float, const char);
5374 vector signed int vec_sld (vector signed int, vector signed int,
5376 vector unsigned int vec_sld (vector unsigned int, vector unsigned int,
5378 vector signed short vec_sld (vector signed short, vector signed short,
5380 vector unsigned short vec_sld (vector unsigned short,
5381 vector unsigned short, const char);
5382 vector signed char vec_sld (vector signed char, vector signed char,
5384 vector unsigned char vec_sld (vector unsigned char,
5385 vector unsigned char,
5388 vector signed int vec_sll (vector signed int, vector unsigned int);
5389 vector signed int vec_sll (vector signed int, vector unsigned short);
5390 vector signed int vec_sll (vector signed int, vector unsigned char);
5391 vector unsigned int vec_sll (vector unsigned int, vector unsigned int);
5392 vector unsigned int vec_sll (vector unsigned int,
5393 vector unsigned short);
5394 vector unsigned int vec_sll (vector unsigned int, vector unsigned char);
5396 vector signed short vec_sll (vector signed short, vector unsigned int);
5397 vector signed short vec_sll (vector signed short,
5398 vector unsigned short);
5399 vector signed short vec_sll (vector signed short, vector unsigned char);
5401 vector unsigned short vec_sll (vector unsigned short,
5402 vector unsigned int);
5403 vector unsigned short vec_sll (vector unsigned short,
5404 vector unsigned short);
5405 vector unsigned short vec_sll (vector unsigned short,
5406 vector unsigned char);
5407 vector signed char vec_sll (vector signed char, vector unsigned int);
5408 vector signed char vec_sll (vector signed char, vector unsigned short);
5409 vector signed char vec_sll (vector signed char, vector unsigned char);
5410 vector unsigned char vec_sll (vector unsigned char,
5411 vector unsigned int);
5412 vector unsigned char vec_sll (vector unsigned char,
5413 vector unsigned short);
5414 vector unsigned char vec_sll (vector unsigned char,
5415 vector unsigned char);
5417 vector float vec_slo (vector float, vector signed char);
5418 vector float vec_slo (vector float, vector unsigned char);
5419 vector signed int vec_slo (vector signed int, vector signed char);
5420 vector signed int vec_slo (vector signed int, vector unsigned char);
5421 vector unsigned int vec_slo (vector unsigned int, vector signed char);
5422 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
5424 vector signed short vec_slo (vector signed short, vector signed char);
5425 vector signed short vec_slo (vector signed short, vector unsigned char);
5427 vector unsigned short vec_slo (vector unsigned short,
5428 vector signed char);
5429 vector unsigned short vec_slo (vector unsigned short,
5430 vector unsigned char);
5431 vector signed char vec_slo (vector signed char, vector signed char);
5432 vector signed char vec_slo (vector signed char, vector unsigned char);
5433 vector unsigned char vec_slo (vector unsigned char, vector signed char);
5435 vector unsigned char vec_slo (vector unsigned char,
5436 vector unsigned char);
5438 vector signed char vec_splat (vector signed char, const char);
5439 vector unsigned char vec_splat (vector unsigned char, const char);
5440 vector signed short vec_splat (vector signed short, const char);
5441 vector unsigned short vec_splat (vector unsigned short, const char);
5442 vector float vec_splat (vector float, const char);
5443 vector signed int vec_splat (vector signed int, const char);
5444 vector unsigned int vec_splat (vector unsigned int, const char);
5446 vector signed char vec_splat_s8 (const char);
5448 vector signed short vec_splat_s16 (const char);
5450 vector signed int vec_splat_s32 (const char);
5452 vector unsigned char vec_splat_u8 (const char);
5454 vector unsigned short vec_splat_u16 (const char);
5456 vector unsigned int vec_splat_u32 (const char);
5458 vector signed char vec_sr (vector signed char, vector unsigned char);
5459 vector unsigned char vec_sr (vector unsigned char,
5460 vector unsigned char);
5461 vector signed short vec_sr (vector signed short, vector unsigned short);
5463 vector unsigned short vec_sr (vector unsigned short,
5464 vector unsigned short);
5465 vector signed int vec_sr (vector signed int, vector unsigned int);
5466 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
5468 vector signed char vec_sra (vector signed char, vector unsigned char);
5469 vector unsigned char vec_sra (vector unsigned char,
5470 vector unsigned char);
5471 vector signed short vec_sra (vector signed short,
5472 vector unsigned short);
5473 vector unsigned short vec_sra (vector unsigned short,
5474 vector unsigned short);
5475 vector signed int vec_sra (vector signed int, vector unsigned int);
5476 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
5478 vector signed int vec_srl (vector signed int, vector unsigned int);
5479 vector signed int vec_srl (vector signed int, vector unsigned short);
5480 vector signed int vec_srl (vector signed int, vector unsigned char);
5481 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
5482 vector unsigned int vec_srl (vector unsigned int,
5483 vector unsigned short);
5484 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
5486 vector signed short vec_srl (vector signed short, vector unsigned int);
5487 vector signed short vec_srl (vector signed short,
5488 vector unsigned short);
5489 vector signed short vec_srl (vector signed short, vector unsigned char);
5491 vector unsigned short vec_srl (vector unsigned short,
5492 vector unsigned int);
5493 vector unsigned short vec_srl (vector unsigned short,
5494 vector unsigned short);
5495 vector unsigned short vec_srl (vector unsigned short,
5496 vector unsigned char);
5497 vector signed char vec_srl (vector signed char, vector unsigned int);
5498 vector signed char vec_srl (vector signed char, vector unsigned short);
5499 vector signed char vec_srl (vector signed char, vector unsigned char);
5500 vector unsigned char vec_srl (vector unsigned char,
5501 vector unsigned int);
5502 vector unsigned char vec_srl (vector unsigned char,
5503 vector unsigned short);
5504 vector unsigned char vec_srl (vector unsigned char,
5505 vector unsigned char);
5507 vector float vec_sro (vector float, vector signed char);
5508 vector float vec_sro (vector float, vector unsigned char);
5509 vector signed int vec_sro (vector signed int, vector signed char);
5510 vector signed int vec_sro (vector signed int, vector unsigned char);
5511 vector unsigned int vec_sro (vector unsigned int, vector signed char);
5512 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
5514 vector signed short vec_sro (vector signed short, vector signed char);
5515 vector signed short vec_sro (vector signed short, vector unsigned char);
5517 vector unsigned short vec_sro (vector unsigned short,
5518 vector signed char);
5519 vector unsigned short vec_sro (vector unsigned short,
5520 vector unsigned char);
5521 vector signed char vec_sro (vector signed char, vector signed char);
5522 vector signed char vec_sro (vector signed char, vector unsigned char);
5523 vector unsigned char vec_sro (vector unsigned char, vector signed char);
5525 vector unsigned char vec_sro (vector unsigned char,
5526 vector unsigned char);
5528 void vec_st (vector float, int, float *);
5529 void vec_st (vector float, int, vector float *);
5530 void vec_st (vector signed int, int, int *);
5531 void vec_st (vector signed int, int, unsigned int *);
5532 void vec_st (vector unsigned int, int, unsigned int *);
5533 void vec_st (vector unsigned int, int, vector unsigned int *);
5534 void vec_st (vector signed short, int, short *);
5535 void vec_st (vector signed short, int, vector unsigned short *);
5536 void vec_st (vector signed short, int, vector signed short *);
5537 void vec_st (vector unsigned short, int, unsigned short *);
5538 void vec_st (vector unsigned short, int, vector unsigned short *);
5539 void vec_st (vector signed char, int, signed char *);
5540 void vec_st (vector signed char, int, unsigned char *);
5541 void vec_st (vector signed char, int, vector signed char *);
5542 void vec_st (vector unsigned char, int, unsigned char *);
5543 void vec_st (vector unsigned char, int, vector unsigned char *);
5545 void vec_ste (vector signed char, int, unsigned char *);
5546 void vec_ste (vector signed char, int, signed char *);
5547 void vec_ste (vector unsigned char, int, unsigned char *);
5548 void vec_ste (vector signed short, int, short *);
5549 void vec_ste (vector signed short, int, unsigned short *);
5550 void vec_ste (vector unsigned short, int, void *);
5551 void vec_ste (vector signed int, int, unsigned int *);
5552 void vec_ste (vector signed int, int, int *);
5553 void vec_ste (vector unsigned int, int, unsigned int *);
5554 void vec_ste (vector float, int, float *);
5556 void vec_stl (vector float, int, vector float *);
5557 void vec_stl (vector float, int, float *);
5558 void vec_stl (vector signed int, int, vector signed int *);
5559 void vec_stl (vector signed int, int, int *);
5560 void vec_stl (vector signed int, int, unsigned int *);
5561 void vec_stl (vector unsigned int, int, vector unsigned int *);
5562 void vec_stl (vector unsigned int, int, unsigned int *);
5563 void vec_stl (vector signed short, int, short *);
5564 void vec_stl (vector signed short, int, unsigned short *);
5565 void vec_stl (vector signed short, int, vector signed short *);
5566 void vec_stl (vector unsigned short, int, unsigned short *);
5567 void vec_stl (vector unsigned short, int, vector signed short *);
5568 void vec_stl (vector signed char, int, signed char *);
5569 void vec_stl (vector signed char, int, unsigned char *);
5570 void vec_stl (vector signed char, int, vector signed char *);
5571 void vec_stl (vector unsigned char, int, unsigned char *);
5572 void vec_stl (vector unsigned char, int, vector unsigned char *);
5574 vector signed char vec_sub (vector signed char, vector signed char);
5575 vector unsigned char vec_sub (vector signed char, vector unsigned char);
5577 vector unsigned char vec_sub (vector unsigned char, vector signed char);
5579 vector unsigned char vec_sub (vector unsigned char,
5580 vector unsigned char);
5581 vector signed short vec_sub (vector signed short, vector signed short);
5582 vector unsigned short vec_sub (vector signed short,
5583 vector unsigned short);
5584 vector unsigned short vec_sub (vector unsigned short,
5585 vector signed short);
5586 vector unsigned short vec_sub (vector unsigned short,
5587 vector unsigned short);
5588 vector signed int vec_sub (vector signed int, vector signed int);
5589 vector unsigned int vec_sub (vector signed int, vector unsigned int);
5590 vector unsigned int vec_sub (vector unsigned int, vector signed int);
5591 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
5592 vector float vec_sub (vector float, vector float);
5594 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
5596 vector unsigned char vec_subs (vector signed char,
5597 vector unsigned char);
5598 vector unsigned char vec_subs (vector unsigned char,
5599 vector signed char);
5600 vector unsigned char vec_subs (vector unsigned char,
5601 vector unsigned char);
5602 vector signed char vec_subs (vector signed char, vector signed char);
5603 vector unsigned short vec_subs (vector signed short,
5604 vector unsigned short);
5605 vector unsigned short vec_subs (vector unsigned short,
5606 vector signed short);
5607 vector unsigned short vec_subs (vector unsigned short,
5608 vector unsigned short);
5609 vector signed short vec_subs (vector signed short, vector signed short);
5611 vector unsigned int vec_subs (vector signed int, vector unsigned int);
5612 vector unsigned int vec_subs (vector unsigned int, vector signed int);
5613 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
5615 vector signed int vec_subs (vector signed int, vector signed int);
5617 vector unsigned int vec_sum4s (vector unsigned char,
5618 vector unsigned int);
5619 vector signed int vec_sum4s (vector signed char, vector signed int);
5620 vector signed int vec_sum4s (vector signed short, vector signed int);
5622 vector signed int vec_sum2s (vector signed int, vector signed int);
5624 vector signed int vec_sums (vector signed int, vector signed int);
5626 vector float vec_trunc (vector float);
5628 vector signed short vec_unpackh (vector signed char);
5629 vector unsigned int vec_unpackh (vector signed short);
5630 vector signed int vec_unpackh (vector signed short);
5632 vector signed short vec_unpackl (vector signed char);
5633 vector unsigned int vec_unpackl (vector signed short);
5634 vector signed int vec_unpackl (vector signed short);
5636 vector float vec_xor (vector float, vector float);
5637 vector float vec_xor (vector float, vector signed int);
5638 vector float vec_xor (vector signed int, vector float);
5639 vector signed int vec_xor (vector signed int, vector signed int);
5640 vector unsigned int vec_xor (vector signed int, vector unsigned int);
5641 vector unsigned int vec_xor (vector unsigned int, vector signed int);
5642 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
5643 vector signed short vec_xor (vector signed short, vector signed short);
5644 vector unsigned short vec_xor (vector signed short,
5645 vector unsigned short);
5646 vector unsigned short vec_xor (vector unsigned short,
5647 vector signed short);
5648 vector unsigned short vec_xor (vector unsigned short,
5649 vector unsigned short);
5650 vector signed char vec_xor (vector signed char, vector signed char);
5651 vector unsigned char vec_xor (vector signed char, vector unsigned char);
5653 vector unsigned char vec_xor (vector unsigned char, vector signed char);
5655 vector unsigned char vec_xor (vector unsigned char,
5656 vector unsigned char);
5658 vector signed int vec_all_eq (vector signed char, vector unsigned char);
5660 vector signed int vec_all_eq (vector signed char, vector signed char);
5661 vector signed int vec_all_eq (vector unsigned char, vector signed char);
5663 vector signed int vec_all_eq (vector unsigned char,
5664 vector unsigned char);
5665 vector signed int vec_all_eq (vector signed short,
5666 vector unsigned short);
5667 vector signed int vec_all_eq (vector signed short, vector signed short);
5669 vector signed int vec_all_eq (vector unsigned short,
5670 vector signed short);
5671 vector signed int vec_all_eq (vector unsigned short,
5672 vector unsigned short);
5673 vector signed int vec_all_eq (vector signed int, vector unsigned int);
5674 vector signed int vec_all_eq (vector signed int, vector signed int);
5675 vector signed int vec_all_eq (vector unsigned int, vector signed int);
5676 vector signed int vec_all_eq (vector unsigned int, vector unsigned int);
5678 vector signed int vec_all_eq (vector float, vector float);
5680 vector signed int vec_all_ge (vector signed char, vector unsigned char);
5682 vector signed int vec_all_ge (vector unsigned char, vector signed char);
5684 vector signed int vec_all_ge (vector unsigned char,
5685 vector unsigned char);
5686 vector signed int vec_all_ge (vector signed char, vector signed char);
5687 vector signed int vec_all_ge (vector signed short,
5688 vector unsigned short);
5689 vector signed int vec_all_ge (vector unsigned short,
5690 vector signed short);
5691 vector signed int vec_all_ge (vector unsigned short,
5692 vector unsigned short);
5693 vector signed int vec_all_ge (vector signed short, vector signed short);
5695 vector signed int vec_all_ge (vector signed int, vector unsigned int);
5696 vector signed int vec_all_ge (vector unsigned int, vector signed int);
5697 vector signed int vec_all_ge (vector unsigned int, vector unsigned int);
5699 vector signed int vec_all_ge (vector signed int, vector signed int);
5700 vector signed int vec_all_ge (vector float, vector float);
5702 vector signed int vec_all_gt (vector signed char, vector unsigned char);
5704 vector signed int vec_all_gt (vector unsigned char, vector signed char);
5706 vector signed int vec_all_gt (vector unsigned char,
5707 vector unsigned char);
5708 vector signed int vec_all_gt (vector signed char, vector signed char);
5709 vector signed int vec_all_gt (vector signed short,
5710 vector unsigned short);
5711 vector signed int vec_all_gt (vector unsigned short,
5712 vector signed short);
5713 vector signed int vec_all_gt (vector unsigned short,
5714 vector unsigned short);
5715 vector signed int vec_all_gt (vector signed short, vector signed short);
5717 vector signed int vec_all_gt (vector signed int, vector unsigned int);
5718 vector signed int vec_all_gt (vector unsigned int, vector signed int);
5719 vector signed int vec_all_gt (vector unsigned int, vector unsigned int);
5721 vector signed int vec_all_gt (vector signed int, vector signed int);
5722 vector signed int vec_all_gt (vector float, vector float);
5724 vector signed int vec_all_in (vector float, vector float);
5726 vector signed int vec_all_le (vector signed char, vector unsigned char);
5728 vector signed int vec_all_le (vector unsigned char, vector signed char);
5730 vector signed int vec_all_le (vector unsigned char,
5731 vector unsigned char);
5732 vector signed int vec_all_le (vector signed char, vector signed char);
5733 vector signed int vec_all_le (vector signed short,
5734 vector unsigned short);
5735 vector signed int vec_all_le (vector unsigned short,
5736 vector signed short);
5737 vector signed int vec_all_le (vector unsigned short,
5738 vector unsigned short);
5739 vector signed int vec_all_le (vector signed short, vector signed short);
5741 vector signed int vec_all_le (vector signed int, vector unsigned int);
5742 vector signed int vec_all_le (vector unsigned int, vector signed int);
5743 vector signed int vec_all_le (vector unsigned int, vector unsigned int);
5745 vector signed int vec_all_le (vector signed int, vector signed int);
5746 vector signed int vec_all_le (vector float, vector float);
5748 vector signed int vec_all_lt (vector signed char, vector unsigned char);
5750 vector signed int vec_all_lt (vector unsigned char, vector signed char);
5752 vector signed int vec_all_lt (vector unsigned char,
5753 vector unsigned char);
5754 vector signed int vec_all_lt (vector signed char, vector signed char);
5755 vector signed int vec_all_lt (vector signed short,
5756 vector unsigned short);
5757 vector signed int vec_all_lt (vector unsigned short,
5758 vector signed short);
5759 vector signed int vec_all_lt (vector unsigned short,
5760 vector unsigned short);
5761 vector signed int vec_all_lt (vector signed short, vector signed short);
5763 vector signed int vec_all_lt (vector signed int, vector unsigned int);
5764 vector signed int vec_all_lt (vector unsigned int, vector signed int);
5765 vector signed int vec_all_lt (vector unsigned int, vector unsigned int);
5767 vector signed int vec_all_lt (vector signed int, vector signed int);
5768 vector signed int vec_all_lt (vector float, vector float);
5770 vector signed int vec_all_nan (vector float);
5772 vector signed int vec_all_ne (vector signed char, vector unsigned char);
5774 vector signed int vec_all_ne (vector signed char, vector signed char);
5775 vector signed int vec_all_ne (vector unsigned char, vector signed char);
5777 vector signed int vec_all_ne (vector unsigned char,
5778 vector unsigned char);
5779 vector signed int vec_all_ne (vector signed short,
5780 vector unsigned short);
5781 vector signed int vec_all_ne (vector signed short, vector signed short);
5783 vector signed int vec_all_ne (vector unsigned short,
5784 vector signed short);
5785 vector signed int vec_all_ne (vector unsigned short,
5786 vector unsigned short);
5787 vector signed int vec_all_ne (vector signed int, vector unsigned int);
5788 vector signed int vec_all_ne (vector signed int, vector signed int);
5789 vector signed int vec_all_ne (vector unsigned int, vector signed int);
5790 vector signed int vec_all_ne (vector unsigned int, vector unsigned int);
5792 vector signed int vec_all_ne (vector float, vector float);
5794 vector signed int vec_all_nge (vector float, vector float);
5796 vector signed int vec_all_ngt (vector float, vector float);
5798 vector signed int vec_all_nle (vector float, vector float);
5800 vector signed int vec_all_nlt (vector float, vector float);
5802 vector signed int vec_all_numeric (vector float);
5804 vector signed int vec_any_eq (vector signed char, vector unsigned char);
5806 vector signed int vec_any_eq (vector signed char, vector signed char);
5807 vector signed int vec_any_eq (vector unsigned char, vector signed char);
5809 vector signed int vec_any_eq (vector unsigned char,
5810 vector unsigned char);
5811 vector signed int vec_any_eq (vector signed short,
5812 vector unsigned short);
5813 vector signed int vec_any_eq (vector signed short, vector signed short);
5815 vector signed int vec_any_eq (vector unsigned short,
5816 vector signed short);
5817 vector signed int vec_any_eq (vector unsigned short,
5818 vector unsigned short);
5819 vector signed int vec_any_eq (vector signed int, vector unsigned int);
5820 vector signed int vec_any_eq (vector signed int, vector signed int);
5821 vector signed int vec_any_eq (vector unsigned int, vector signed int);
5822 vector signed int vec_any_eq (vector unsigned int, vector unsigned int);
5824 vector signed int vec_any_eq (vector float, vector float);
5826 vector signed int vec_any_ge (vector signed char, vector unsigned char);
5828 vector signed int vec_any_ge (vector unsigned char, vector signed char);
5830 vector signed int vec_any_ge (vector unsigned char,
5831 vector unsigned char);
5832 vector signed int vec_any_ge (vector signed char, vector signed char);
5833 vector signed int vec_any_ge (vector signed short,
5834 vector unsigned short);
5835 vector signed int vec_any_ge (vector unsigned short,
5836 vector signed short);
5837 vector signed int vec_any_ge (vector unsigned short,
5838 vector unsigned short);
5839 vector signed int vec_any_ge (vector signed short, vector signed short);
5841 vector signed int vec_any_ge (vector signed int, vector unsigned int);
5842 vector signed int vec_any_ge (vector unsigned int, vector signed int);
5843 vector signed int vec_any_ge (vector unsigned int, vector unsigned int);
5845 vector signed int vec_any_ge (vector signed int, vector signed int);
5846 vector signed int vec_any_ge (vector float, vector float);
5848 vector signed int vec_any_gt (vector signed char, vector unsigned char);
5850 vector signed int vec_any_gt (vector unsigned char, vector signed char);
5852 vector signed int vec_any_gt (vector unsigned char,
5853 vector unsigned char);
5854 vector signed int vec_any_gt (vector signed char, vector signed char);
5855 vector signed int vec_any_gt (vector signed short,
5856 vector unsigned short);
5857 vector signed int vec_any_gt (vector unsigned short,
5858 vector signed short);
5859 vector signed int vec_any_gt (vector unsigned short,
5860 vector unsigned short);
5861 vector signed int vec_any_gt (vector signed short, vector signed short);
5863 vector signed int vec_any_gt (vector signed int, vector unsigned int);
5864 vector signed int vec_any_gt (vector unsigned int, vector signed int);
5865 vector signed int vec_any_gt (vector unsigned int, vector unsigned int);
5867 vector signed int vec_any_gt (vector signed int, vector signed int);
5868 vector signed int vec_any_gt (vector float, vector float);
5870 vector signed int vec_any_le (vector signed char, vector unsigned char);
5872 vector signed int vec_any_le (vector unsigned char, vector signed char);
5874 vector signed int vec_any_le (vector unsigned char,
5875 vector unsigned char);
5876 vector signed int vec_any_le (vector signed char, vector signed char);
5877 vector signed int vec_any_le (vector signed short,
5878 vector unsigned short);
5879 vector signed int vec_any_le (vector unsigned short,
5880 vector signed short);
5881 vector signed int vec_any_le (vector unsigned short,
5882 vector unsigned short);
5883 vector signed int vec_any_le (vector signed short, vector signed short);
5885 vector signed int vec_any_le (vector signed int, vector unsigned int);
5886 vector signed int vec_any_le (vector unsigned int, vector signed int);
5887 vector signed int vec_any_le (vector unsigned int, vector unsigned int);
5889 vector signed int vec_any_le (vector signed int, vector signed int);
5890 vector signed int vec_any_le (vector float, vector float);
5892 vector signed int vec_any_lt (vector signed char, vector unsigned char);
5894 vector signed int vec_any_lt (vector unsigned char, vector signed char);
5896 vector signed int vec_any_lt (vector unsigned char,
5897 vector unsigned char);
5898 vector signed int vec_any_lt (vector signed char, vector signed char);
5899 vector signed int vec_any_lt (vector signed short,
5900 vector unsigned short);
5901 vector signed int vec_any_lt (vector unsigned short,
5902 vector signed short);
5903 vector signed int vec_any_lt (vector unsigned short,
5904 vector unsigned short);
5905 vector signed int vec_any_lt (vector signed short, vector signed short);
5907 vector signed int vec_any_lt (vector signed int, vector unsigned int);
5908 vector signed int vec_any_lt (vector unsigned int, vector signed int);
5909 vector signed int vec_any_lt (vector unsigned int, vector unsigned int);
5911 vector signed int vec_any_lt (vector signed int, vector signed int);
5912 vector signed int vec_any_lt (vector float, vector float);
5914 vector signed int vec_any_nan (vector float);
5916 vector signed int vec_any_ne (vector signed char, vector unsigned char);
5918 vector signed int vec_any_ne (vector signed char, vector signed char);
5919 vector signed int vec_any_ne (vector unsigned char, vector signed char);
5921 vector signed int vec_any_ne (vector unsigned char,
5922 vector unsigned char);
5923 vector signed int vec_any_ne (vector signed short,
5924 vector unsigned short);
5925 vector signed int vec_any_ne (vector signed short, vector signed short);
5927 vector signed int vec_any_ne (vector unsigned short,
5928 vector signed short);
5929 vector signed int vec_any_ne (vector unsigned short,
5930 vector unsigned short);
5931 vector signed int vec_any_ne (vector signed int, vector unsigned int);
5932 vector signed int vec_any_ne (vector signed int, vector signed int);
5933 vector signed int vec_any_ne (vector unsigned int, vector signed int);
5934 vector signed int vec_any_ne (vector unsigned int, vector unsigned int);
5936 vector signed int vec_any_ne (vector float, vector float);
5938 vector signed int vec_any_nge (vector float, vector float);
5940 vector signed int vec_any_ngt (vector float, vector float);
5942 vector signed int vec_any_nle (vector float, vector float);
5944 vector signed int vec_any_nlt (vector float, vector float);
5946 vector signed int vec_any_numeric (vector float);
5948 vector signed int vec_any_out (vector float, vector float);
5952 @section Pragmas Accepted by GCC
5956 GCC supports several types of pragmas, primarily in order to compile
5957 code originally written for other compilers. Note that in general
5958 we do not recommend the use of pragmas; @xref{Function Attributes},
5959 for further explanation.
5969 @subsection ARM Pragmas
5971 The ARM target defines pragmas for controlling the default addition of
5972 @code{long_call} and @code{short_call} attributes to functions.
5973 @xref{Function Attributes}, for information about the effects of these
5978 @cindex pragma, long_calls
5979 Set all subsequent functions to have the @code{long_call} attribute.
5982 @cindex pragma, no_long_calls
5983 Set all subsequent functions to have the @code{short_call} attribute.
5985 @item long_calls_off
5986 @cindex pragma, long_calls_off
5987 Do not affect the @code{long_call} or @code{short_call} attributes of
5988 subsequent functions.
5991 @c Describe c4x pragmas here.
5992 @c Describe h8300 pragmas here.
5993 @c Describe i370 pragmas here.
5994 @c Describe i960 pragmas here.
5995 @c Describe sh pragmas here.
5996 @c Describe v850 pragmas here.
5998 @node Darwin Pragmas
5999 @subsection Darwin Pragmas
6001 The following pragmas are available for all architectures running the
6002 Darwin operating system. These are useful for compatibility with other
6006 @item mark @var{tokens}@dots{}
6007 @cindex pragma, mark
6008 This pragma is accepted, but has no effect.
6010 @item options align=@var{alignment}
6011 @cindex pragma, options align
6012 This pragma sets the alignment of fields in structures. The values of
6013 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
6014 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
6015 properly; to restore the previous setting, use @code{reset} for the
6018 @item segment @var{tokens}@dots{}
6019 @cindex pragma, segment
6020 This pragma is accepted, but has no effect.
6022 @item unused (@var{var} [, @var{var}]@dots{})
6023 @cindex pragma, unused
6024 This pragma declares variables to be possibly unused. GCC will not
6025 produce warnings for the listed variables. The effect is similar to
6026 that of the @code{unused} attribute, except that this pragma may appear
6027 anywhere within the variables' scopes.
6030 @node Solaris Pragmas
6031 @subsection Solaris Pragmas
6033 For compatibility with the SunPRO compiler, the following pragma
6037 @item redefine_extname @var{oldname} @var{newname}
6038 @cindex pragma, redefine_extname
6040 This pragma gives the C function @var{oldname} the assembler label
6041 @var{newname}. The pragma must appear before the function declaration.
6042 This pragma is equivalent to the asm labels extension (@pxref{Asm
6043 Labels}). The preprocessor defines @code{__PRAGMA_REDEFINE_EXTNAME}
6044 if the pragma is available.
6048 @subsection Tru64 Pragmas
6050 For compatibility with the Compaq C compiler, the following pragma
6054 @item extern_prefix @var{string}
6055 @cindex pragma, extern_prefix
6057 This pragma renames all subsequent function and variable declarations
6058 such that @var{string} is prepended to the name. This effect may be
6059 terminated by using another @code{extern_prefix} pragma with the
6062 This pragma is similar in intent to to the asm labels extension
6063 (@pxref{Asm Labels}) in that the system programmer wants to change
6064 the assembly-level ABI without changing the source-level API. The
6065 preprocessor defines @code{__EXTERN_PREFIX} if the pragma is available.
6068 @node Unnamed Fields
6069 @section Unnamed struct/union fields within structs/unions.
6073 For compatibility with other compilers, GCC allows you to define
6074 a structure or union that contains, as fields, structures and unions
6075 without names. For example:
6088 In this example, the user would be able to access members of the unnamed
6089 union with code like @samp{foo.b}. Note that only unnamed structs and
6090 unions are allowed, you may not have, for example, an unnamed
6093 You must never create such structures that cause ambiguous field definitions.
6094 For example, this structure:
6105 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
6106 Such constructs are not supported and must be avoided. In the future,
6107 such constructs may be detected and treated as compilation errors.
6109 @node C++ Extensions
6110 @chapter Extensions to the C++ Language
6111 @cindex extensions, C++ language
6112 @cindex C++ language extensions
6114 The GNU compiler provides these extensions to the C++ language (and you
6115 can also use most of the C language extensions in your C++ programs). If you
6116 want to write code that checks whether these features are available, you can
6117 test for the GNU compiler the same way as for C programs: check for a
6118 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
6119 test specifically for GNU C++ (@pxref{Standard Predefined,,Standard
6120 Predefined Macros,cpp.info,The C Preprocessor}).
6123 * Min and Max:: C++ Minimum and maximum operators.
6124 * Volatiles:: What constitutes an access to a volatile object.
6125 * Restricted Pointers:: C99 restricted pointers and references.
6126 * Vague Linkage:: Where G++ puts inlines, vtables and such.
6127 * C++ Interface:: You can use a single C++ header file for both
6128 declarations and definitions.
6129 * Template Instantiation:: Methods for ensuring that exactly one copy of
6130 each needed template instantiation is emitted.
6131 * Bound member functions:: You can extract a function pointer to the
6132 method denoted by a @samp{->*} or @samp{.*} expression.
6133 * C++ Attributes:: Variable, function, and type attributes for C++ only.
6134 * Java Exceptions:: Tweaking exception handling to work with Java.
6135 * Deprecated Features:: Things might disappear from g++.
6136 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
6140 @section Minimum and Maximum Operators in C++
6142 It is very convenient to have operators which return the ``minimum'' or the
6143 ``maximum'' of two arguments. In GNU C++ (but not in GNU C),
6146 @item @var{a} <? @var{b}
6148 @cindex minimum operator
6149 is the @dfn{minimum}, returning the smaller of the numeric values
6150 @var{a} and @var{b};
6152 @item @var{a} >? @var{b}
6154 @cindex maximum operator
6155 is the @dfn{maximum}, returning the larger of the numeric values @var{a}
6159 These operations are not primitive in ordinary C++, since you can
6160 use a macro to return the minimum of two things in C++, as in the
6164 #define MIN(X,Y) ((X) < (Y) ? : (X) : (Y))
6168 You might then use @w{@samp{int min = MIN (i, j);}} to set @var{min} to
6169 the minimum value of variables @var{i} and @var{j}.
6171 However, side effects in @code{X} or @code{Y} may cause unintended
6172 behavior. For example, @code{MIN (i++, j++)} will fail, incrementing
6173 the smaller counter twice. A GNU C extension allows you to write safe
6174 macros that avoid this kind of problem (@pxref{Naming Types,,Naming an
6175 Expression's Type}). However, writing @code{MIN} and @code{MAX} as
6176 macros also forces you to use function-call notation for a
6177 fundamental arithmetic operation. Using GNU C++ extensions, you can
6178 write @w{@samp{int min = i <? j;}} instead.
6180 Since @code{<?} and @code{>?} are built into the compiler, they properly
6181 handle expressions with side-effects; @w{@samp{int min = i++ <? j++;}}
6185 @section When is a Volatile Object Accessed?
6186 @cindex accessing volatiles
6187 @cindex volatile read
6188 @cindex volatile write
6189 @cindex volatile access
6191 Both the C and C++ standard have the concept of volatile objects. These
6192 are normally accessed by pointers and used for accessing hardware. The
6193 standards encourage compilers to refrain from optimizations
6194 concerning accesses to volatile objects that it might perform on
6195 non-volatile objects. The C standard leaves it implementation defined
6196 as to what constitutes a volatile access. The C++ standard omits to
6197 specify this, except to say that C++ should behave in a similar manner
6198 to C with respect to volatiles, where possible. The minimum either
6199 standard specifies is that at a sequence point all previous accesses to
6200 volatile objects have stabilized and no subsequent accesses have
6201 occurred. Thus an implementation is free to reorder and combine
6202 volatile accesses which occur between sequence points, but cannot do so
6203 for accesses across a sequence point. The use of volatiles does not
6204 allow you to violate the restriction on updating objects multiple times
6205 within a sequence point.
6207 In most expressions, it is intuitively obvious what is a read and what is
6208 a write. For instance
6211 volatile int *dst = @var{somevalue};
6212 volatile int *src = @var{someothervalue};
6217 will cause a read of the volatile object pointed to by @var{src} and stores the
6218 value into the volatile object pointed to by @var{dst}. There is no
6219 guarantee that these reads and writes are atomic, especially for objects
6220 larger than @code{int}.
6222 Less obvious expressions are where something which looks like an access
6223 is used in a void context. An example would be,
6226 volatile int *src = @var{somevalue};
6230 With C, such expressions are rvalues, and as rvalues cause a read of
6231 the object, GCC interprets this as a read of the volatile being pointed
6232 to. The C++ standard specifies that such expressions do not undergo
6233 lvalue to rvalue conversion, and that the type of the dereferenced
6234 object may be incomplete. The C++ standard does not specify explicitly
6235 that it is this lvalue to rvalue conversion which is responsible for
6236 causing an access. However, there is reason to believe that it is,
6237 because otherwise certain simple expressions become undefined. However,
6238 because it would surprise most programmers, G++ treats dereferencing a
6239 pointer to volatile object of complete type in a void context as a read
6240 of the object. When the object has incomplete type, G++ issues a
6245 struct T @{int m;@};
6246 volatile S *ptr1 = @var{somevalue};
6247 volatile T *ptr2 = @var{somevalue};
6252 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
6253 causes a read of the object pointed to. If you wish to force an error on
6254 the first case, you must force a conversion to rvalue with, for instance
6255 a static cast, @code{static_cast<S>(*ptr1)}.
6257 When using a reference to volatile, G++ does not treat equivalent
6258 expressions as accesses to volatiles, but instead issues a warning that
6259 no volatile is accessed. The rationale for this is that otherwise it
6260 becomes difficult to determine where volatile access occur, and not
6261 possible to ignore the return value from functions returning volatile
6262 references. Again, if you wish to force a read, cast the reference to
6265 @node Restricted Pointers
6266 @section Restricting Pointer Aliasing
6267 @cindex restricted pointers
6268 @cindex restricted references
6269 @cindex restricted this pointer
6271 As with gcc, g++ understands the C99 feature of restricted pointers,
6272 specified with the @code{__restrict__}, or @code{__restrict} type
6273 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
6274 language flag, @code{restrict} is not a keyword in C++.
6276 In addition to allowing restricted pointers, you can specify restricted
6277 references, which indicate that the reference is not aliased in the local
6281 void fn (int *__restrict__ rptr, int &__restrict__ rref)
6288 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
6289 @var{rref} refers to a (different) unaliased integer.
6291 You may also specify whether a member function's @var{this} pointer is
6292 unaliased by using @code{__restrict__} as a member function qualifier.
6295 void T::fn () __restrict__
6302 Within the body of @code{T::fn}, @var{this} will have the effective
6303 definition @code{T *__restrict__ const this}. Notice that the
6304 interpretation of a @code{__restrict__} member function qualifier is
6305 different to that of @code{const} or @code{volatile} qualifier, in that it
6306 is applied to the pointer rather than the object. This is consistent with
6307 other compilers which implement restricted pointers.
6309 As with all outermost parameter qualifiers, @code{__restrict__} is
6310 ignored in function definition matching. This means you only need to
6311 specify @code{__restrict__} in a function definition, rather than
6312 in a function prototype as well.
6315 @section Vague Linkage
6316 @cindex vague linkage
6318 There are several constructs in C++ which require space in the object
6319 file but are not clearly tied to a single translation unit. We say that
6320 these constructs have ``vague linkage''. Typically such constructs are
6321 emitted wherever they are needed, though sometimes we can be more
6325 @item Inline Functions
6326 Inline functions are typically defined in a header file which can be
6327 included in many different compilations. Hopefully they can usually be
6328 inlined, but sometimes an out-of-line copy is necessary, if the address
6329 of the function is taken or if inlining fails. In general, we emit an
6330 out-of-line copy in all translation units where one is needed. As an
6331 exception, we only emit inline virtual functions with the vtable, since
6332 it will always require a copy.
6334 Local static variables and string constants used in an inline function
6335 are also considered to have vague linkage, since they must be shared
6336 between all inlined and out-of-line instances of the function.
6340 C++ virtual functions are implemented in most compilers using a lookup
6341 table, known as a vtable. The vtable contains pointers to the virtual
6342 functions provided by a class, and each object of the class contains a
6343 pointer to its vtable (or vtables, in some multiple-inheritance
6344 situations). If the class declares any non-inline, non-pure virtual
6345 functions, the first one is chosen as the ``key method'' for the class,
6346 and the vtable is only emitted in the translation unit where the key
6349 @emph{Note:} If the chosen key method is later defined as inline, the
6350 vtable will still be emitted in every translation unit which defines it.
6351 Make sure that any inline virtuals are declared inline in the class
6352 body, even if they are not defined there.
6354 @item type_info objects
6357 C++ requires information about types to be written out in order to
6358 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
6359 For polymorphic classes (classes with virtual functions), the type_info
6360 object is written out along with the vtable so that @samp{dynamic_cast}
6361 can determine the dynamic type of a class object at runtime. For all
6362 other types, we write out the type_info object when it is used: when
6363 applying @samp{typeid} to an expression, throwing an object, or
6364 referring to a type in a catch clause or exception specification.
6366 @item Template Instantiations
6367 Most everything in this section also applies to template instantiations,
6368 but there are other options as well.
6369 @xref{Template Instantiation,,Where's the Template?}.
6373 When used with GNU ld version 2.8 or later on an ELF system such as
6374 Linux/GNU or Solaris 2, or on Microsoft Windows, duplicate copies of
6375 these constructs will be discarded at link time. This is known as
6378 On targets that don't support COMDAT, but do support weak symbols, GCC
6379 will use them. This way one copy will override all the others, but
6380 the unused copies will still take up space in the executable.
6382 For targets which do not support either COMDAT or weak symbols,
6383 most entities with vague linkage will be emitted as local symbols to
6384 avoid duplicate definition errors from the linker. This will not happen
6385 for local statics in inlines, however, as having multiple copies will
6386 almost certainly break things.
6388 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
6389 another way to control placement of these constructs.
6392 @section Declarations and Definitions in One Header
6394 @cindex interface and implementation headers, C++
6395 @cindex C++ interface and implementation headers
6396 C++ object definitions can be quite complex. In principle, your source
6397 code will need two kinds of things for each object that you use across
6398 more than one source file. First, you need an @dfn{interface}
6399 specification, describing its structure with type declarations and
6400 function prototypes. Second, you need the @dfn{implementation} itself.
6401 It can be tedious to maintain a separate interface description in a
6402 header file, in parallel to the actual implementation. It is also
6403 dangerous, since separate interface and implementation definitions may
6404 not remain parallel.
6406 @cindex pragmas, interface and implementation
6407 With GNU C++, you can use a single header file for both purposes.
6410 @emph{Warning:} The mechanism to specify this is in transition. For the
6411 nonce, you must use one of two @code{#pragma} commands; in a future
6412 release of GNU C++, an alternative mechanism will make these
6413 @code{#pragma} commands unnecessary.
6416 The header file contains the full definitions, but is marked with
6417 @samp{#pragma interface} in the source code. This allows the compiler
6418 to use the header file only as an interface specification when ordinary
6419 source files incorporate it with @code{#include}. In the single source
6420 file where the full implementation belongs, you can use either a naming
6421 convention or @samp{#pragma implementation} to indicate this alternate
6422 use of the header file.
6425 @item #pragma interface
6426 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
6427 @kindex #pragma interface
6428 Use this directive in @emph{header files} that define object classes, to save
6429 space in most of the object files that use those classes. Normally,
6430 local copies of certain information (backup copies of inline member
6431 functions, debugging information, and the internal tables that implement
6432 virtual functions) must be kept in each object file that includes class
6433 definitions. You can use this pragma to avoid such duplication. When a
6434 header file containing @samp{#pragma interface} is included in a
6435 compilation, this auxiliary information will not be generated (unless
6436 the main input source file itself uses @samp{#pragma implementation}).
6437 Instead, the object files will contain references to be resolved at link
6440 The second form of this directive is useful for the case where you have
6441 multiple headers with the same name in different directories. If you
6442 use this form, you must specify the same string to @samp{#pragma
6445 @item #pragma implementation
6446 @itemx #pragma implementation "@var{objects}.h"
6447 @kindex #pragma implementation
6448 Use this pragma in a @emph{main input file}, when you want full output from
6449 included header files to be generated (and made globally visible). The
6450 included header file, in turn, should use @samp{#pragma interface}.
6451 Backup copies of inline member functions, debugging information, and the
6452 internal tables used to implement virtual functions are all generated in
6453 implementation files.
6455 @cindex implied @code{#pragma implementation}
6456 @cindex @code{#pragma implementation}, implied
6457 @cindex naming convention, implementation headers
6458 If you use @samp{#pragma implementation} with no argument, it applies to
6459 an include file with the same basename@footnote{A file's @dfn{basename}
6460 was the name stripped of all leading path information and of trailing
6461 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
6462 file. For example, in @file{allclass.cc}, giving just
6463 @samp{#pragma implementation}
6464 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
6466 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
6467 an implementation file whenever you would include it from
6468 @file{allclass.cc} even if you never specified @samp{#pragma
6469 implementation}. This was deemed to be more trouble than it was worth,
6470 however, and disabled.
6472 If you use an explicit @samp{#pragma implementation}, it must appear in
6473 your source file @emph{before} you include the affected header files.
6475 Use the string argument if you want a single implementation file to
6476 include code from multiple header files. (You must also use
6477 @samp{#include} to include the header file; @samp{#pragma
6478 implementation} only specifies how to use the file---it doesn't actually
6481 There is no way to split up the contents of a single header file into
6482 multiple implementation files.
6485 @cindex inlining and C++ pragmas
6486 @cindex C++ pragmas, effect on inlining
6487 @cindex pragmas in C++, effect on inlining
6488 @samp{#pragma implementation} and @samp{#pragma interface} also have an
6489 effect on function inlining.
6491 If you define a class in a header file marked with @samp{#pragma
6492 interface}, the effect on a function defined in that class is similar to
6493 an explicit @code{extern} declaration---the compiler emits no code at
6494 all to define an independent version of the function. Its definition
6495 is used only for inlining with its callers.
6497 @opindex fno-implement-inlines
6498 Conversely, when you include the same header file in a main source file
6499 that declares it as @samp{#pragma implementation}, the compiler emits
6500 code for the function itself; this defines a version of the function
6501 that can be found via pointers (or by callers compiled without
6502 inlining). If all calls to the function can be inlined, you can avoid
6503 emitting the function by compiling with @option{-fno-implement-inlines}.
6504 If any calls were not inlined, you will get linker errors.
6506 @node Template Instantiation
6507 @section Where's the Template?
6509 @cindex template instantiation
6511 C++ templates are the first language feature to require more
6512 intelligence from the environment than one usually finds on a UNIX
6513 system. Somehow the compiler and linker have to make sure that each
6514 template instance occurs exactly once in the executable if it is needed,
6515 and not at all otherwise. There are two basic approaches to this
6516 problem, which I will refer to as the Borland model and the Cfront model.
6520 Borland C++ solved the template instantiation problem by adding the code
6521 equivalent of common blocks to their linker; the compiler emits template
6522 instances in each translation unit that uses them, and the linker
6523 collapses them together. The advantage of this model is that the linker
6524 only has to consider the object files themselves; there is no external
6525 complexity to worry about. This disadvantage is that compilation time
6526 is increased because the template code is being compiled repeatedly.
6527 Code written for this model tends to include definitions of all
6528 templates in the header file, since they must be seen to be
6532 The AT&T C++ translator, Cfront, solved the template instantiation
6533 problem by creating the notion of a template repository, an
6534 automatically maintained place where template instances are stored. A
6535 more modern version of the repository works as follows: As individual
6536 object files are built, the compiler places any template definitions and
6537 instantiations encountered in the repository. At link time, the link
6538 wrapper adds in the objects in the repository and compiles any needed
6539 instances that were not previously emitted. The advantages of this
6540 model are more optimal compilation speed and the ability to use the
6541 system linker; to implement the Borland model a compiler vendor also
6542 needs to replace the linker. The disadvantages are vastly increased
6543 complexity, and thus potential for error; for some code this can be
6544 just as transparent, but in practice it can been very difficult to build
6545 multiple programs in one directory and one program in multiple
6546 directories. Code written for this model tends to separate definitions
6547 of non-inline member templates into a separate file, which should be
6548 compiled separately.
6551 When used with GNU ld version 2.8 or later on an ELF system such as
6552 Linux/GNU or Solaris 2, or on Microsoft Windows, g++ supports the
6553 Borland model. On other systems, g++ implements neither automatic
6556 A future version of g++ will support a hybrid model whereby the compiler
6557 will emit any instantiations for which the template definition is
6558 included in the compile, and store template definitions and
6559 instantiation context information into the object file for the rest.
6560 The link wrapper will extract that information as necessary and invoke
6561 the compiler to produce the remaining instantiations. The linker will
6562 then combine duplicate instantiations.
6564 In the mean time, you have the following options for dealing with
6565 template instantiations:
6570 Compile your template-using code with @option{-frepo}. The compiler will
6571 generate files with the extension @samp{.rpo} listing all of the
6572 template instantiations used in the corresponding object files which
6573 could be instantiated there; the link wrapper, @samp{collect2}, will
6574 then update the @samp{.rpo} files to tell the compiler where to place
6575 those instantiations and rebuild any affected object files. The
6576 link-time overhead is negligible after the first pass, as the compiler
6577 will continue to place the instantiations in the same files.
6579 This is your best option for application code written for the Borland
6580 model, as it will just work. Code written for the Cfront model will
6581 need to be modified so that the template definitions are available at
6582 one or more points of instantiation; usually this is as simple as adding
6583 @code{#include <tmethods.cc>} to the end of each template header.
6585 For library code, if you want the library to provide all of the template
6586 instantiations it needs, just try to link all of its object files
6587 together; the link will fail, but cause the instantiations to be
6588 generated as a side effect. Be warned, however, that this may cause
6589 conflicts if multiple libraries try to provide the same instantiations.
6590 For greater control, use explicit instantiation as described in the next
6594 @opindex fno-implicit-templates
6595 Compile your code with @option{-fno-implicit-templates} to disable the
6596 implicit generation of template instances, and explicitly instantiate
6597 all the ones you use. This approach requires more knowledge of exactly
6598 which instances you need than do the others, but it's less
6599 mysterious and allows greater control. You can scatter the explicit
6600 instantiations throughout your program, perhaps putting them in the
6601 translation units where the instances are used or the translation units
6602 that define the templates themselves; you can put all of the explicit
6603 instantiations you need into one big file; or you can create small files
6610 template class Foo<int>;
6611 template ostream& operator <<
6612 (ostream&, const Foo<int>&);
6615 for each of the instances you need, and create a template instantiation
6618 If you are using Cfront-model code, you can probably get away with not
6619 using @option{-fno-implicit-templates} when compiling files that don't
6620 @samp{#include} the member template definitions.
6622 If you use one big file to do the instantiations, you may want to
6623 compile it without @option{-fno-implicit-templates} so you get all of the
6624 instances required by your explicit instantiations (but not by any
6625 other files) without having to specify them as well.
6627 g++ has extended the template instantiation syntax outlined in the
6628 Working Paper to allow forward declaration of explicit instantiations
6629 (with @code{extern}), instantiation of the compiler support data for a
6630 template class (i.e.@: the vtable) without instantiating any of its
6631 members (with @code{inline}), and instantiation of only the static data
6632 members of a template class, without the support data or member
6633 functions (with (@code{static}):
6636 extern template int max (int, int);
6637 inline template class Foo<int>;
6638 static template class Foo<int>;
6642 Do nothing. Pretend g++ does implement automatic instantiation
6643 management. Code written for the Borland model will work fine, but
6644 each translation unit will contain instances of each of the templates it
6645 uses. In a large program, this can lead to an unacceptable amount of code
6649 @opindex fexternal-templates
6650 Add @samp{#pragma interface} to all files containing template
6651 definitions. For each of these files, add @samp{#pragma implementation
6652 "@var{filename}"} to the top of some @samp{.C} file which
6653 @samp{#include}s it. Then compile everything with
6654 @option{-fexternal-templates}. The templates will then only be expanded
6655 in the translation unit which implements them (i.e.@: has a @samp{#pragma
6656 implementation} line for the file where they live); all other files will
6657 use external references. If you're lucky, everything should work
6658 properly. If you get undefined symbol errors, you need to make sure
6659 that each template instance which is used in the program is used in the
6660 file which implements that template. If you don't have any use for a
6661 particular instance in that file, you can just instantiate it
6662 explicitly, using the syntax from the latest C++ working paper:
6665 template class A<int>;
6666 template ostream& operator << (ostream&, const A<int>&);
6669 This strategy will work with code written for either model. If you are
6670 using code written for the Cfront model, the file containing a class
6671 template and the file containing its member templates should be
6672 implemented in the same translation unit.
6675 @opindex falt-external-templates
6676 A slight variation on this approach is to use the flag
6677 @option{-falt-external-templates} instead. This flag causes template
6678 instances to be emitted in the translation unit that implements the
6679 header where they are first instantiated, rather than the one which
6680 implements the file where the templates are defined. This header must
6681 be the same in all translation units, or things are likely to break.
6683 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
6684 more discussion of these pragmas.
6687 @node Bound member functions
6688 @section Extracting the function pointer from a bound pointer to member function
6691 @cindex pointer to member function
6692 @cindex bound pointer to member function
6694 In C++, pointer to member functions (PMFs) are implemented using a wide
6695 pointer of sorts to handle all the possible call mechanisms; the PMF
6696 needs to store information about how to adjust the @samp{this} pointer,
6697 and if the function pointed to is virtual, where to find the vtable, and
6698 where in the vtable to look for the member function. If you are using
6699 PMFs in an inner loop, you should really reconsider that decision. If
6700 that is not an option, you can extract the pointer to the function that
6701 would be called for a given object/PMF pair and call it directly inside
6702 the inner loop, to save a bit of time.
6704 Note that you will still be paying the penalty for the call through a
6705 function pointer; on most modern architectures, such a call defeats the
6706 branch prediction features of the CPU@. This is also true of normal
6707 virtual function calls.
6709 The syntax for this extension is
6713 extern int (A::*fp)();
6714 typedef int (*fptr)(A *);
6716 fptr p = (fptr)(a.*fp);
6719 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
6720 no object is needed to obtain the address of the function. They can be
6721 converted to function pointers directly:
6724 fptr p1 = (fptr)(&A::foo);
6727 @opindex Wno-pmf-conversions
6728 You must specify @option{-Wno-pmf-conversions} to use this extension.
6730 @node C++ Attributes
6731 @section C++-Specific Variable, Function, and Type Attributes
6733 Some attributes only make sense for C++ programs.
6736 @item init_priority (@var{priority})
6737 @cindex init_priority attribute
6740 In Standard C++, objects defined at namespace scope are guaranteed to be
6741 initialized in an order in strict accordance with that of their definitions
6742 @emph{in a given translation unit}. No guarantee is made for initializations
6743 across translation units. However, GNU C++ allows users to control the
6744 order of initialization of objects defined at namespace scope with the
6745 @code{init_priority} attribute by specifying a relative @var{priority},
6746 a constant integral expression currently bounded between 101 and 65535
6747 inclusive. Lower numbers indicate a higher priority.
6749 In the following example, @code{A} would normally be created before
6750 @code{B}, but the @code{init_priority} attribute has reversed that order:
6753 Some_Class A __attribute__ ((init_priority (2000)));
6754 Some_Class B __attribute__ ((init_priority (543)));
6758 Note that the particular values of @var{priority} do not matter; only their
6761 @item java_interface
6762 @cindex java_interface attribute
6764 This type attribute informs C++ that the class is a Java interface. It may
6765 only be applied to classes declared within an @code{extern "Java"} block.
6766 Calls to methods declared in this interface will be dispatched using GCJ's
6767 interface table mechanism, instead of regular virtual table dispatch.
6771 @node Java Exceptions
6772 @section Java Exceptions
6774 The Java language uses a slightly different exception handling model
6775 from C++. Normally, GNU C++ will automatically detect when you are
6776 writing C++ code that uses Java exceptions, and handle them
6777 appropriately. However, if C++ code only needs to execute destructors
6778 when Java exceptions are thrown through it, GCC will guess incorrectly.
6779 Sample problematic code is:
6782 struct S @{ ~S(); @};
6783 extern void bar(); // is written in Java, and may throw exceptions
6792 The usual effect of an incorrect guess is a link failure, complaining of
6793 a missing routine called @samp{__gxx_personality_v0}.
6795 You can inform the compiler that Java exceptions are to be used in a
6796 translation unit, irrespective of what it might think, by writing
6797 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
6798 @samp{#pragma} must appear before any functions that throw or catch
6799 exceptions, or run destructors when exceptions are thrown through them.
6801 You cannot mix Java and C++ exceptions in the same translation unit. It
6802 is believed to be safe to throw a C++ exception from one file through
6803 another file compiled for the Java exception model, or vice versa, but
6804 there may be bugs in this area.
6806 @node Deprecated Features
6807 @section Deprecated Features
6809 In the past, the GNU C++ compiler was extended to experiment with new
6810 features, at a time when the C++ language was still evolving. Now that
6811 the C++ standard is complete, some of those features are superseded by
6812 superior alternatives. Using the old features might cause a warning in
6813 some cases that the feature will be dropped in the future. In other
6814 cases, the feature might be gone already.
6816 While the list below is not exhaustive, it documents some of the options
6817 that are now deprecated:
6820 @item -fexternal-templates
6821 @itemx -falt-external-templates
6822 These are two of the many ways for g++ to implement template
6823 instantiation. @xref{Template Instantiation}. The C++ standard clearly
6824 defines how template definitions have to be organized across
6825 implementation units. g++ has an implicit instantiation mechanism that
6826 should work just fine for standard-conforming code.
6828 @item -fstrict-prototype
6829 @itemx -fno-strict-prototype
6830 Previously it was possible to use an empty prototype parameter list to
6831 indicate an unspecified number of parameters (like C), rather than no
6832 parameters, as C++ demands. This feature has been removed, except where
6833 it is required for backwards compatibility @xref{Backwards Compatibility}.
6836 The named return value extension has been deprecated, and is now
6839 The use of initializer lists with new expressions has been deprecated,
6840 and is now removed from g++.
6842 Floating and complex non-type template parameters have been deprecated,
6843 and are now removed from g++.
6845 The implicit typename extension has been deprecated and will be removed
6846 from g++ at some point. In some cases g++ determines that a dependant
6847 type such as @code{TPL<T>::X} is a type without needing a
6848 @code{typename} keyword, contrary to the standard.
6850 @node Backwards Compatibility
6851 @section Backwards Compatibility
6852 @cindex Backwards Compatibility
6853 @cindex ARM [Annotated C++ Reference Manual]
6855 Now that there is a definitive ISO standard C++, G++ has a specification
6856 to adhere to. The C++ language evolved over time, and features that
6857 used to be acceptable in previous drafts of the standard, such as the ARM
6858 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
6859 compilation of C++ written to such drafts, G++ contains some backwards
6860 compatibilities. @emph{All such backwards compatibility features are
6861 liable to disappear in future versions of G++.} They should be considered
6862 deprecated @xref{Deprecated Features}.
6866 If a variable is declared at for scope, it used to remain in scope until
6867 the end of the scope which contained the for statement (rather than just
6868 within the for scope). G++ retains this, but issues a warning, if such a
6869 variable is accessed outside the for scope.
6871 @item Implicit C language
6872 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
6873 scope to set the language. On such systems, all header files are
6874 implicitly scoped inside a C language scope. Also, an empty prototype
6875 @code{()} will be treated as an unspecified number of arguments, rather
6876 than no arguments, as C++ demands.