3 @settitle The C Preprocessor
9 @include gcc-common.texi
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1987-2023 Free Software Foundation, Inc.
15 Permission is granted to copy, distribute and/or modify this document
16 under the terms of the GNU Free Documentation License, Version 1.3 or
17 any later version published by the Free Software Foundation. A copy of
18 the license is included in the
20 section entitled ``GNU Free Documentation License''.
22 @c man begin COPYRIGHT
27 @c man begin COPYRIGHT
28 This manual contains no Invariant Sections. The Front-Cover Texts are
29 (a) (see below), and the Back-Cover Texts are (b) (see below).
31 (a) The FSF's Front-Cover Text is:
35 (b) The FSF's Back-Cover Text is:
37 You have freedom to copy and modify this GNU Manual, like GNU
38 software. Copies published by the Free Software Foundation raise
39 funds for GNU development.
43 @c Create a separate index for command line options.
47 @c Used in cppopts.texi and cppenv.texi.
51 @dircategory Software development
53 * Cpp: (cpp). The GNU C preprocessor.
58 @title The C Preprocessor
60 @author Richard M. Stallman, Zachary Weinberg
62 @c There is a fill at the bottom of the page, so we need a filll to
64 @vskip 0pt plus 1filll
73 The C preprocessor implements the macro language used to transform C,
74 C++, and Objective-C programs before they are compiled. It can also be
86 * Preprocessor Output::
88 * Implementation Details::
90 * Environment Variables::
91 * GNU Free Documentation License::
92 * Index of Directives::
97 --- The Detailed Node Listing ---
102 * Initial processing::
104 * The preprocessing language::
109 * Include Operation::
111 * Once-Only Headers::
112 * Alternatives to Wrapper #ifndef::
113 * Computed Includes::
119 * Object-like Macros::
120 * Function-like Macros::
125 * Predefined Macros::
126 * Undefining and Redefining Macros::
127 * Directives Within Macro Arguments::
132 * Standard Predefined Macros::
133 * Common Predefined Macros::
134 * System-specific Predefined Macros::
135 * C++ Named Operators::
140 * Operator Precedence Problems::
141 * Swallowing the Semicolon::
142 * Duplication of Side Effects::
143 * Self-Referential Macros::
145 * Newlines in Arguments::
150 * Conditional Syntax::
161 Implementation Details
163 * Implementation-defined behavior::
164 * Implementation limits::
165 * Obsolete Features::
169 * Obsolete Features::
179 @c man begin DESCRIPTION
180 The C preprocessor, often known as @dfn{cpp}, is a @dfn{macro processor}
181 that is used automatically by the C compiler to transform your program
182 before compilation. It is called a macro processor because it allows
183 you to define @dfn{macros}, which are brief abbreviations for longer
186 The C preprocessor is intended to be used only with C, C++, and
187 Objective-C source code. In the past, it has been abused as a general
188 text processor. It will choke on input which does not obey C's lexical
189 rules. For example, apostrophes will be interpreted as the beginning of
190 character constants, and cause errors. Also, you cannot rely on it
191 preserving characteristics of the input which are not significant to
192 C-family languages. If a Makefile is preprocessed, all the hard tabs
193 will be removed, and the Makefile will not work.
195 Having said that, you can often get away with using cpp on things which
196 are not C@. Other Algol-ish programming languages are often safe
197 (Ada, etc.) So is assembly, with caution. @option{-traditional-cpp}
198 mode preserves more white space, and is otherwise more permissive. Many
199 of the problems can be avoided by writing C or C++ style comments
200 instead of native language comments, and keeping macros simple.
202 Wherever possible, you should use a preprocessor geared to the language
203 you are writing in. Modern versions of the GNU assembler have macro
204 facilities. Most high level programming languages have their own
205 conditional compilation and inclusion mechanism. If all else fails,
206 try a true general text processor, such as GNU M4.
208 C preprocessors vary in some details. This manual discusses the GNU C
209 preprocessor, which provides a small superset of the features of ISO
210 Standard C@. In its default mode, the GNU C preprocessor does not do a
211 few things required by the standard. These are features which are
212 rarely, if ever, used, and may cause surprising changes to the meaning
213 of a program which does not expect them. To get strict ISO Standard C,
214 you should use the @option{-std=c90}, @option{-std=c99},
215 @option{-std=c11} or @option{-std=c17} options, depending
216 on which version of the standard you want. To get all the mandatory
217 diagnostics, you must also use @option{-pedantic}. @xref{Invocation}.
219 This manual describes the behavior of the ISO preprocessor. To
220 minimize gratuitous differences, where the ISO preprocessor's
221 behavior does not conflict with traditional semantics, the
222 traditional preprocessor should behave the same way. The various
223 differences that do exist are detailed in the section @ref{Traditional
226 For clarity, unless noted otherwise, references to @samp{CPP} in this
227 manual refer to GNU CPP@.
232 * Initial processing::
234 * The preprocessing language::
238 @section Character sets
240 Source code character set processing in C and related languages is
241 rather complicated. The C standard discusses two character sets, but
242 there are really at least four.
244 The files input to CPP might be in any character set at all. CPP's
245 very first action, before it even looks for line boundaries, is to
246 convert the file into the character set it uses for internal
247 processing. That set is what the C standard calls the @dfn{source}
248 character set. It must be isomorphic with ISO 10646, also known as
249 Unicode. CPP uses the UTF-8 encoding of Unicode.
251 The character sets of the input files are specified using the
252 @option{-finput-charset=} option.
254 All preprocessing work (the subject of the rest of this manual) is
255 carried out in the source character set. If you request textual
256 output from the preprocessor with the @option{-E} option, it will be
259 After preprocessing is complete, string and character constants are
260 converted again, into the @dfn{execution} character set. This
261 character set is under control of the user; the default is UTF-8,
262 matching the source character set. Wide string and character
263 constants have their own character set, which is not called out
264 specifically in the standard. Again, it is under control of the user.
265 The default is UTF-16 or UTF-32, whichever fits in the target's
266 @code{wchar_t} type, in the target machine's byte
267 order.@footnote{UTF-16 does not meet the requirements of the C
268 standard for a wide character set, but the choice of 16-bit
269 @code{wchar_t} is enshrined in some system ABIs so we cannot fix
270 this.} Octal and hexadecimal escape sequences do not undergo
271 conversion; @t{'\x12'} has the value 0x12 regardless of the currently
272 selected execution character set. All other escapes are replaced by
273 the character in the source character set that they represent, then
274 converted to the execution character set, just like unescaped
277 In identifiers, characters outside the ASCII range can be specified
278 with the @samp{\u} and @samp{\U} escapes or used directly in the input
279 encoding. If strict ISO C90 conformance is specified with an option
280 such as @option{-std=c90}, or @option{-fno-extended-identifiers} is
281 used, then those constructs are not permitted in identifiers.
283 @node Initial processing
284 @section Initial processing
286 The preprocessor performs a series of textual transformations on its
287 input. These happen before all other processing. Conceptually, they
288 happen in a rigid order, and the entire file is run through each
289 transformation before the next one begins. CPP actually does them
290 all at once, for performance reasons. These transformations correspond
291 roughly to the first three ``phases of translation'' described in the C
297 The input file is read into memory and broken into lines.
299 Different systems use different conventions to indicate the end of a
300 line. GCC accepts the ASCII control sequences @kbd{LF}, @kbd{@w{CR
301 LF}} and @kbd{CR} as end-of-line markers. These are the canonical
302 sequences used by Unix, DOS and VMS, and the classic Mac OS (before
303 OSX) respectively. You may therefore safely copy source code written
304 on any of those systems to a different one and use it without
305 conversion. (GCC may lose track of the current line number if a file
306 doesn't consistently use one convention, as sometimes happens when it
307 is edited on computers with different conventions that share a network
310 If the last line of any input file lacks an end-of-line marker, the end
311 of the file is considered to implicitly supply one. The C standard says
312 that this condition provokes undefined behavior, so GCC will emit a
317 @anchor{trigraphs}If trigraphs are enabled, they are replaced by their
318 corresponding single characters. By default GCC ignores trigraphs,
319 but if you request a strictly conforming mode with the @option{-std}
320 option, or you specify the @option{-trigraphs} option, then it
323 These are nine three-character sequences, all starting with @samp{??},
324 that are defined by ISO C to stand for single characters. They permit
325 obsolete systems that lack some of C's punctuation to use C@. For
326 example, @samp{??/} stands for @samp{\}, so @t{'??/n'} is a character
327 constant for a newline.
329 Trigraphs are not popular and many compilers implement them
330 incorrectly. Portable code should not rely on trigraphs being either
331 converted or ignored. With @option{-Wtrigraphs} GCC will warn you
332 when a trigraph may change the meaning of your program if it were
333 converted. @xref{Wtrigraphs}.
335 In a string constant, you can prevent a sequence of question marks
336 from being confused with a trigraph by inserting a backslash between
337 the question marks, or by separating the string literal at the
338 trigraph and making use of string literal concatenation. @t{"(??\?)"}
339 is the string @samp{(???)}, not @samp{(?]}. Traditional C compilers
340 do not recognize these idioms.
342 The nine trigraphs and their replacements are
345 Trigraph: ??( ??) ??< ??> ??= ??/ ??' ??! ??-
346 Replacement: [ ] @{ @} # \ ^ | ~
349 @cindex continued lines
350 @cindex backslash-newline
352 Continued lines are merged into one long line.
354 A continued line is a line which ends with a backslash, @samp{\}. The
355 backslash is removed and the following line is joined with the current
356 one. No space is inserted, so you may split a line anywhere, even in
357 the middle of a word. (It is generally more readable to split lines
358 only at white space.)
360 The trailing backslash on a continued line is commonly referred to as a
361 @dfn{backslash-newline}.
363 If there is white space between a backslash and the end of a line, that
364 is still a continued line. However, as this is usually the result of an
365 editing mistake, and many compilers will not accept it as a continued
366 line, GCC will warn you about it.
369 @cindex line comments
370 @cindex block comments
372 All comments are replaced with single spaces.
374 There are two kinds of comments. @dfn{Block comments} begin with
375 @samp{/*} and continue until the next @samp{*/}. Block comments do not
379 /* @r{this is} /* @r{one comment} */ @r{text outside comment}
382 @dfn{Line comments} begin with @samp{//} and continue to the end of the
383 current line. Line comments do not nest either, but it does not matter,
384 because they would end in the same place anyway.
387 // @r{this is} // @r{one comment}
388 @r{text outside comment}
392 It is safe to put line comments inside block comments, or vice versa.
397 // @r{contains line comment}
399 */ @r{outside comment}
401 // @r{line comment} /* @r{contains block comment} */
405 But beware of commenting out one end of a block comment with a line
410 // @r{l.c.} /* @r{block comment begins}
411 @r{oops! this isn't a comment anymore} */
415 Comments are not recognized within string literals.
416 @t{@w{"/* blah */"}} is the string constant @samp{@w{/* blah */}}, not
419 Line comments are not in the 1989 edition of the C standard, but they
420 are recognized by GCC as an extension. In C++ and in the 1999 edition
421 of the C standard, they are an official part of the language.
423 Since these transformations happen before all other processing, you can
424 split a line mechanically with backslash-newline anywhere. You can
425 comment out the end of a line. You can continue a line comment onto the
426 next line with backslash-newline. You can even split @samp{/*},
427 @samp{*/}, and @samp{//} onto multiple lines with backslash-newline.
443 is equivalent to @code{@w{#define FOO 1020}}. All these tricks are
444 extremely confusing and should not be used in code intended to be
447 There is no way to prevent a backslash at the end of a line from being
448 interpreted as a backslash-newline. This cannot affect any correct
452 @section Tokenization
455 @cindex preprocessing tokens
456 After the textual transformations are finished, the input file is
457 converted into a sequence of @dfn{preprocessing tokens}. These mostly
458 correspond to the syntactic tokens used by the C compiler, but there are
459 a few differences. White space separates tokens; it is not itself a
460 token of any kind. Tokens do not have to be separated by white space,
461 but it is often necessary to avoid ambiguities.
463 When faced with a sequence of characters that has more than one possible
464 tokenization, the preprocessor is greedy. It always makes each token,
465 starting from the left, as big as possible before moving on to the next
466 token. For instance, @code{a+++++b} is interpreted as
467 @code{@w{a ++ ++ + b}}, not as @code{@w{a ++ + ++ b}}, even though the
468 latter tokenization could be part of a valid C program and the former
471 Once the input file is broken into tokens, the token boundaries never
472 change, except when the @samp{##} preprocessing operator is used to paste
473 tokens together. @xref{Concatenation}. For example,
485 The compiler does not re-tokenize the preprocessor's output. Each
486 preprocessing token becomes one compiler token.
489 Preprocessing tokens fall into five broad classes: identifiers,
490 preprocessing numbers, string literals, punctuators, and other. An
491 @dfn{identifier} is the same as an identifier in C: any sequence of
492 letters, digits, or underscores, which begins with a letter or
493 underscore. Keywords of C have no significance to the preprocessor;
494 they are ordinary identifiers. You can define a macro whose name is a
495 keyword, for instance. The only identifier which can be considered a
496 preprocessing keyword is @code{defined}. @xref{Defined}.
498 This is mostly true of other languages which use the C preprocessor.
499 However, a few of the keywords of C++ are significant even in the
500 preprocessor. @xref{C++ Named Operators}.
502 In the 1999 C standard, identifiers may contain letters which are not
503 part of the ``basic source character set'', at the implementation's
504 discretion (such as accented Latin letters, Greek letters, or Chinese
505 ideograms). This may be done with an extended character set, or the
506 @samp{\u} and @samp{\U} escape sequences.
508 As an extension, GCC treats @samp{$} as a letter. This is for
509 compatibility with some systems, such as VMS, where @samp{$} is commonly
510 used in system-defined function and object names. @samp{$} is not a
511 letter in strictly conforming mode, or if you specify the @option{-$}
512 option. @xref{Invocation}.
515 @cindex preprocessing numbers
516 A @dfn{preprocessing number} has a rather bizarre definition. The
517 category includes all the normal integer and floating point constants
518 one expects of C, but also a number of other things one might not
519 initially recognize as a number. Formally, preprocessing numbers begin
520 with an optional period, a required decimal digit, and then continue
521 with any sequence of letters, digits, underscores, periods, and
522 exponents. Exponents are the two-character sequences @samp{e+},
523 @samp{e-}, @samp{E+}, @samp{E-}, @samp{p+}, @samp{p-}, @samp{P+}, and
524 @samp{P-}. (The exponents that begin with @samp{p} or @samp{P} are
525 used for hexadecimal floating-point constants.)
527 The purpose of this unusual definition is to isolate the preprocessor
528 from the full complexity of numeric constants. It does not have to
529 distinguish between lexically valid and invalid floating-point numbers,
530 which is complicated. The definition also permits you to split an
531 identifier at any position and get exactly two tokens, which can then be
532 pasted back together with the @samp{##} operator.
534 It's possible for preprocessing numbers to cause programs to be
535 misinterpreted. For example, @code{0xE+12} is a preprocessing number
536 which does not translate to any valid numeric constant, therefore a
537 syntax error. It does not mean @code{@w{0xE + 12}}, which is what you
540 @cindex string literals
541 @cindex string constants
542 @cindex character constants
543 @cindex header file names
544 @c the @: prevents makeinfo from turning '' into ".
545 @dfn{String literals} are string constants, character constants, and
546 header file names (the argument of @samp{#include}).@footnote{The C
547 standard uses the term @dfn{string literal} to refer only to what we are
548 calling @dfn{string constants}.} String constants and character
549 constants are straightforward: @t{"@dots{}"} or @t{'@dots{}'}. In
550 either case embedded quotes should be escaped with a backslash:
551 @t{'\'@:'} is the character constant for @samp{'}. There is no limit on
552 the length of a character constant, but the value of a character
553 constant that contains more than one character is
554 implementation-defined. @xref{Implementation Details}.
556 Header file names either look like string constants, @t{"@dots{}"}, or are
557 written with angle brackets instead, @t{<@dots{}>}. In either case,
558 backslash is an ordinary character. There is no way to escape the
559 closing quote or angle bracket. The preprocessor looks for the header
560 file in different places depending on which form you use. @xref{Include
563 No string literal may extend past the end of a line. You may use continued
564 lines instead, or string constant concatenation.
568 @cindex alternative tokens
569 @dfn{Punctuators} are all the usual bits of punctuation which are
570 meaningful to C and C++. All but three of the punctuation characters in
571 ASCII are C punctuators. The exceptions are @samp{@@}, @samp{$}, and
572 @samp{`}. In addition, all the two- and three-character operators are
573 punctuators. There are also six @dfn{digraphs}, which the C++ standard
574 calls @dfn{alternative tokens}, which are merely alternate ways to spell
575 other punctuators. This is a second attempt to work around missing
576 punctuation in obsolete systems. It has no negative side effects,
577 unlike trigraphs, but does not cover as much ground. The digraphs and
578 their corresponding normal punctuators are:
581 Digraph: <% %> <: :> %: %:%:
582 Punctuator: @{ @} [ ] # ##
586 Any other single byte is considered ``other'' and passed on to the
587 preprocessor's output unchanged. The C compiler will almost certainly
588 reject source code containing ``other'' tokens. In ASCII, the only
589 ``other'' characters are @samp{@@}, @samp{$}, @samp{`}, and control
590 characters other than NUL (all bits zero). (Note that @samp{$} is
591 normally considered a letter.) All bytes with the high bit set
592 (numeric range 0x7F--0xFF) that were not succesfully interpreted as
593 part of an extended character in the input encoding are also ``other''
594 in the present implementation.
596 NUL is a special case because of the high probability that its
597 appearance is accidental, and because it may be invisible to the user
598 (many terminals do not display NUL at all). Within comments, NULs are
599 silently ignored, just as any other character would be. In running
600 text, NUL is considered white space. For example, these two directives
601 have the same meaning.
609 (where @samp{^@@} is ASCII NUL)@. Within string or character constants,
610 NULs are preserved. In the latter two cases the preprocessor emits a
613 @node The preprocessing language
614 @section The preprocessing language
616 @cindex preprocessing directives
617 @cindex directive line
618 @cindex directive name
620 After tokenization, the stream of tokens may simply be passed straight
621 to the compiler's parser. However, if it contains any operations in the
622 @dfn{preprocessing language}, it will be transformed first. This stage
623 corresponds roughly to the standard's ``translation phase 4'' and is
624 what most people think of as the preprocessor's job.
626 The preprocessing language consists of @dfn{directives} to be executed
627 and @dfn{macros} to be expanded. Its primary capabilities are:
631 Inclusion of header files. These are files of declarations that can be
632 substituted into your program.
635 Macro expansion. You can define @dfn{macros}, which are abbreviations
636 for arbitrary fragments of C code. The preprocessor will replace the
637 macros with their definitions throughout the program. Some macros are
638 automatically defined for you.
641 Conditional compilation. You can include or exclude parts of the
642 program according to various conditions.
645 Line control. If you use a program to combine or rearrange source files
646 into an intermediate file which is then compiled, you can use line
647 control to inform the compiler where each source line originally came
651 Diagnostics. You can detect problems at compile time and issue errors
655 There are a few more, less useful, features.
657 Except for expansion of predefined macros, all these operations are
658 triggered with @dfn{preprocessing directives}. Preprocessing directives
659 are lines in your program that start with @samp{#}. Whitespace is
660 allowed before and after the @samp{#}. The @samp{#} is followed by an
661 identifier, the @dfn{directive name}. It specifies the operation to
662 perform. Directives are commonly referred to as @samp{#@var{name}}
663 where @var{name} is the directive name. For example, @samp{#define} is
664 the directive that defines a macro.
666 The @samp{#} which begins a directive cannot come from a macro
667 expansion. Also, the directive name is not macro expanded. Thus, if
668 @code{foo} is defined as a macro expanding to @code{define}, that does
669 not make @samp{#foo} a valid preprocessing directive.
671 The set of valid directive names is fixed. Programs cannot define new
672 preprocessing directives.
674 Some directives require arguments; these make up the rest of the
675 directive line and must be separated from the directive name by
676 whitespace. For example, @samp{#define} must be followed by a macro
677 name and the intended expansion of the macro.
679 A preprocessing directive cannot cover more than one line. The line
680 may, however, be continued with backslash-newline, or by a block comment
681 which extends past the end of the line. In either case, when the
682 directive is processed, the continuations have already been merged with
683 the first line to make one long line.
686 @chapter Header Files
689 A header file is a file containing C declarations and macro definitions
690 (@pxref{Macros}) to be shared between several source files. You request
691 the use of a header file in your program by @dfn{including} it, with the
692 C preprocessing directive @samp{#include}.
694 Header files serve two purposes.
697 @cindex system header files
699 System header files declare the interfaces to parts of the operating
700 system. You include them in your program to supply the definitions and
701 declarations you need to invoke system calls and libraries.
704 Your own header files contain declarations for interfaces between the
705 source files of your program. Each time you have a group of related
706 declarations and macro definitions all or most of which are needed in
707 several different source files, it is a good idea to create a header
711 Including a header file produces the same results as copying the header
712 file into each source file that needs it. Such copying would be
713 time-consuming and error-prone. With a header file, the related
714 declarations appear in only one place. If they need to be changed, they
715 can be changed in one place, and programs that include the header file
716 will automatically use the new version when next recompiled. The header
717 file eliminates the labor of finding and changing all the copies as well
718 as the risk that a failure to find one copy will result in
719 inconsistencies within a program.
721 In C, the usual convention is to give header files names that end with
722 @file{.h}. It is most portable to use only letters, digits, dashes, and
723 underscores in header file names, and at most one dot.
727 * Include Operation::
729 * Once-Only Headers::
730 * Alternatives to Wrapper #ifndef::
731 * Computed Includes::
737 @section Include Syntax
740 Both user and system header files are included using the preprocessing
741 directive @samp{#include}. It has two variants:
744 @item #include <@var{file}>
745 This variant is used for system header files. It searches for a file
746 named @var{file} in a standard list of system directories. You can prepend
747 directories to this list with the @option{-I} option (@pxref{Invocation}).
749 @item #include "@var{file}"
750 This variant is used for header files of your own program. It
751 searches for a file named @var{file} first in the directory containing
752 the current file, then in the quote directories and then the same
753 directories used for @code{<@var{file}>}. You can prepend directories
754 to the list of quote directories with the @option{-iquote} option.
757 The argument of @samp{#include}, whether delimited with quote marks or
758 angle brackets, behaves like a string constant in that comments are not
759 recognized, and macro names are not expanded. Thus, @code{@w{#include
760 <x/*y>}} specifies inclusion of a system header file named @file{x/*y}.
762 However, if backslashes occur within @var{file}, they are considered
763 ordinary text characters, not escape characters. None of the character
764 escape sequences appropriate to string constants in C are processed.
765 Thus, @code{@w{#include "x\n\\y"}} specifies a filename containing three
766 backslashes. (Some systems interpret @samp{\} as a pathname separator.
767 All of these also interpret @samp{/} the same way. It is most portable
768 to use only @samp{/}.)
770 It is an error if there is anything (other than comments) on the line
773 @node Include Operation
774 @section Include Operation
776 The @samp{#include} directive works by directing the C preprocessor to
777 scan the specified file as input before continuing with the rest of the
778 current file. The output from the preprocessor contains the output
779 already generated, followed by the output resulting from the included
780 file, followed by the output that comes from the text after the
781 @samp{#include} directive. For example, if you have a header file
782 @file{header.h} as follows,
789 and a main program called @file{program.c} that uses the header file,
804 the compiler will see the same token stream as it would if
805 @file{program.c} read
818 Included files are not limited to declarations and macro definitions;
819 those are merely the typical uses. Any fragment of a C program can be
820 included from another file. The include file could even contain the
821 beginning of a statement that is concluded in the containing file, or
822 the end of a statement that was started in the including file. However,
823 an included file must consist of complete tokens. Comments and string
824 literals which have not been closed by the end of an included file are
825 invalid. For error recovery, they are considered to end at the end of
828 To avoid confusion, it is best if header files contain only complete
829 syntactic units---function declarations or definitions, type
832 The line following the @samp{#include} directive is always treated as a
833 separate line by the C preprocessor, even if the included file lacks a
839 By default, the preprocessor looks for header files included by the quote
840 form of the directive @code{@w{#include "@var{file}"}} first relative to
841 the directory of the current file, and then in a preconfigured list
842 of standard system directories.
843 For example, if @file{/usr/include/sys/stat.h} contains
844 @code{@w{#include "types.h"}}, GCC looks for @file{types.h} first in
845 @file{/usr/include/sys}, then in its usual search path.
847 For the angle-bracket form @code{@w{#include <@var{file}>}}, the
848 preprocessor's default behavior is to look only in the standard system
849 directories. The exact search directory list depends on the target
850 system, how GCC is configured, and where it is installed. You can
851 find the default search directory list for your version of CPP by
852 invoking it with the @option{-v} option. For example,
855 cpp -v /dev/null -o /dev/null
858 There are a number of command-line options you can use to add
859 additional directories to the search path.
860 The most commonly-used option is @option{-I@var{dir}}, which causes
861 @var{dir} to be searched after the current directory (for the quote
862 form of the directive) and ahead of the standard system directories.
863 You can specify multiple @option{-I} options on the command line,
864 in which case the directories are searched in left-to-right order.
866 If you need separate control over the search paths for the quote and
867 angle-bracket forms of the @samp{#include} directive, you can use the
868 @option{-iquote} and/or @option{-isystem} options instead of @option{-I}.
869 @xref{Invocation}, for a detailed description of these options, as
870 well as others that are less generally useful.
872 If you specify other options on the command line, such as @option{-I},
873 that affect where the preprocessor searches for header files, the
874 directory list printed by the @option{-v} option reflects the actual
875 search path used by the preprocessor.
877 Note that you can also prevent the preprocessor from searching any of
878 the default system header directories with the @option{-nostdinc}
879 option. This is useful when you are compiling an operating system
880 kernel or some other program that does not use the standard C library
881 facilities, or the standard C library itself.
883 @node Once-Only Headers
884 @section Once-Only Headers
885 @cindex repeated inclusion
886 @cindex including just once
887 @cindex wrapper @code{#ifndef}
889 If a header file happens to be included twice, the compiler will process
890 its contents twice. This is very likely to cause an error, e.g.@: when the
891 compiler sees the same structure definition twice. Even if it does not,
892 it will certainly waste time.
894 The standard way to prevent this is to enclose the entire real contents
895 of the file in a conditional, like this:
900 #ifndef FILE_FOO_SEEN
901 #define FILE_FOO_SEEN
903 @var{the entire file}
905 #endif /* !FILE_FOO_SEEN */
909 This construct is commonly known as a @dfn{wrapper #ifndef}.
910 When the header is included again, the conditional will be false,
911 because @code{FILE_FOO_SEEN} is defined. The preprocessor will skip
912 over the entire contents of the file, and the compiler will not see it
915 CPP optimizes even further. It remembers when a header file has a
916 wrapper @samp{#ifndef}. If a subsequent @samp{#include} specifies that
917 header, and the macro in the @samp{#ifndef} is still defined, it does
918 not bother to rescan the file at all.
920 You can put comments outside the wrapper. They will not interfere with
923 @cindex controlling macro
925 The macro @code{FILE_FOO_SEEN} is called the @dfn{controlling macro} or
926 @dfn{guard macro}. In a user header file, the macro name should not
927 begin with @samp{_}. In a system header file, it should begin with
928 @samp{__} to avoid conflicts with user programs. In any kind of header
929 file, the macro name should contain the name of the file and some
930 additional text, to avoid conflicts with other header files.
932 @node Alternatives to Wrapper #ifndef
933 @section Alternatives to Wrapper #ifndef
935 CPP supports two more ways of indicating that a header file should be
936 read only once. Neither one is as portable as a wrapper @samp{#ifndef}
937 and we recommend you do not use them in new programs, with the caveat
938 that @samp{#import} is standard practice in Objective-C.
941 CPP supports a variant of @samp{#include} called @samp{#import} which
942 includes a file, but does so at most once. If you use @samp{#import}
943 instead of @samp{#include}, then you don't need the conditionals
944 inside the header file to prevent multiple inclusion of the contents.
945 @samp{#import} is standard in Objective-C, but is considered a
946 deprecated extension in C and C++.
948 @samp{#import} is not a well designed feature. It requires the users of
949 a header file to know that it should only be included once. It is much
950 better for the header file's implementor to write the file so that users
951 don't need to know this. Using a wrapper @samp{#ifndef} accomplishes
954 In the present implementation, a single use of @samp{#import} will
955 prevent the file from ever being read again, by either @samp{#import} or
956 @samp{#include}. You should not rely on this; do not use both
957 @samp{#import} and @samp{#include} to refer to the same header file.
959 Another way to prevent a header file from being included more than once
960 is with the @samp{#pragma once} directive (@pxref{Pragmas}).
961 @samp{#pragma once} does not have the problems that @samp{#import} does,
962 but it is not recognized by all preprocessors, so you cannot rely on it
963 in a portable program.
965 @node Computed Includes
966 @section Computed Includes
967 @cindex computed includes
968 @cindex macros in include
970 Sometimes it is necessary to select one of several different header
971 files to be included into your program. They might specify
972 configuration parameters to be used on different sorts of operating
973 systems, for instance. You could do this with a series of conditionals,
977 # include "system_1.h"
979 # include "system_2.h"
985 That rapidly becomes tedious. Instead, the preprocessor offers the
986 ability to use a macro for the header name. This is called a
987 @dfn{computed include}. Instead of writing a header name as the direct
988 argument of @samp{#include}, you simply put a macro name there instead:
991 #define SYSTEM_H "system_1.h"
997 @code{SYSTEM_H} will be expanded, and the preprocessor will look for
998 @file{system_1.h} as if the @samp{#include} had been written that way
999 originally. @code{SYSTEM_H} could be defined by your Makefile with a
1002 You must be careful when you define the macro. @samp{#define} saves
1003 tokens, not text. The preprocessor has no way of knowing that the macro
1004 will be used as the argument of @samp{#include}, so it generates
1005 ordinary tokens, not a header name. This is unlikely to cause problems
1006 if you use double-quote includes, which are close enough to string
1007 constants. If you use angle brackets, however, you may have trouble.
1009 The syntax of a computed include is actually a bit more general than the
1010 above. If the first non-whitespace character after @samp{#include} is
1011 not @samp{"} or @samp{<}, then the entire line is macro-expanded
1012 like running text would be.
1014 If the line expands to a single string constant, the contents of that
1015 string constant are the file to be included. CPP does not re-examine the
1016 string for embedded quotes, but neither does it process backslash
1017 escapes in the string. Therefore
1020 #define HEADER "a\"b"
1025 looks for a file named @file{a\"b}. CPP searches for the file according
1026 to the rules for double-quoted includes.
1028 If the line expands to a token stream beginning with a @samp{<} token
1029 and including a @samp{>} token, then the tokens between the @samp{<} and
1030 the first @samp{>} are combined to form the filename to be included.
1031 Any whitespace between tokens is reduced to a single space; then any
1032 space after the initial @samp{<} is retained, but a trailing space
1033 before the closing @samp{>} is ignored. CPP searches for the file
1034 according to the rules for angle-bracket includes.
1036 In either case, if there are any tokens on the line after the file name,
1037 an error occurs and the directive is not processed. It is also an error
1038 if the result of expansion does not match either of the two expected
1041 These rules are implementation-defined behavior according to the C
1042 standard. To minimize the risk of different compilers interpreting your
1043 computed includes differently, we recommend you use only a single
1044 object-like macro which expands to a string constant. This will also
1045 minimize confusion for people reading your program.
1047 @node Wrapper Headers
1048 @section Wrapper Headers
1049 @cindex wrapper headers
1050 @cindex overriding a header file
1051 @findex #include_next
1053 Sometimes it is necessary to adjust the contents of a system-provided
1054 header file without editing it directly. GCC's @command{fixincludes}
1055 operation does this, for example. One way to do that would be to create
1056 a new header file with the same name and insert it in the search path
1057 before the original header. That works fine as long as you're willing
1058 to replace the old header entirely. But what if you want to refer to
1059 the old header from the new one?
1061 You cannot simply include the old header with @samp{#include}. That
1062 will start from the beginning, and find your new header again. If your
1063 header is not protected from multiple inclusion (@pxref{Once-Only
1064 Headers}), it will recurse infinitely and cause a fatal error.
1066 You could include the old header with an absolute pathname:
1068 #include "/usr/include/old-header.h"
1071 This works, but is not clean; should the system headers ever move, you
1072 would have to edit the new headers to match.
1074 There is no way to solve this problem within the C standard, but you can
1075 use the GNU extension @samp{#include_next}. It means, ``Include the
1076 @emph{next} file with this name''. This directive works like
1077 @samp{#include} except in searching for the specified file: it starts
1078 searching the list of header file directories @emph{after} the directory
1079 in which the current file was found.
1081 Suppose you specify @option{-I /usr/local/include}, and the list of
1082 directories to search also includes @file{/usr/include}; and suppose
1083 both directories contain @file{signal.h}. Ordinary @code{@w{#include
1084 <signal.h>}} finds the file under @file{/usr/local/include}. If that
1085 file contains @code{@w{#include_next <signal.h>}}, it starts searching
1086 after that directory, and finds the file in @file{/usr/include}.
1088 @samp{#include_next} does not distinguish between @code{<@var{file}>}
1089 and @code{"@var{file}"} inclusion, nor does it check that the file you
1090 specify has the same name as the current file. It simply looks for the
1091 file named, starting with the directory in the search path after the one
1092 where the current file was found.
1094 The use of @samp{#include_next} can lead to great confusion. We
1095 recommend it be used only when there is no other alternative. In
1096 particular, it should not be used in the headers belonging to a specific
1097 program; it should be used only to make global corrections along the
1098 lines of @command{fixincludes}.
1100 @node System Headers
1101 @section System Headers
1102 @cindex system header files
1104 The header files declaring interfaces to the operating system and
1105 runtime libraries often cannot be written in strictly conforming C@.
1106 Therefore, GCC gives code found in @dfn{system headers} special
1107 treatment. All warnings, other than those generated by @samp{#warning}
1108 (@pxref{Diagnostics}), are suppressed while GCC is processing a system
1109 header. Macros defined in a system header are immune to a few warnings
1110 wherever they are expanded. This immunity is granted on an ad-hoc
1111 basis, when we find that a warning generates lots of false positives
1112 because of code in macros defined in system headers.
1114 Normally, only the headers found in specific directories are considered
1115 system headers. These directories are determined when GCC is compiled.
1116 There are, however, two ways to make normal headers into system headers:
1120 Header files found in directories added to the search path with the
1121 @option{-isystem} and @option{-idirafter} command-line options are
1122 treated as system headers for the purposes of diagnostics.
1124 @findex #pragma GCC system_header
1126 There is also a directive, @code{@w{#pragma GCC system_header}}, which
1127 tells GCC to consider the rest of the current include file a system
1128 header, no matter where it was found. Code that comes before the
1129 @samp{#pragma} in the file is not affected. @code{@w{#pragma GCC
1130 system_header}} has no effect in the primary source file.
1133 On some targets, such as RS/6000 AIX, GCC implicitly surrounds all
1134 system headers with an @samp{extern "C"} block when compiling as C++.
1139 A @dfn{macro} is a fragment of code which has been given a name.
1140 Whenever the name is used, it is replaced by the contents of the macro.
1141 There are two kinds of macros. They differ mostly in what they look
1142 like when they are used. @dfn{Object-like} macros resemble data objects
1143 when used, @dfn{function-like} macros resemble function calls.
1145 You may define any valid identifier as a macro, even if it is a C
1146 keyword. The preprocessor does not know anything about keywords. This
1147 can be useful if you wish to hide a keyword such as @code{const} from an
1148 older compiler that does not understand it. However, the preprocessor
1149 operator @code{defined} (@pxref{Defined}) can never be defined as a
1150 macro, and C++'s named operators (@pxref{C++ Named Operators}) cannot be
1151 macros when you are compiling C++.
1154 * Object-like Macros::
1155 * Function-like Macros::
1160 * Predefined Macros::
1161 * Undefining and Redefining Macros::
1162 * Directives Within Macro Arguments::
1166 @node Object-like Macros
1167 @section Object-like Macros
1168 @cindex object-like macro
1169 @cindex symbolic constants
1170 @cindex manifest constants
1172 An @dfn{object-like macro} is a simple identifier which will be replaced
1173 by a code fragment. It is called object-like because it looks like a
1174 data object in code that uses it. They are most commonly used to give
1175 symbolic names to numeric constants.
1178 You create macros with the @samp{#define} directive. @samp{#define} is
1179 followed by the name of the macro and then the token sequence it should
1180 be an abbreviation for, which is variously referred to as the macro's
1181 @dfn{body}, @dfn{expansion} or @dfn{replacement list}. For example,
1184 #define BUFFER_SIZE 1024
1188 defines a macro named @code{BUFFER_SIZE} as an abbreviation for the
1189 token @code{1024}. If somewhere after this @samp{#define} directive
1190 there comes a C statement of the form
1193 foo = (char *) malloc (BUFFER_SIZE);
1197 then the C preprocessor will recognize and @dfn{expand} the macro
1198 @code{BUFFER_SIZE}. The C compiler will see the same tokens as it would
1202 foo = (char *) malloc (1024);
1205 By convention, macro names are written in uppercase. Programs are
1206 easier to read when it is possible to tell at a glance which names are
1209 The macro's body ends at the end of the @samp{#define} line. You may
1210 continue the definition onto multiple lines, if necessary, using
1211 backslash-newline. When the macro is expanded, however, it will all
1212 come out on one line. For example,
1215 #define NUMBERS 1, \
1218 int x[] = @{ NUMBERS @};
1219 @expansion{} int x[] = @{ 1, 2, 3 @};
1223 The most common visible consequence of this is surprising line numbers
1226 There is no restriction on what can go in a macro body provided it
1227 decomposes into valid preprocessing tokens. Parentheses need not
1228 balance, and the body need not resemble valid C code. (If it does not,
1229 you may get error messages from the C compiler when you use the macro.)
1231 The C preprocessor scans your program sequentially. Macro definitions
1232 take effect at the place you write them. Therefore, the following input
1233 to the C preprocessor
1249 When the preprocessor expands a macro name, the macro's expansion
1250 replaces the macro invocation, then the expansion is examined for more
1251 macros to expand. For example,
1255 #define TABLESIZE BUFSIZE
1256 #define BUFSIZE 1024
1258 @expansion{} BUFSIZE
1264 @code{TABLESIZE} is expanded first to produce @code{BUFSIZE}, then that
1265 macro is expanded to produce the final result, @code{1024}.
1267 Notice that @code{BUFSIZE} was not defined when @code{TABLESIZE} was
1268 defined. The @samp{#define} for @code{TABLESIZE} uses exactly the
1269 expansion you specify---in this case, @code{BUFSIZE}---and does not
1270 check to see whether it too contains macro names. Only when you
1271 @emph{use} @code{TABLESIZE} is the result of its expansion scanned for
1274 This makes a difference if you change the definition of @code{BUFSIZE}
1275 at some point in the source file. @code{TABLESIZE}, defined as shown,
1276 will always expand using the definition of @code{BUFSIZE} that is
1277 currently in effect:
1280 #define BUFSIZE 1020
1281 #define TABLESIZE BUFSIZE
1287 Now @code{TABLESIZE} expands (in two stages) to @code{37}.
1289 If the expansion of a macro contains its own name, either directly or
1290 via intermediate macros, it is not expanded again when the expansion is
1291 examined for more macros. This prevents infinite recursion.
1292 @xref{Self-Referential Macros}, for the precise details.
1294 @node Function-like Macros
1295 @section Function-like Macros
1296 @cindex function-like macros
1298 You can also define macros whose use looks like a function call. These
1299 are called @dfn{function-like macros}. To define a function-like macro,
1300 you use the same @samp{#define} directive, but you put a pair of
1301 parentheses immediately after the macro name. For example,
1304 #define lang_init() c_init()
1306 @expansion{} c_init()
1309 A function-like macro is only expanded if its name appears with a pair
1310 of parentheses after it. If you write just the name, it is left alone.
1311 This can be useful when you have a function and a macro of the same
1312 name, and you wish to use the function sometimes.
1315 extern void foo(void);
1316 #define foo() /* @r{optimized inline version} */
1322 Here the call to @code{foo()} will use the macro, but the function
1323 pointer will get the address of the real function. If the macro were to
1324 be expanded, it would cause a syntax error.
1326 If you put spaces between the macro name and the parentheses in the
1327 macro definition, that does not define a function-like macro, it defines
1328 an object-like macro whose expansion happens to begin with a pair of
1332 #define lang_init () c_init()
1334 @expansion{} () c_init()()
1337 The first two pairs of parentheses in this expansion come from the
1338 macro. The third is the pair that was originally after the macro
1339 invocation. Since @code{lang_init} is an object-like macro, it does not
1340 consume those parentheses.
1342 @node Macro Arguments
1343 @section Macro Arguments
1345 @cindex macros with arguments
1346 @cindex arguments in macro definitions
1348 Function-like macros can take @dfn{arguments}, just like true functions.
1349 To define a macro that uses arguments, you insert @dfn{parameters}
1350 between the pair of parentheses in the macro definition that make the
1351 macro function-like. The parameters must be valid C identifiers,
1352 separated by commas and optionally whitespace.
1354 To invoke a macro that takes arguments, you write the name of the macro
1355 followed by a list of @dfn{actual arguments} in parentheses, separated
1356 by commas. The invocation of the macro need not be restricted to a
1357 single logical line---it can cross as many lines in the source file as
1358 you wish. The number of arguments you give must match the number of
1359 parameters in the macro definition. When the macro is expanded, each
1360 use of a parameter in its body is replaced by the tokens of the
1361 corresponding argument. (You need not use all of the parameters in the
1364 As an example, here is a macro that computes the minimum of two numeric
1365 values, as it is defined in many C programs, and some uses.
1368 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
1369 x = min(a, b); @expansion{} x = ((a) < (b) ? (a) : (b));
1370 y = min(1, 2); @expansion{} y = ((1) < (2) ? (1) : (2));
1371 z = min(a + 28, *p); @expansion{} z = ((a + 28) < (*p) ? (a + 28) : (*p));
1375 (In this small example you can already see several of the dangers of
1376 macro arguments. @xref{Macro Pitfalls}, for detailed explanations.)
1378 Leading and trailing whitespace in each argument is dropped, and all
1379 whitespace between the tokens of an argument is reduced to a single
1380 space. Parentheses within each argument must balance; a comma within
1381 such parentheses does not end the argument. However, there is no
1382 requirement for square brackets or braces to balance, and they do not
1383 prevent a comma from separating arguments. Thus,
1386 macro (array[x = y, x + 1])
1390 passes two arguments to @code{macro}: @code{array[x = y} and @code{x +
1391 1]}. If you want to supply @code{array[x = y, x + 1]} as an argument,
1392 you can write it as @code{array[(x = y, x + 1)]}, which is equivalent C
1395 All arguments to a macro are completely macro-expanded before they are
1396 substituted into the macro body. After substitution, the complete text
1397 is scanned again for macros to expand, including the arguments. This rule
1398 may seem strange, but it is carefully designed so you need not worry
1399 about whether any function call is actually a macro invocation. You can
1400 run into trouble if you try to be too clever, though. @xref{Argument
1401 Prescan}, for detailed discussion.
1403 For example, @code{min (min (a, b), c)} is first expanded to
1406 min (((a) < (b) ? (a) : (b)), (c))
1414 ((((a) < (b) ? (a) : (b))) < (c)
1415 ? (((a) < (b) ? (a) : (b)))
1421 (Line breaks shown here for clarity would not actually be generated.)
1423 @cindex empty macro arguments
1424 You can leave macro arguments empty; this is not an error to the
1425 preprocessor (but many macros will then expand to invalid code).
1426 You cannot leave out arguments entirely; if a macro takes two arguments,
1427 there must be exactly one comma at the top level of its argument list.
1428 Here are some silly examples using @code{min}:
1431 min(, b) @expansion{} (( ) < (b) ? ( ) : (b))
1432 min(a, ) @expansion{} ((a ) < ( ) ? (a ) : ( ))
1433 min(,) @expansion{} (( ) < ( ) ? ( ) : ( ))
1434 min((,),) @expansion{} (((,)) < ( ) ? ((,)) : ( ))
1436 min() @error{} macro "min" requires 2 arguments, but only 1 given
1437 min(,,) @error{} macro "min" passed 3 arguments, but takes just 2
1440 Whitespace is not a preprocessing token, so if a macro @code{foo} takes
1441 one argument, @code{@w{foo ()}} and @code{@w{foo ( )}} both supply it an
1442 empty argument. Previous GNU preprocessor implementations and
1443 documentation were incorrect on this point, insisting that a
1444 function-like macro that takes a single argument be passed a space if an
1445 empty argument was required.
1447 Macro parameters appearing inside string literals are not replaced by
1448 their corresponding actual arguments.
1451 #define foo(x) x, "x"
1452 foo(bar) @expansion{} bar, "x"
1456 @section Stringizing
1458 @cindex @samp{#} operator
1460 Sometimes you may want to convert a macro argument into a string
1461 constant. Parameters are not replaced inside string constants, but you
1462 can use the @samp{#} preprocessing operator instead. When a macro
1463 parameter is used with a leading @samp{#}, the preprocessor replaces it
1464 with the literal text of the actual argument, converted to a string
1465 constant. Unlike normal parameter replacement, the argument is not
1466 macro-expanded first. This is called @dfn{stringizing}.
1468 There is no way to combine an argument with surrounding text and
1469 stringize it all together. Instead, you can write a series of adjacent
1470 string constants and stringized arguments. The preprocessor
1471 replaces the stringized arguments with string constants. The C
1472 compiler then combines all the adjacent string constants into one
1475 Here is an example of a macro definition that uses stringizing:
1479 #define WARN_IF(EXP) \
1481 fprintf (stderr, "Warning: " #EXP "\n"); @} \
1484 @expansion{} do @{ if (x == 0)
1485 fprintf (stderr, "Warning: " "x == 0" "\n"); @} while (0);
1490 The argument for @code{EXP} is substituted once, as-is, into the
1491 @code{if} statement, and once, stringized, into the argument to
1492 @code{fprintf}. If @code{x} were a macro, it would be expanded in the
1493 @code{if} statement, but not in the string.
1495 The @code{do} and @code{while (0)} are a kludge to make it possible to
1496 write @code{WARN_IF (@var{arg});}, which the resemblance of
1497 @code{WARN_IF} to a function would make C programmers want to do; see
1498 @ref{Swallowing the Semicolon}.
1500 Stringizing in C involves more than putting double-quote characters
1501 around the fragment. The preprocessor backslash-escapes the quotes
1502 surrounding embedded string constants, and all backslashes within string and
1503 character constants, in order to get a valid C string constant with the
1504 proper contents. Thus, stringizing @code{@w{p = "foo\n";}} results in
1505 @t{@w{"p = \"foo\\n\";"}}. However, backslashes that are not inside string
1506 or character constants are not duplicated: @samp{\n} by itself
1507 stringizes to @t{"\n"}.
1509 All leading and trailing whitespace in text being stringized is
1510 ignored. Any sequence of whitespace in the middle of the text is
1511 converted to a single space in the stringized result. Comments are
1512 replaced by whitespace long before stringizing happens, so they
1513 never appear in stringized text.
1515 There is no way to convert a macro argument into a character constant.
1517 If you want to stringize the result of expansion of a macro argument,
1518 you have to use two levels of macros.
1521 #define xstr(s) str(s)
1527 @expansion{} xstr (4)
1528 @expansion{} str (4)
1532 @code{s} is stringized when it is used in @code{str}, so it is not
1533 macro-expanded first. But @code{s} is an ordinary argument to
1534 @code{xstr}, so it is completely macro-expanded before @code{xstr}
1535 itself is expanded (@pxref{Argument Prescan}). Therefore, by the time
1536 @code{str} gets to its argument, it has already been macro-expanded.
1539 @section Concatenation
1540 @cindex concatenation
1541 @cindex token pasting
1542 @cindex token concatenation
1543 @cindex @samp{##} operator
1545 It is often useful to merge two tokens into one while expanding macros.
1546 This is called @dfn{token pasting} or @dfn{token concatenation}. The
1547 @samp{##} preprocessing operator performs token pasting. When a macro
1548 is expanded, the two tokens on either side of each @samp{##} operator
1549 are combined into a single token, which then replaces the @samp{##} and
1550 the two original tokens in the macro expansion. Usually both will be
1551 identifiers, or one will be an identifier and the other a preprocessing
1552 number. When pasted, they make a longer identifier. This isn't the
1553 only valid case. It is also possible to concatenate two numbers (or a
1554 number and a name, such as @code{1.5} and @code{e3}) into a number.
1555 Also, multi-character operators such as @code{+=} can be formed by
1558 However, two tokens that don't together form a valid token cannot be
1559 pasted together. For example, you cannot concatenate @code{x} with
1560 @code{+} in either order. If you try, the preprocessor issues a warning
1561 and emits the two tokens. Whether it puts white space between the
1562 tokens is undefined. It is common to find unnecessary uses of @samp{##}
1563 in complex macros. If you get this warning, it is likely that you can
1564 simply remove the @samp{##}.
1566 Both the tokens combined by @samp{##} could come from the macro body,
1567 but you could just as well write them as one token in the first place.
1568 Token pasting is most useful when one or both of the tokens comes from a
1569 macro argument. If either of the tokens next to an @samp{##} is a
1570 parameter name, it is replaced by its actual argument before @samp{##}
1571 executes. As with stringizing, the actual argument is not
1572 macro-expanded first. If the argument is empty, that @samp{##} has no
1575 Keep in mind that the C preprocessor converts comments to whitespace
1576 before macros are even considered. Therefore, you cannot create a
1577 comment by concatenating @samp{/} and @samp{*}. You can put as much
1578 whitespace between @samp{##} and its operands as you like, including
1579 comments, and you can put comments in arguments that will be
1580 concatenated. However, it is an error if @samp{##} appears at either
1581 end of a macro body.
1583 Consider a C program that interprets named commands. There probably
1584 needs to be a table of commands, perhaps an array of structures declared
1592 void (*function) (void);
1597 struct command commands[] =
1599 @{ "quit", quit_command @},
1600 @{ "help", help_command @},
1606 It would be cleaner not to have to give each command name twice, once in
1607 the string constant and once in the function name. A macro which takes the
1608 name of a command as an argument can make this unnecessary. The string
1609 constant can be created with stringizing, and the function name by
1610 concatenating the argument with @samp{_command}. Here is how it is done:
1613 #define COMMAND(NAME) @{ #NAME, NAME ## _command @}
1615 struct command commands[] =
1623 @node Variadic Macros
1624 @section Variadic Macros
1625 @cindex variable number of arguments
1626 @cindex macros with variable arguments
1627 @cindex variadic macros
1629 A macro can be declared to accept a variable number of arguments much as
1630 a function can. The syntax for defining the macro is similar to that of
1631 a function. Here is an example:
1634 #define eprintf(...) fprintf (stderr, __VA_ARGS__)
1637 This kind of macro is called @dfn{variadic}. When the macro is invoked,
1638 all the tokens in its argument list after the last named argument (this
1639 macro has none), including any commas, become the @dfn{variable
1640 argument}. This sequence of tokens replaces the identifier
1641 @code{@w{__VA_ARGS__}} in the macro body wherever it appears. Thus, we
1642 have this expansion:
1645 eprintf ("%s:%d: ", input_file, lineno)
1646 @expansion{} fprintf (stderr, "%s:%d: ", input_file, lineno)
1649 The variable argument is completely macro-expanded before it is inserted
1650 into the macro expansion, just like an ordinary argument. You may use
1651 the @samp{#} and @samp{##} operators to stringize the variable argument
1652 or to paste its leading or trailing token with another token. (But see
1653 below for an important special case for @samp{##}.)
1655 If your macro is complicated, you may want a more descriptive name for
1656 the variable argument than @code{@w{__VA_ARGS__}}. CPP permits
1657 this, as an extension. You may write an argument name immediately
1658 before the @samp{...}; that name is used for the variable argument.
1659 The @code{eprintf} macro above could be written
1662 #define eprintf(args...) fprintf (stderr, args)
1666 using this extension. You cannot use @code{@w{__VA_ARGS__}} and this
1667 extension in the same macro.
1669 You can have named arguments as well as variable arguments in a variadic
1670 macro. We could define @code{eprintf} like this, instead:
1673 #define eprintf(format, ...) fprintf (stderr, format, __VA_ARGS__)
1677 This formulation looks more descriptive, but historically it was less
1678 flexible: you had to supply at least one argument after the format
1679 string. In standard C, you could not omit the comma separating the
1680 named argument from the variable arguments. (Note that this
1681 restriction has been lifted in C++20, and never existed in GNU C; see
1684 Furthermore, if you left the variable argument empty, you would have
1685 gotten a syntax error, because there would have been an extra comma
1686 after the format string.
1689 eprintf("success!\n", );
1690 @expansion{} fprintf(stderr, "success!\n", );
1693 This has been fixed in C++20, and GNU CPP also has a pair of
1694 extensions which deal with this problem.
1696 First, in GNU CPP, and in C++ beginning in C++20, you are allowed to
1697 leave the variable argument out entirely:
1700 eprintf ("success!\n")
1701 @expansion{} fprintf(stderr, "success!\n", );
1705 Second, C++20 introduces the @code{@w{__VA_OPT__}} function macro.
1706 This macro may only appear in the definition of a variadic macro. If
1707 the variable argument has any tokens, then a @code{@w{__VA_OPT__}}
1708 invocation expands to its argument; but if the variable argument does
1709 not have any tokens, the @code{@w{__VA_OPT__}} expands to nothing:
1712 #define eprintf(format, ...) \
1713 fprintf (stderr, format __VA_OPT__(,) __VA_ARGS__)
1716 @code{@w{__VA_OPT__}} is also available in GNU C and GNU C++.
1718 Historically, GNU CPP has also had another extension to handle the
1719 trailing comma: the @samp{##} token paste operator has a special
1720 meaning when placed between a comma and a variable argument. Despite
1721 the introduction of @code{@w{__VA_OPT__}}, this extension remains
1722 supported in GNU CPP, for backward compatibility. If you write
1725 #define eprintf(format, ...) fprintf (stderr, format, ##__VA_ARGS__)
1729 and the variable argument is left out when the @code{eprintf} macro is
1730 used, then the comma before the @samp{##} will be deleted. This does
1731 @emph{not} happen if you pass an empty argument, nor does it happen if
1732 the token preceding @samp{##} is anything other than a comma.
1735 eprintf ("success!\n")
1736 @expansion{} fprintf(stderr, "success!\n");
1740 The above explanation is ambiguous about the case where the only macro
1741 parameter is a variable arguments parameter, as it is meaningless to
1742 try to distinguish whether no argument at all is an empty argument or
1744 CPP retains the comma when conforming to a specific C
1745 standard. Otherwise the comma is dropped as an extension to the standard.
1748 mandates that the only place the identifier @code{@w{__VA_ARGS__}}
1749 can appear is in the replacement list of a variadic macro. It may not
1750 be used as a macro name, macro argument name, or within a different type
1751 of macro. It may also be forbidden in open text; the standard is
1752 ambiguous. We recommend you avoid using it except for its defined
1755 Likewise, C++ forbids @code{@w{__VA_OPT__}} anywhere outside the
1756 replacement list of a variadic macro.
1758 Variadic macros became a standard part of the C language with C99.
1759 GNU CPP previously supported them
1760 with a named variable argument
1761 (@samp{args...}, not @samp{...} and @code{@w{__VA_ARGS__}}), which
1762 is still supported for backward compatibility.
1764 @node Predefined Macros
1765 @section Predefined Macros
1767 @cindex predefined macros
1768 Several object-like macros are predefined; you use them without
1769 supplying their definitions. They fall into three classes: standard,
1770 common, and system-specific.
1772 In C++, there is a fourth category, the named operators. They act like
1773 predefined macros, but you cannot undefine them.
1776 * Standard Predefined Macros::
1777 * Common Predefined Macros::
1778 * System-specific Predefined Macros::
1779 * C++ Named Operators::
1782 @node Standard Predefined Macros
1783 @subsection Standard Predefined Macros
1784 @cindex standard predefined macros.
1786 The standard predefined macros are specified by the relevant
1787 language standards, so they are available with all compilers that
1788 implement those standards. Older compilers may not provide all of
1789 them. Their names all start with double underscores.
1793 This macro expands to the name of the current input file, in the form of
1794 a C string constant. This is the path by which the preprocessor opened
1795 the file, not the short name specified in @samp{#include} or as the
1796 input file name argument. For example,
1797 @code{"/usr/local/include/myheader.h"} is a possible expansion of this
1801 This macro expands to the current input line number, in the form of a
1802 decimal integer constant. While we call it a predefined macro, it's
1803 a pretty strange macro, since its ``definition'' changes with each
1804 new line of source code.
1807 @code{__FILE__} and @code{__LINE__} are useful in generating an error
1808 message to report an inconsistency detected by the program; the message
1809 can state the source line at which the inconsistency was detected. For
1813 fprintf (stderr, "Internal error: "
1814 "negative string length "
1815 "%d at %s, line %d.",
1816 length, __FILE__, __LINE__);
1819 An @samp{#include} directive changes the expansions of @code{__FILE__}
1820 and @code{__LINE__} to correspond to the included file. At the end of
1821 that file, when processing resumes on the input file that contained
1822 the @samp{#include} directive, the expansions of @code{__FILE__} and
1823 @code{__LINE__} revert to the values they had before the
1824 @samp{#include} (but @code{__LINE__} is then incremented by one as
1825 processing moves to the line after the @samp{#include}).
1827 A @samp{#line} directive changes @code{__LINE__}, and may change
1828 @code{__FILE__} as well. @xref{Line Control}.
1830 C99 introduced @code{__func__}, and GCC has provided @code{__FUNCTION__}
1831 for a long time. Both of these are strings containing the name of the
1832 current function (there are slight semantic differences; see the GCC
1833 manual). Neither of them is a macro; the preprocessor does not know the
1834 name of the current function. They tend to be useful in conjunction
1835 with @code{__FILE__} and @code{__LINE__}, though.
1840 This macro expands to a string constant that describes the date on which
1841 the preprocessor is being run. The string constant contains eleven
1842 characters and looks like @code{@w{"Feb 12 1996"}}. If the day of the
1843 month is less than 10, it is padded with a space on the left.
1845 If GCC cannot determine the current date, it will emit a warning message
1846 (once per compilation) and @code{__DATE__} will expand to
1847 @code{@w{"??? ?? ????"}}.
1850 This macro expands to a string constant that describes the time at
1851 which the preprocessor is being run. The string constant contains
1852 eight characters and looks like @code{"23:59:01"}.
1854 If GCC cannot determine the current time, it will emit a warning message
1855 (once per compilation) and @code{__TIME__} will expand to
1859 In normal operation, this macro expands to the constant 1, to signify
1860 that this compiler conforms to ISO Standard C@. If GNU CPP is used with
1861 a compiler other than GCC, this is not necessarily true; however, the
1862 preprocessor always conforms to the standard unless the
1863 @option{-traditional-cpp} option is used.
1865 This macro is not defined if the @option{-traditional-cpp} option is used.
1867 On some hosts, the system compiler uses a different convention, where
1868 @code{__STDC__} is normally 0, but is 1 if the user specifies strict
1869 conformance to the C Standard. CPP follows the host convention when
1870 processing system header files, but when processing user files
1871 @code{__STDC__} is always 1. This has been reported to cause problems;
1872 for instance, some versions of Solaris provide X Windows headers that
1873 expect @code{__STDC__} to be either undefined or 1. @xref{Invocation}.
1875 @item __STDC_VERSION__
1876 This macro expands to the C Standard's version number, a long integer
1877 constant of the form @code{@var{yyyy}@var{mm}L} where @var{yyyy} and
1878 @var{mm} are the year and month of the Standard version. This signifies
1879 which version of the C Standard the compiler conforms to. Like
1880 @code{__STDC__}, this is not necessarily accurate for the entire
1881 implementation, unless GNU CPP is being used with GCC@.
1883 The value @code{199409L} signifies the 1989 C standard as amended in
1884 1994, which is the current default; the value @code{199901L} signifies
1885 the 1999 revision of the C standard; the value @code{201112L}
1886 signifies the 2011 revision of the C standard; the value
1887 @code{201710L} signifies the 2017 revision of the C standard (which is
1888 otherwise identical to the 2011 version apart from correction of
1889 defects). An unspecified value larger than @code{201710L} is used for
1890 the experimental @option{-std=c23} and @option{-std=gnu23} modes.
1892 This macro is not defined if the @option{-traditional-cpp} option is
1893 used, nor when compiling C++ or Objective-C@.
1895 @item __STDC_HOSTED__
1896 This macro is defined, with value 1, if the compiler's target is a
1897 @dfn{hosted environment}. A hosted environment has the complete
1898 facilities of the standard C library available.
1901 This macro is defined when the C++ compiler is in use. You can use
1902 @code{__cplusplus} to test whether a header is compiled by a C compiler
1903 or a C++ compiler. This macro is similar to @code{__STDC_VERSION__}, in
1904 that it expands to a version number. Depending on the language standard
1905 selected, the value of the macro is
1906 @code{199711L} for the 1998 C++ standard,
1907 @code{201103L} for the 2011 C++ standard,
1908 @code{201402L} for the 2014 C++ standard,
1909 @code{201703L} for the 2017 C++ standard,
1910 @code{202002L} for the 2020 C++ standard,
1911 @code{202302L} for the 2023 C++ standard,
1912 or an unspecified value strictly larger than @code{202302L} for the
1913 experimental languages enabled by @option{-std=c++26} and
1914 @option{-std=gnu++26}.
1917 This macro is defined, with value 1, when the Objective-C compiler is in
1918 use. You can use @code{__OBJC__} to test whether a header is compiled
1919 by a C compiler or an Objective-C compiler.
1922 This macro is defined with value 1 when preprocessing assembly
1927 @node Common Predefined Macros
1928 @subsection Common Predefined Macros
1929 @cindex common predefined macros
1931 The common predefined macros are GNU C extensions. They are available
1932 with the same meanings regardless of the machine or operating system on
1933 which you are using GNU C or GNU Fortran. Their names all start with
1939 This macro expands to sequential integral values starting from 0. In
1940 conjunction with the @code{##} operator, this provides a convenient means to
1941 generate unique identifiers. Care must be taken to ensure that
1942 @code{__COUNTER__} is not expanded prior to inclusion of precompiled headers
1943 which use it. Otherwise, the precompiled headers will not be used.
1946 The GNU Fortran compiler defines this.
1949 @itemx __GNUC_MINOR__
1950 @itemx __GNUC_PATCHLEVEL__
1951 These macros are defined by all GNU compilers that use the C
1952 preprocessor: C, C++, Objective-C and Fortran. Their values are the major
1953 version, minor version, and patch level of the compiler, as integer
1954 constants. For example, GCC version @var{x}.@var{y}.@var{z}
1955 defines @code{__GNUC__} to @var{x}, @code{__GNUC_MINOR__} to @var{y},
1956 and @code{__GNUC_PATCHLEVEL__} to @var{z}. These
1957 macros are also defined if you invoke the preprocessor directly.
1959 If all you need to know is whether or not your program is being compiled
1960 by GCC, or a non-GCC compiler that claims to accept the GNU C dialects,
1961 you can simply test @code{__GNUC__}. If you need to write code
1962 which depends on a specific version, you must be more careful. Each
1963 time the minor version is increased, the patch level is reset to zero;
1964 each time the major version is increased, the
1965 minor version and patch level are reset. If you wish to use the
1966 predefined macros directly in the conditional, you will need to write it
1970 /* @r{Test for GCC > 3.2.0} */
1971 #if __GNUC__ > 3 || \
1972 (__GNUC__ == 3 && (__GNUC_MINOR__ > 2 || \
1973 (__GNUC_MINOR__ == 2 && \
1974 __GNUC_PATCHLEVEL__ > 0))
1978 Another approach is to use the predefined macros to
1979 calculate a single number, then compare that against a threshold:
1982 #define GCC_VERSION (__GNUC__ * 10000 \
1983 + __GNUC_MINOR__ * 100 \
1984 + __GNUC_PATCHLEVEL__)
1986 /* @r{Test for GCC > 3.2.0} */
1987 #if GCC_VERSION > 30200
1991 Many people find this form easier to understand.
1994 The GNU C++ compiler defines this. Testing it is equivalent to
1995 testing @code{@w{(__GNUC__ && __cplusplus)}}.
1997 @item __STRICT_ANSI__
1998 GCC defines this macro if and only if the @option{-ansi} switch, or a
1999 @option{-std} switch specifying strict conformance to some version of ISO C
2000 or ISO C++, was specified when GCC was invoked. It is defined to @samp{1}.
2001 This macro exists primarily to direct GNU libc's header files to use only
2002 definitions found in standard C.
2005 This macro expands to the name of the main input file, in the form
2006 of a C string constant. This is the source file that was specified
2007 on the command line of the preprocessor or C compiler.
2010 This macro expands to the basename of the current input file, in the
2011 form of a C string constant. This is the last path component by which
2012 the preprocessor opened the file. For example, processing
2013 @code{"/usr/local/include/myheader.h"} would set this
2014 macro to @code{"myheader.h"}.
2016 @item __INCLUDE_LEVEL__
2017 This macro expands to a decimal integer constant that represents the
2018 depth of nesting in include files. The value of this macro is
2019 incremented on every @samp{#include} directive and decremented at the
2020 end of every included file. It starts out at 0, its value within the
2021 base file specified on the command line.
2024 This macro is defined if the target uses the ELF object format.
2027 This macro expands to a string constant which describes the version of
2028 the compiler in use. You should not rely on its contents having any
2029 particular form, but it can be counted on to contain at least the
2033 @itemx __OPTIMIZE_SIZE__
2034 @itemx __NO_INLINE__
2035 These macros describe the compilation mode. @code{__OPTIMIZE__} is
2036 defined in all optimizing compilations. @code{__OPTIMIZE_SIZE__} is
2037 defined if the compiler is optimizing for size, not speed.
2038 @code{__NO_INLINE__} is defined if no functions will be inlined into
2039 their callers (when not optimizing, or when inlining has been
2040 specifically disabled by @option{-fno-inline}).
2042 These macros cause certain GNU header files to provide optimized
2043 definitions, using macros or inline functions, of system library
2044 functions. You should not use these macros in any way unless you make
2045 sure that programs will execute with the same effect whether or not they
2046 are defined. If they are defined, their value is 1.
2048 @item __GNUC_GNU_INLINE__
2049 GCC defines this macro if functions declared @code{inline} will be
2050 handled in GCC's traditional gnu90 mode. Object files will contain
2051 externally visible definitions of all functions declared @code{inline}
2052 without @code{extern} or @code{static}. They will not contain any
2053 definitions of any functions declared @code{extern inline}.
2055 @item __GNUC_STDC_INLINE__
2056 GCC defines this macro if functions declared @code{inline} will be
2057 handled according to the ISO C99 or later standards. Object files will contain
2058 externally visible definitions of all functions declared @code{extern
2059 inline}. They will not contain definitions of any functions declared
2060 @code{inline} without @code{extern}.
2062 If this macro is defined, GCC supports the @code{gnu_inline} function
2063 attribute as a way to always get the gnu90 behavior.
2065 @item __CHAR_UNSIGNED__
2066 GCC defines this macro if and only if the data type @code{char} is
2067 unsigned on the target machine. It exists to cause the standard header
2068 file @file{limits.h} to work correctly. You should not use this macro
2069 yourself; instead, refer to the standard macros defined in @file{limits.h}.
2071 @item __WCHAR_UNSIGNED__
2072 Like @code{__CHAR_UNSIGNED__}, this macro is defined if and only if the
2073 data type @code{wchar_t} is unsigned and the front-end is in C++ mode.
2075 @item __REGISTER_PREFIX__
2076 This macro expands to a single token (not a string constant) which is
2077 the prefix applied to CPU register names in assembly language for this
2078 target. You can use it to write assembly that is usable in multiple
2079 environments. For example, in the @code{m68k-aout} environment it
2080 expands to nothing, but in the @code{m68k-coff} environment it expands
2081 to a single @samp{%}.
2083 @item __USER_LABEL_PREFIX__
2084 This macro expands to a single token which is the prefix applied to
2085 user labels (symbols visible to C code) in assembly. For example, in
2086 the @code{m68k-aout} environment it expands to an @samp{_}, but in the
2087 @code{m68k-coff} environment it expands to nothing.
2089 This macro will have the correct definition even if
2090 @option{-f(no-)underscores} is in use, but it will not be correct if
2091 target-specific options that adjust this prefix are used (e.g.@: the
2092 OSF/rose @option{-mno-underscores} option).
2095 @itemx __PTRDIFF_TYPE__
2096 @itemx __WCHAR_TYPE__
2097 @itemx __WINT_TYPE__
2098 @itemx __INTMAX_TYPE__
2099 @itemx __UINTMAX_TYPE__
2100 @itemx __SIG_ATOMIC_TYPE__
2101 @itemx __INT8_TYPE__
2102 @itemx __INT16_TYPE__
2103 @itemx __INT32_TYPE__
2104 @itemx __INT64_TYPE__
2105 @itemx __UINT8_TYPE__
2106 @itemx __UINT16_TYPE__
2107 @itemx __UINT32_TYPE__
2108 @itemx __UINT64_TYPE__
2109 @itemx __INT_LEAST8_TYPE__
2110 @itemx __INT_LEAST16_TYPE__
2111 @itemx __INT_LEAST32_TYPE__
2112 @itemx __INT_LEAST64_TYPE__
2113 @itemx __UINT_LEAST8_TYPE__
2114 @itemx __UINT_LEAST16_TYPE__
2115 @itemx __UINT_LEAST32_TYPE__
2116 @itemx __UINT_LEAST64_TYPE__
2117 @itemx __INT_FAST8_TYPE__
2118 @itemx __INT_FAST16_TYPE__
2119 @itemx __INT_FAST32_TYPE__
2120 @itemx __INT_FAST64_TYPE__
2121 @itemx __UINT_FAST8_TYPE__
2122 @itemx __UINT_FAST16_TYPE__
2123 @itemx __UINT_FAST32_TYPE__
2124 @itemx __UINT_FAST64_TYPE__
2125 @itemx __INTPTR_TYPE__
2126 @itemx __UINTPTR_TYPE__
2127 These macros are defined to the correct underlying types for the
2128 @code{size_t}, @code{ptrdiff_t}, @code{wchar_t}, @code{wint_t},
2129 @code{intmax_t}, @code{uintmax_t}, @code{sig_atomic_t}, @code{int8_t},
2130 @code{int16_t}, @code{int32_t}, @code{int64_t}, @code{uint8_t},
2131 @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
2132 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
2133 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
2134 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
2135 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
2136 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
2137 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t} typedefs,
2138 respectively. They exist to make the standard header files
2139 @file{stddef.h}, @file{stdint.h}, and @file{wchar.h} work correctly.
2140 You should not use these macros directly; instead, include the
2141 appropriate headers and use the typedefs. Some of these macros may
2142 not be defined on particular systems if GCC does not provide a
2143 @file{stdint.h} header on those systems.
2146 Defined to the number of bits used in the representation of the
2147 @code{char} data type. It exists to make the standard header given
2148 numerical limits work correctly. You should not use
2149 this macro directly; instead, include the appropriate headers.
2152 @itemx __WCHAR_MAX__
2156 @itemx __LONG_LONG_MAX__
2159 @itemx __PTRDIFF_MAX__
2160 @itemx __INTMAX_MAX__
2161 @itemx __UINTMAX_MAX__
2162 @itemx __SIG_ATOMIC_MAX__
2164 @itemx __INT16_MAX__
2165 @itemx __INT32_MAX__
2166 @itemx __INT64_MAX__
2167 @itemx __UINT8_MAX__
2168 @itemx __UINT16_MAX__
2169 @itemx __UINT32_MAX__
2170 @itemx __UINT64_MAX__
2171 @itemx __INT_LEAST8_MAX__
2172 @itemx __INT_LEAST16_MAX__
2173 @itemx __INT_LEAST32_MAX__
2174 @itemx __INT_LEAST64_MAX__
2175 @itemx __UINT_LEAST8_MAX__
2176 @itemx __UINT_LEAST16_MAX__
2177 @itemx __UINT_LEAST32_MAX__
2178 @itemx __UINT_LEAST64_MAX__
2179 @itemx __INT_FAST8_MAX__
2180 @itemx __INT_FAST16_MAX__
2181 @itemx __INT_FAST32_MAX__
2182 @itemx __INT_FAST64_MAX__
2183 @itemx __UINT_FAST8_MAX__
2184 @itemx __UINT_FAST16_MAX__
2185 @itemx __UINT_FAST32_MAX__
2186 @itemx __UINT_FAST64_MAX__
2187 @itemx __INTPTR_MAX__
2188 @itemx __UINTPTR_MAX__
2189 @itemx __WCHAR_MIN__
2191 @itemx __SIG_ATOMIC_MIN__
2192 Defined to the maximum value of the @code{signed char}, @code{wchar_t},
2193 @code{signed short},
2194 @code{signed int}, @code{signed long}, @code{signed long long},
2195 @code{wint_t}, @code{size_t}, @code{ptrdiff_t},
2196 @code{intmax_t}, @code{uintmax_t}, @code{sig_atomic_t}, @code{int8_t},
2197 @code{int16_t}, @code{int32_t}, @code{int64_t}, @code{uint8_t},
2198 @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
2199 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
2200 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
2201 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
2202 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
2203 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
2204 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t} types and
2205 to the minimum value of the @code{wchar_t}, @code{wint_t}, and
2206 @code{sig_atomic_t} types respectively. They exist to make the
2207 standard header given numerical limits work correctly. You should not
2208 use these macros directly; instead, include the appropriate headers.
2209 Some of these macros may not be defined on particular systems if GCC
2210 does not provide a @file{stdint.h} header on those systems.
2222 Defined to implementations of the standard @file{stdint.h} macros with
2223 the same names without the leading @code{__}. They exist the make the
2224 implementation of that header work correctly. You should not use
2225 these macros directly; instead, include the appropriate headers. Some
2226 of these macros may not be defined on particular systems if GCC does
2227 not provide a @file{stdint.h} header on those systems.
2229 @item __SCHAR_WIDTH__
2230 @itemx __SHRT_WIDTH__
2231 @itemx __INT_WIDTH__
2232 @itemx __LONG_WIDTH__
2233 @itemx __LONG_LONG_WIDTH__
2234 @itemx __PTRDIFF_WIDTH__
2235 @itemx __SIG_ATOMIC_WIDTH__
2236 @itemx __SIZE_WIDTH__
2237 @itemx __WCHAR_WIDTH__
2238 @itemx __WINT_WIDTH__
2239 @itemx __INT_LEAST8_WIDTH__
2240 @itemx __INT_LEAST16_WIDTH__
2241 @itemx __INT_LEAST32_WIDTH__
2242 @itemx __INT_LEAST64_WIDTH__
2243 @itemx __INT_FAST8_WIDTH__
2244 @itemx __INT_FAST16_WIDTH__
2245 @itemx __INT_FAST32_WIDTH__
2246 @itemx __INT_FAST64_WIDTH__
2247 @itemx __INTPTR_WIDTH__
2248 @itemx __INTMAX_WIDTH__
2249 Defined to the bit widths of the corresponding types. They exist to
2250 make the implementations of @file{limits.h} and @file{stdint.h} behave
2251 correctly. You should not use these macros directly; instead, include
2252 the appropriate headers. Some of these macros may not be defined on
2253 particular systems if GCC does not provide a @file{stdint.h} header on
2256 @item __SIZEOF_INT__
2257 @itemx __SIZEOF_LONG__
2258 @itemx __SIZEOF_LONG_LONG__
2259 @itemx __SIZEOF_SHORT__
2260 @itemx __SIZEOF_POINTER__
2261 @itemx __SIZEOF_FLOAT__
2262 @itemx __SIZEOF_DOUBLE__
2263 @itemx __SIZEOF_LONG_DOUBLE__
2264 @itemx __SIZEOF_SIZE_T__
2265 @itemx __SIZEOF_WCHAR_T__
2266 @itemx __SIZEOF_WINT_T__
2267 @itemx __SIZEOF_PTRDIFF_T__
2268 Defined to the number of bytes of the C standard data types: @code{int},
2269 @code{long}, @code{long long}, @code{short}, @code{void *}, @code{float},
2270 @code{double}, @code{long double}, @code{size_t}, @code{wchar_t}, @code{wint_t}
2271 and @code{ptrdiff_t}.
2273 @item __BYTE_ORDER__
2274 @itemx __ORDER_LITTLE_ENDIAN__
2275 @itemx __ORDER_BIG_ENDIAN__
2276 @itemx __ORDER_PDP_ENDIAN__
2277 @code{__BYTE_ORDER__} is defined to one of the values
2278 @code{__ORDER_LITTLE_ENDIAN__}, @code{__ORDER_BIG_ENDIAN__}, or
2279 @code{__ORDER_PDP_ENDIAN__} to reflect the layout of multi-byte and
2280 multi-word quantities in memory. If @code{__BYTE_ORDER__} is equal to
2281 @code{__ORDER_LITTLE_ENDIAN__} or @code{__ORDER_BIG_ENDIAN__}, then
2282 multi-byte and multi-word quantities are laid out identically: the
2283 byte (word) at the lowest address is the least significant or most
2284 significant byte (word) of the quantity, respectively. If
2285 @code{__BYTE_ORDER__} is equal to @code{__ORDER_PDP_ENDIAN__}, then
2286 bytes in 16-bit words are laid out in a little-endian fashion, whereas
2287 the 16-bit subwords of a 32-bit quantity are laid out in big-endian
2290 You should use these macros for testing like this:
2293 /* @r{Test for a little-endian machine} */
2294 #if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
2297 @item __FLOAT_WORD_ORDER__
2298 @code{__FLOAT_WORD_ORDER__} is defined to one of the values
2299 @code{__ORDER_LITTLE_ENDIAN__} or @code{__ORDER_BIG_ENDIAN__} to reflect
2300 the layout of the words of multi-word floating-point quantities.
2303 This macro is defined, with value 1, when compiling a C++ source file
2304 with warnings about deprecated constructs enabled. These warnings are
2305 enabled by default, but can be disabled with @option{-Wno-deprecated}.
2308 This macro is defined, with value 1, when compiling a C++ source file
2309 with exceptions enabled. If @option{-fno-exceptions} is used when
2310 compiling the file, then this macro is not defined.
2313 This macro is defined, with value 1, when compiling a C++ source file
2314 with runtime type identification enabled. If @option{-fno-rtti} is
2315 used when compiling the file, then this macro is not defined.
2317 @item __USING_SJLJ_EXCEPTIONS__
2318 This macro is defined, with value 1, if the compiler uses the old
2319 mechanism based on @code{setjmp} and @code{longjmp} for exception
2322 @item __GXX_EXPERIMENTAL_CXX0X__
2323 This macro is defined when compiling a C++ source file with C++11 features
2324 enabled, i.e., for all C++ language dialects except @option{-std=c++98}
2325 and @option{-std=gnu++98}. This macro is obsolete, but can be used to
2326 detect experimental C++0x features in very old versions of GCC. Since
2327 GCC 4.7.0 the @code{__cplusplus} macro is defined correctly, so most
2328 code should test @code{__cplusplus >= 201103L} instead of using this
2332 This macro is defined when compiling a C++ source file. It has the
2333 value 1 if the compiler will use weak symbols, COMDAT sections, or
2334 other similar techniques to collapse symbols with ``vague linkage''
2335 that are defined in multiple translation units. If the compiler will
2336 not collapse such symbols, this macro is defined with value 0. In
2337 general, user code should not need to make use of this macro; the
2338 purpose of this macro is to ease implementation of the C++ runtime
2339 library provided with G++.
2341 @item __NEXT_RUNTIME__
2342 This macro is defined, with value 1, if (and only if) the NeXT runtime
2343 (as in @option{-fnext-runtime}) is in use for Objective-C@. If the GNU
2344 runtime is used, this macro is not defined, so that you can use this
2345 macro to determine which runtime (NeXT or GNU) is being used.
2349 These macros are defined, with value 1, if (and only if) the compilation
2350 is for a target where @code{long int} and pointer both use 64-bits and
2351 @code{int} uses 32-bit.
2354 This macro is defined, with value 1, when @option{-fstack-protector} is in
2358 This macro is defined, with value 2, when @option{-fstack-protector-all} is
2361 @item __SSP_STRONG__
2362 This macro is defined, with value 3, when @option{-fstack-protector-strong} is
2365 @item __SSP_EXPLICIT__
2366 This macro is defined, with value 4, when @option{-fstack-protector-explicit} is
2369 @item __SANITIZE_ADDRESS__
2370 This macro is defined, with value 1, when @option{-fsanitize=address}
2371 or @option{-fsanitize=kernel-address} are in use.
2373 @item __SANITIZE_THREAD__
2374 This macro is defined, with value 1, when @option{-fsanitize=thread} is in use.
2377 This macro expands to a string constant that describes the date and time
2378 of the last modification of the current source file. The string constant
2379 contains abbreviated day of the week, month, day of the month, time in
2380 hh:mm:ss form, year and looks like @code{@w{"Sun Sep 16 01:03:52 1973"}}.
2381 If the day of the month is less than 10, it is padded with a space on the left.
2383 If GCC cannot determine the current date, it will emit a warning message
2384 (once per compilation) and @code{__TIMESTAMP__} will expand to
2385 @code{@w{"??? ??? ?? ??:??:?? ????"}}.
2387 @item __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1
2388 @itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2
2389 @itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4
2390 @itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8
2391 @itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_16
2392 These macros are defined when the target processor supports atomic compare
2393 and swap operations on operands 1, 2, 4, 8 or 16 bytes in length, respectively.
2395 @item __HAVE_SPECULATION_SAFE_VALUE
2396 This macro is defined with the value 1 to show that this version of GCC
2397 supports @code{__builtin_speculation_safe_value}.
2399 @item __GCC_HAVE_DWARF2_CFI_ASM
2400 This macro is defined when the compiler is emitting DWARF CFI directives
2401 to the assembler. When this is defined, it is possible to emit those same
2402 directives in inline assembly.
2405 @itemx __FP_FAST_FMAF
2406 @itemx __FP_FAST_FMAL
2407 These macros are defined with value 1 if the backend supports the
2408 @code{fma}, @code{fmaf}, and @code{fmal} builtin functions, so that
2409 the include file @file{math.h} can define the macros
2410 @code{FP_FAST_FMA}, @code{FP_FAST_FMAF}, and @code{FP_FAST_FMAL}
2411 for compatibility with the 1999 C standard.
2413 @item __FP_FAST_FMAF16
2414 @itemx __FP_FAST_FMAF32
2415 @itemx __FP_FAST_FMAF64
2416 @itemx __FP_FAST_FMAF128
2417 @itemx __FP_FAST_FMAF32X
2418 @itemx __FP_FAST_FMAF64X
2419 @itemx __FP_FAST_FMAF128X
2420 These macros are defined with the value 1 if the backend supports the
2421 @code{fma} functions using the additional @code{_Float@var{n}} and
2422 @code{_Float@var{n}x} types that are defined in ISO/IEC TS
2423 18661-3:2015. The include file @file{math.h} can define the
2424 @code{FP_FAST_FMAF@var{n}} and @code{FP_FAST_FMAF@var{n}x} macros if
2425 the user defined @code{__STDC_WANT_IEC_60559_TYPES_EXT__} before
2426 including @file{math.h}.
2429 This macro is defined to indicate the intended level of support for
2430 IEEE 754 (IEC 60559) floating-point arithmetic. It expands to a
2431 nonnegative integer value. If 0, it indicates that the combination of
2432 the compiler configuration and the command-line options is not
2433 intended to support IEEE 754 arithmetic for @code{float} and
2434 @code{double} as defined in C99 and C11 Annex F (for example, that the
2435 standard rounding modes and exceptions are not supported, or that
2436 optimizations are enabled that conflict with IEEE 754 semantics). If
2437 1, it indicates that IEEE 754 arithmetic is intended to be supported;
2438 this does not mean that all relevant language features are supported
2439 by GCC. If 2 or more, it additionally indicates support for IEEE
2440 754-2008 (in particular, that the binary encodings for quiet and
2441 signaling NaNs are as specified in IEEE 754-2008).
2443 This macro does not indicate the default state of command-line options
2444 that control optimizations that C99 and C11 permit to be controlled by
2445 standard pragmas, where those standards do not require a particular
2446 default state. It does not indicate whether optimizations respect
2447 signaling NaN semantics (the macro for that is
2448 @code{__SUPPORT_SNAN__}). It does not indicate support for decimal
2449 floating point or the IEEE 754 binary16 and binary128 types.
2451 @item __GCC_IEC_559_COMPLEX
2452 This macro is defined to indicate the intended level of support for
2453 IEEE 754 (IEC 60559) floating-point arithmetic for complex numbers, as
2454 defined in C99 and C11 Annex G. It expands to a nonnegative integer
2455 value. If 0, it indicates that the combination of the compiler
2456 configuration and the command-line options is not intended to support
2457 Annex G requirements (for example, because @option{-fcx-limited-range}
2458 was used). If 1 or more, it indicates that it is intended to support
2459 those requirements; this does not mean that all relevant language
2460 features are supported by GCC.
2462 @item __NO_MATH_ERRNO__
2463 This macro is defined if @option{-fno-math-errno} is used, or enabled
2464 by another option such as @option{-ffast-math} or by default.
2466 @item __RECIPROCAL_MATH__
2467 This macro is defined if @option{-freciprocal-math} is used, or enabled
2468 by another option such as @option{-ffast-math} or by default.
2470 @item __NO_SIGNED_ZEROS__
2471 This macro is defined if @option{-fno-signed-zeros} is used, or enabled
2472 by another option such as @option{-ffast-math} or by default.
2474 @item __NO_TRAPPING_MATH__
2475 This macro is defined if @option{-fno-trapping-math} is used.
2477 @item __ASSOCIATIVE_MATH__
2478 This macro is defined if @option{-fassociative-math} is used, or enabled
2479 by another option such as @option{-ffast-math} or by default.
2481 @item __ROUNDING_MATH__
2482 This macro is defined if @option{-frounding-math} is used.
2484 @item __GNUC_EXECUTION_CHARSET_NAME
2485 @itemx __GNUC_WIDE_EXECUTION_CHARSET_NAME
2486 These macros are defined to expand to a narrow string literal of
2487 the name of the narrow and wide compile-time execution character
2488 set used. It directly reflects the name passed to the options
2489 @option{-fexec-charset} and @option{-fwide-exec-charset}, or the defaults
2490 documented for those options (that is, it can expand to something like
2491 @code{"UTF-8"}). @xref{Invocation}.
2494 @node System-specific Predefined Macros
2495 @subsection System-specific Predefined Macros
2497 @cindex system-specific predefined macros
2498 @cindex predefined macros, system-specific
2499 @cindex reserved namespace
2501 The C preprocessor normally predefines several macros that indicate what
2502 type of system and machine is in use. They are obviously different on
2503 each target supported by GCC@. This manual, being for all systems and
2504 machines, cannot tell you what their names are, but you can use
2505 @command{cpp -dM} to see them all. @xref{Invocation}. All system-specific
2506 predefined macros expand to a constant value, so you can test them with
2507 either @samp{#ifdef} or @samp{#if}.
2509 The C standard requires that all system-specific macros be part of the
2510 @dfn{reserved namespace}. All names which begin with two underscores,
2511 or an underscore and a capital letter, are reserved for the compiler and
2512 library to use as they wish. However, historically system-specific
2513 macros have had names with no special prefix; for instance, it is common
2514 to find @code{unix} defined on Unix systems. For all such macros, GCC
2515 provides a parallel macro with two underscores added at the beginning
2516 and the end. If @code{unix} is defined, @code{__unix__} will be defined
2517 too. There will never be more than two underscores; the parallel of
2518 @code{_mips} is @code{__mips__}.
2520 When the @option{-ansi} option, or any @option{-std} option that
2521 requests strict conformance, is given to the compiler, all the
2522 system-specific predefined macros outside the reserved namespace are
2523 suppressed. The parallel macros, inside the reserved namespace, remain
2526 We are slowly phasing out all predefined macros which are outside the
2527 reserved namespace. You should never use them in new programs, and we
2528 encourage you to correct older code to use the parallel macros whenever
2529 you find it. We don't recommend you use the system-specific macros that
2530 are in the reserved namespace, either. It is better in the long run to
2531 check specifically for features you need, using a tool such as
2534 @node C++ Named Operators
2535 @subsection C++ Named Operators
2536 @cindex named operators
2537 @cindex C++ named operators
2538 @cindex @file{iso646.h}
2540 In C++, there are eleven keywords which are simply alternate spellings
2541 of operators normally written with punctuation. These keywords are
2542 treated as such even in the preprocessor. They function as operators in
2543 @samp{#if}, and they cannot be defined as macros or poisoned. In C, you
2544 can request that those keywords take their C++ meaning by including
2545 @file{iso646.h}. That header defines each one as a normal object-like
2546 macro expanding to the appropriate punctuator.
2548 These are the named operators and their corresponding punctuators:
2550 @multitable {Named Operator} {Punctuator}
2551 @item Named Operator @tab Punctuator
2552 @item @code{and} @tab @code{&&}
2553 @item @code{and_eq} @tab @code{&=}
2554 @item @code{bitand} @tab @code{&}
2555 @item @code{bitor} @tab @code{|}
2556 @item @code{compl} @tab @code{~}
2557 @item @code{not} @tab @code{!}
2558 @item @code{not_eq} @tab @code{!=}
2559 @item @code{or} @tab @code{||}
2560 @item @code{or_eq} @tab @code{|=}
2561 @item @code{xor} @tab @code{^}
2562 @item @code{xor_eq} @tab @code{^=}
2565 @node Undefining and Redefining Macros
2566 @section Undefining and Redefining Macros
2567 @cindex undefining macros
2568 @cindex redefining macros
2571 If a macro ceases to be useful, it may be @dfn{undefined} with the
2572 @samp{#undef} directive. @samp{#undef} takes a single argument, the
2573 name of the macro to undefine. You use the bare macro name, even if the
2574 macro is function-like. It is an error if anything appears on the line
2575 after the macro name. @samp{#undef} has no effect if the name is not a
2580 x = FOO; @expansion{} x = 4;
2582 x = FOO; @expansion{} x = FOO;
2585 Once a macro has been undefined, that identifier may be @dfn{redefined}
2586 as a macro by a subsequent @samp{#define} directive. The new definition
2587 need not have any resemblance to the old definition.
2589 However, if an identifier which is currently a macro is redefined, then
2590 the new definition must be @dfn{effectively the same} as the old one.
2591 Two macro definitions are effectively the same if:
2593 @item Both are the same type of macro (object- or function-like).
2594 @item All the tokens of the replacement list are the same.
2595 @item If there are any parameters, they are the same.
2596 @item Whitespace appears in the same places in both. It need not be
2597 exactly the same amount of whitespace, though. Remember that comments
2598 count as whitespace.
2602 These definitions are effectively the same:
2604 #define FOUR (2 + 2)
2605 #define FOUR (2 + 2)
2606 #define FOUR (2 /* @r{two} */ + 2)
2611 #define FOUR (2 + 2)
2612 #define FOUR ( 2+2 )
2613 #define FOUR (2 * 2)
2614 #define FOUR(score,and,seven,years,ago) (2 + 2)
2617 If a macro is redefined with a definition that is not effectively the
2618 same as the old one, the preprocessor issues a warning and changes the
2619 macro to use the new definition. If the new definition is effectively
2620 the same, the redefinition is silently ignored. This allows, for
2621 instance, two different headers to define a common macro. The
2622 preprocessor will only complain if the definitions do not match.
2624 @node Directives Within Macro Arguments
2625 @section Directives Within Macro Arguments
2626 @cindex macro arguments and directives
2628 Occasionally it is convenient to use preprocessor directives within
2629 the arguments of a macro. The C and C++ standards declare that
2630 behavior in these cases is undefined. GNU CPP
2631 processes arbitrary directives within macro arguments in
2632 exactly the same way as it would have processed the directive were the
2633 function-like macro invocation not present.
2635 If, within a macro invocation, that macro is redefined, then the new
2636 definition takes effect in time for argument pre-expansion, but the
2637 original definition is still used for argument replacement. Here is a
2638 pathological example:
2656 with the semantics described above.
2658 @node Macro Pitfalls
2659 @section Macro Pitfalls
2660 @cindex problems with macros
2661 @cindex pitfalls of macros
2663 In this section we describe some special rules that apply to macros and
2664 macro expansion, and point out certain cases in which the rules have
2665 counter-intuitive consequences that you must watch out for.
2669 * Operator Precedence Problems::
2670 * Swallowing the Semicolon::
2671 * Duplication of Side Effects::
2672 * Self-Referential Macros::
2673 * Argument Prescan::
2674 * Newlines in Arguments::
2678 @subsection Misnesting
2680 When a macro is called with arguments, the arguments are substituted
2681 into the macro body and the result is checked, together with the rest of
2682 the input file, for more macro calls. It is possible to piece together
2683 a macro call coming partially from the macro body and partially from the
2684 arguments. For example,
2687 #define twice(x) (2*(x))
2688 #define call_with_1(x) x(1)
2690 @expansion{} twice(1)
2691 @expansion{} (2*(1))
2694 Macro definitions do not have to have balanced parentheses. By writing
2695 an unbalanced open parenthesis in a macro body, it is possible to create
2696 a macro call that begins inside the macro body but ends outside of it.
2700 #define strange(file) fprintf (file, "%s %d",
2702 strange(stderr) p, 35)
2703 @expansion{} fprintf (stderr, "%s %d", p, 35)
2706 The ability to piece together a macro call can be useful, but the use of
2707 unbalanced open parentheses in a macro body is just confusing, and
2710 @node Operator Precedence Problems
2711 @subsection Operator Precedence Problems
2712 @cindex parentheses in macro bodies
2714 You may have noticed that in most of the macro definition examples shown
2715 above, each occurrence of a macro argument name had parentheses around
2716 it. In addition, another pair of parentheses usually surround the
2717 entire macro definition. Here is why it is best to write macros that
2720 Suppose you define a macro as follows,
2723 #define ceil_div(x, y) (x + y - 1) / y
2727 whose purpose is to divide, rounding up. (One use for this operation is
2728 to compute how many @code{int} objects are needed to hold a certain
2729 number of @code{char} objects.) Then suppose it is used as follows:
2732 a = ceil_div (b & c, sizeof (int));
2733 @expansion{} a = (b & c + sizeof (int) - 1) / sizeof (int);
2737 This does not do what is intended. The operator-precedence rules of
2738 C make it equivalent to this:
2741 a = (b & (c + sizeof (int) - 1)) / sizeof (int);
2745 What we want is this:
2748 a = ((b & c) + sizeof (int) - 1)) / sizeof (int);
2752 Defining the macro as
2755 #define ceil_div(x, y) ((x) + (y) - 1) / (y)
2759 provides the desired result.
2761 Unintended grouping can result in another way. Consider @code{sizeof
2762 ceil_div(1, 2)}. That has the appearance of a C expression that would
2763 compute the size of the type of @code{ceil_div (1, 2)}, but in fact it
2764 means something very different. Here is what it expands to:
2767 sizeof ((1) + (2) - 1) / (2)
2771 This would take the size of an integer and divide it by two. The
2772 precedence rules have put the division outside the @code{sizeof} when it
2773 was intended to be inside.
2775 Parentheses around the entire macro definition prevent such problems.
2776 Here, then, is the recommended way to define @code{ceil_div}:
2779 #define ceil_div(x, y) (((x) + (y) - 1) / (y))
2782 @node Swallowing the Semicolon
2783 @subsection Swallowing the Semicolon
2784 @cindex semicolons (after macro calls)
2786 Often it is desirable to define a macro that expands into a compound
2787 statement. Consider, for example, the following macro, that advances a
2788 pointer (the argument @code{p} says where to find it) across whitespace
2792 #define SKIP_SPACES(p, limit) \
2793 @{ char *lim = (limit); \
2794 while (p < lim) @{ \
2795 if (*p++ != ' ') @{ \
2800 Here backslash-newline is used to split the macro definition, which must
2801 be a single logical line, so that it resembles the way such code would
2802 be laid out if not part of a macro definition.
2804 A call to this macro might be @code{SKIP_SPACES (p, lim)}. Strictly
2805 speaking, the call expands to a compound statement, which is a complete
2806 statement with no need for a semicolon to end it. However, since it
2807 looks like a function call, it minimizes confusion if you can use it
2808 like a function call, writing a semicolon afterward, as in
2809 @code{SKIP_SPACES (p, lim);}
2811 This can cause trouble before @code{else} statements, because the
2812 semicolon is actually a null statement. Suppose you write
2816 SKIP_SPACES (p, lim);
2821 The presence of two statements---the compound statement and a null
2822 statement---in between the @code{if} condition and the @code{else}
2823 makes invalid C code.
2825 The definition of the macro @code{SKIP_SPACES} can be altered to solve
2826 this problem, using a @code{do @dots{} while} statement. Here is how:
2829 #define SKIP_SPACES(p, limit) \
2830 do @{ char *lim = (limit); \
2831 while (p < lim) @{ \
2832 if (*p++ != ' ') @{ \
2833 p--; break; @}@}@} \
2837 Now @code{SKIP_SPACES (p, lim);} expands into
2840 do @{@dots{}@} while (0);
2844 which is one statement. The loop executes exactly once; most compilers
2845 generate no extra code for it.
2847 @node Duplication of Side Effects
2848 @subsection Duplication of Side Effects
2850 @cindex side effects (in macro arguments)
2851 @cindex unsafe macros
2852 Many C programs define a macro @code{min}, for ``minimum'', like this:
2855 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
2858 When you use this macro with an argument containing a side effect,
2862 next = min (x + y, foo (z));
2866 it expands as follows:
2869 next = ((x + y) < (foo (z)) ? (x + y) : (foo (z)));
2873 where @code{x + y} has been substituted for @code{X} and @code{foo (z)}
2876 The function @code{foo} is used only once in the statement as it appears
2877 in the program, but the expression @code{foo (z)} has been substituted
2878 twice into the macro expansion. As a result, @code{foo} might be called
2879 two times when the statement is executed. If it has side effects or if
2880 it takes a long time to compute, the results might not be what you
2881 intended. We say that @code{min} is an @dfn{unsafe} macro.
2883 The best solution to this problem is to define @code{min} in a way that
2884 computes the value of @code{foo (z)} only once. The C language offers
2885 no standard way to do this, but it can be done with GNU extensions as
2890 (@{ typeof (X) x_ = (X); \
2891 typeof (Y) y_ = (Y); \
2892 (x_ < y_) ? x_ : y_; @})
2895 The @samp{(@{ @dots{} @})} notation produces a compound statement that
2896 acts as an expression. Its value is the value of its last statement.
2897 This permits us to define local variables and assign each argument to
2898 one. The local variables have underscores after their names to reduce
2899 the risk of conflict with an identifier of wider scope (it is impossible
2900 to avoid this entirely). Now each argument is evaluated exactly once.
2902 If you do not wish to use GNU C extensions, the only solution is to be
2903 careful when @emph{using} the macro @code{min}. For example, you can
2904 calculate the value of @code{foo (z)}, save it in a variable, and use
2905 that variable in @code{min}:
2909 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
2913 next = min (x + y, tem);
2919 (where we assume that @code{foo} returns type @code{int}).
2921 @node Self-Referential Macros
2922 @subsection Self-Referential Macros
2923 @cindex self-reference
2925 A @dfn{self-referential} macro is one whose name appears in its
2926 definition. Recall that all macro definitions are rescanned for more
2927 macros to replace. If the self-reference were considered a use of the
2928 macro, it would produce an infinitely large expansion. To prevent this,
2929 the self-reference is not considered a macro call. It is passed into
2930 the preprocessor output unchanged. Consider an example:
2933 #define foo (4 + foo)
2937 where @code{foo} is also a variable in your program.
2939 Following the ordinary rules, each reference to @code{foo} will expand
2940 into @code{(4 + foo)}; then this will be rescanned and will expand into
2941 @code{(4 + (4 + foo))}; and so on until the computer runs out of memory.
2943 The self-reference rule cuts this process short after one step, at
2944 @code{(4 + foo)}. Therefore, this macro definition has the possibly
2945 useful effect of causing the program to add 4 to the value of @code{foo}
2946 wherever @code{foo} is referred to.
2948 In most cases, it is a bad idea to take advantage of this feature. A
2949 person reading the program who sees that @code{foo} is a variable will
2950 not expect that it is a macro as well. The reader will come across the
2951 identifier @code{foo} in the program and think its value should be that
2952 of the variable @code{foo}, whereas in fact the value is four greater.
2954 One common, useful use of self-reference is to create a macro which
2955 expands to itself. If you write
2962 then the macro @code{EPERM} expands to @code{EPERM}. Effectively, it is
2963 left alone by the preprocessor whenever it's used in running text. You
2964 can tell that it's a macro with @samp{#ifdef}. You might do this if you
2965 want to define numeric constants with an @code{enum}, but have
2966 @samp{#ifdef} be true for each constant.
2968 If a macro @code{x} expands to use a macro @code{y}, and the expansion of
2969 @code{y} refers to the macro @code{x}, that is an @dfn{indirect
2970 self-reference} of @code{x}. @code{x} is not expanded in this case
2971 either. Thus, if we have
2979 then @code{x} and @code{y} expand as follows:
2983 x @expansion{} (4 + y)
2984 @expansion{} (4 + (2 * x))
2986 y @expansion{} (2 * x)
2987 @expansion{} (2 * (4 + y))
2992 Each macro is expanded when it appears in the definition of the other
2993 macro, but not when it indirectly appears in its own definition.
2995 @node Argument Prescan
2996 @subsection Argument Prescan
2997 @cindex expansion of arguments
2998 @cindex macro argument expansion
2999 @cindex prescan of macro arguments
3001 Macro arguments are completely macro-expanded before they are
3002 substituted into a macro body, unless they are stringized or pasted
3003 with other tokens. After substitution, the entire macro body, including
3004 the substituted arguments, is scanned again for macros to be expanded.
3005 The result is that the arguments are scanned @emph{twice} to expand
3006 macro calls in them.
3008 Most of the time, this has no effect. If the argument contained any
3009 macro calls, they are expanded during the first scan. The result
3010 therefore contains no macro calls, so the second scan does not change
3011 it. If the argument were substituted as given, with no prescan, the
3012 single remaining scan would find the same macro calls and produce the
3015 You might expect the double scan to change the results when a
3016 self-referential macro is used in an argument of another macro
3017 (@pxref{Self-Referential Macros}): the self-referential macro would be
3018 expanded once in the first scan, and a second time in the second scan.
3019 However, this is not what happens. The self-references that do not
3020 expand in the first scan are marked so that they will not expand in the
3023 You might wonder, ``Why mention the prescan, if it makes no difference?
3024 And why not skip it and make the preprocessor faster?'' The answer is
3025 that the prescan does make a difference in three special cases:
3029 Nested calls to a macro.
3031 We say that @dfn{nested} calls to a macro occur when a macro's argument
3032 contains a call to that very macro. For example, if @code{f} is a macro
3033 that expects one argument, @code{f (f (1))} is a nested pair of calls to
3034 @code{f}. The desired expansion is made by expanding @code{f (1)} and
3035 substituting that into the definition of @code{f}. The prescan causes
3036 the expected result to happen. Without the prescan, @code{f (1)} itself
3037 would be substituted as an argument, and the inner use of @code{f} would
3038 appear during the main scan as an indirect self-reference and would not
3042 Macros that call other macros that stringize or concatenate.
3044 If an argument is stringized or concatenated, the prescan does not
3045 occur. If you @emph{want} to expand a macro, then stringize or
3046 concatenate its expansion, you can do that by causing one macro to call
3047 another macro that does the stringizing or concatenation. For
3048 instance, if you have
3051 #define AFTERX(x) X_ ## x
3052 #define XAFTERX(x) AFTERX(x)
3053 #define TABLESIZE 1024
3054 #define BUFSIZE TABLESIZE
3057 then @code{AFTERX(BUFSIZE)} expands to @code{X_BUFSIZE}, and
3058 @code{XAFTERX(BUFSIZE)} expands to @code{X_1024}. (Not to
3059 @code{X_TABLESIZE}. Prescan always does a complete expansion.)
3062 Macros used in arguments, whose expansions contain unshielded commas.
3064 This can cause a macro expanded on the second scan to be called with the
3065 wrong number of arguments. Here is an example:
3069 #define bar(x) lose(x)
3070 #define lose(x) (1 + (x))
3073 We would like @code{bar(foo)} to turn into @code{(1 + (foo))}, which
3074 would then turn into @code{(1 + (a,b))}. Instead, @code{bar(foo)}
3075 expands into @code{lose(a,b)}, and you get an error because @code{lose}
3076 requires a single argument. In this case, the problem is easily solved
3077 by the same parentheses that ought to be used to prevent misnesting of
3078 arithmetic operations:
3083 #define bar(x) lose((x))
3086 The extra pair of parentheses prevents the comma in @code{foo}'s
3087 definition from being interpreted as an argument separator.
3091 @node Newlines in Arguments
3092 @subsection Newlines in Arguments
3093 @cindex newlines in macro arguments
3095 The invocation of a function-like macro can extend over many logical
3096 lines. However, in the present implementation, the entire expansion
3097 comes out on one line. Thus line numbers emitted by the compiler or
3098 debugger refer to the line the invocation started on, which might be
3099 different to the line containing the argument causing the problem.
3101 Here is an example illustrating this:
3104 #define ignore_second_arg(a,b,c) a; c
3106 ignore_second_arg (foo (),
3112 The syntax error triggered by the tokens @code{syntax error} results in
3113 an error message citing line three---the line of ignore_second_arg---
3114 even though the problematic code comes from line five.
3116 We consider this a bug, and intend to fix it in the near future.
3119 @chapter Conditionals
3120 @cindex conditionals
3122 A @dfn{conditional} is a directive that instructs the preprocessor to
3123 select whether or not to include a chunk of code in the final token
3124 stream passed to the compiler. Preprocessor conditionals can test
3125 arithmetic expressions, or whether a name is defined as a macro, or both
3126 simultaneously using the special @code{defined} operator.
3128 A conditional in the C preprocessor resembles in some ways an @code{if}
3129 statement in C, but it is important to understand the difference between
3130 them. The condition in an @code{if} statement is tested during the
3131 execution of your program. Its purpose is to allow your program to
3132 behave differently from run to run, depending on the data it is
3133 operating on. The condition in a preprocessing conditional directive is
3134 tested when your program is compiled. Its purpose is to allow different
3135 code to be included in the program depending on the situation at the
3136 time of compilation.
3138 However, the distinction is becoming less clear. Modern compilers often
3139 do test @code{if} statements when a program is compiled, if their
3140 conditions are known not to vary at run time, and eliminate code which
3141 can never be executed. If you can count on your compiler to do this,
3142 you may find that your program is more readable if you use @code{if}
3143 statements with constant conditions (perhaps determined by macros). Of
3144 course, you can only use this to exclude code, not type definitions or
3145 other preprocessing directives, and you can only do it if the code
3146 remains syntactically valid when it is not to be used.
3149 * Conditional Uses::
3150 * Conditional Syntax::
3154 @node Conditional Uses
3155 @section Conditional Uses
3157 There are three general reasons to use a conditional.
3161 A program may need to use different code depending on the machine or
3162 operating system it is to run on. In some cases the code for one
3163 operating system may be erroneous on another operating system; for
3164 example, it might refer to data types or constants that do not exist on
3165 the other system. When this happens, it is not enough to avoid
3166 executing the invalid code. Its mere presence will cause the compiler
3167 to reject the program. With a preprocessing conditional, the offending
3168 code can be effectively excised from the program when it is not valid.
3171 You may want to be able to compile the same source file into two
3172 different programs. One version might make frequent time-consuming
3173 consistency checks on its intermediate data, or print the values of
3174 those data for debugging, and the other not.
3177 A conditional whose condition is always false is one way to exclude code
3178 from the program but keep it as a sort of comment for future reference.
3181 Simple programs that do not need system-specific logic or complex
3182 debugging hooks generally will not need to use preprocessing
3185 @node Conditional Syntax
3186 @section Conditional Syntax
3189 A conditional in the C preprocessor begins with a @dfn{conditional
3190 directive}: @samp{#if}, @samp{#ifdef} or @samp{#ifndef}.
3198 * @code{__has_attribute}::
3199 * @code{__has_cpp_attribute}::
3200 * @code{__has_c_attribute}::
3201 * @code{__has_builtin}::
3202 * @code{__has_feature}::
3203 * @code{__has_extension}::
3204 * @code{__has_include}::
3212 The simplest sort of conditional is
3218 @var{controlled text}
3220 #endif /* @var{MACRO} */
3224 @cindex conditional group
3225 This block is called a @dfn{conditional group}. @var{controlled text}
3226 will be included in the output of the preprocessor if and only if
3227 @var{MACRO} is defined. We say that the conditional @dfn{succeeds} if
3228 @var{MACRO} is defined, @dfn{fails} if it is not.
3230 The @var{controlled text} inside of a conditional can include
3231 preprocessing directives. They are executed only if the conditional
3232 succeeds. You can nest conditional groups inside other conditional
3233 groups, but they must be completely nested. In other words,
3234 @samp{#endif} always matches the nearest @samp{#ifdef} (or
3235 @samp{#ifndef}, or @samp{#if}). Also, you cannot start a conditional
3236 group in one file and end it in another.
3238 Even if a conditional fails, the @var{controlled text} inside it is
3239 still run through initial transformations and tokenization. Therefore,
3240 it must all be lexically valid C@. Normally the only way this matters is
3241 that all comments and string literals inside a failing conditional group
3242 must still be properly ended.
3244 The comment following the @samp{#endif} is not required, but it is a
3245 good practice if there is a lot of @var{controlled text}, because it
3246 helps people match the @samp{#endif} to the corresponding @samp{#ifdef}.
3247 Older programs sometimes put @var{MACRO} directly after the
3248 @samp{#endif} without enclosing it in a comment. This is invalid code
3249 according to the C standard. CPP accepts it with a warning. It
3250 never affects which @samp{#ifndef} the @samp{#endif} matches.
3253 Sometimes you wish to use some code if a macro is @emph{not} defined.
3254 You can do this by writing @samp{#ifndef} instead of @samp{#ifdef}.
3255 One common use of @samp{#ifndef} is to include code only the first
3256 time a header file is included. @xref{Once-Only Headers}.
3258 Macro definitions can vary between compilations for several reasons.
3259 Here are some samples.
3263 Some macros are predefined on each kind of machine
3264 (@pxref{System-specific Predefined Macros}). This allows you to provide
3265 code specially tuned for a particular machine.
3268 System header files define more macros, associated with the features
3269 they implement. You can test these macros with conditionals to avoid
3270 using a system feature on a machine where it is not implemented.
3273 Macros can be defined or undefined with the @option{-D} and @option{-U}
3274 command-line options when you compile the program. You can arrange to
3275 compile the same source file into two different programs by choosing a
3276 macro name to specify which program you want, writing conditionals to
3277 test whether or how this macro is defined, and then controlling the
3278 state of the macro with command-line options, perhaps set in the
3279 Makefile. @xref{Invocation}.
3282 Your program might have a special header file (often called
3283 @file{config.h}) that is adjusted when the program is compiled. It can
3284 define or not define macros depending on the features of the system and
3285 the desired capabilities of the program. The adjustment can be
3286 automated by a tool such as @command{autoconf}, or done by hand.
3292 The @samp{#if} directive allows you to test the value of an arithmetic
3293 expression, rather than the mere existence of one macro. Its syntax is
3297 #if @var{expression}
3299 @var{controlled text}
3301 #endif /* @var{expression} */
3305 @var{expression} is a C expression of integer type, subject to stringent
3306 restrictions. It may contain
3313 Character constants, which are interpreted as they would be in normal
3317 Arithmetic operators for addition, subtraction, multiplication,
3318 division, bitwise operations, shifts, comparisons, and logical
3319 operations (@code{&&} and @code{||}). The latter two obey the usual
3320 short-circuiting rules of standard C@.
3323 Macros. All macros in the expression are expanded before actual
3324 computation of the expression's value begins.
3327 Uses of the @code{defined} operator, which lets you check whether macros
3328 are defined in the middle of an @samp{#if}.
3331 Identifiers that are not macros, which are all considered to be the
3332 number zero. This allows you to write @code{@w{#if MACRO}} instead of
3333 @code{@w{#ifdef MACRO}}, if you know that MACRO, when defined, will
3334 always have a nonzero value. Function-like macros used without their
3335 function call parentheses are also treated as zero.
3337 In some contexts this shortcut is undesirable. The @option{-Wundef}
3338 option causes GCC to warn whenever it encounters an identifier which is
3339 not a macro in an @samp{#if}.
3342 The preprocessor does not know anything about types in the language.
3343 Therefore, @code{sizeof} operators are not recognized in @samp{#if}, and
3344 neither are @code{enum} constants. They will be taken as identifiers
3345 which are not macros, and replaced by zero. In the case of
3346 @code{sizeof}, this is likely to cause the expression to be invalid.
3348 The preprocessor calculates the value of @var{expression}. It carries
3349 out all calculations in the widest integer type known to the compiler;
3350 on most machines supported by GCC this is 64 bits. This is not the same
3351 rule as the compiler uses to calculate the value of a constant
3352 expression, and may give different results in some cases. If the value
3353 comes out to be nonzero, the @samp{#if} succeeds and the @var{controlled
3354 text} is included; otherwise it is skipped.
3359 @cindex @code{defined}
3360 The special operator @code{defined} is used in @samp{#if} and
3361 @samp{#elif} expressions to test whether a certain name is defined as a
3362 macro. @code{defined @var{name}} and @code{defined (@var{name})} are
3363 both expressions whose value is 1 if @var{name} is defined as a macro at
3364 the current point in the program, and 0 otherwise. Thus, @code{@w{#if
3365 defined MACRO}} is precisely equivalent to @code{@w{#ifdef MACRO}}.
3367 @code{defined} is useful when you wish to test more than one macro for
3368 existence at once. For example,
3371 #if defined (__vax__) || defined (__ns16000__)
3375 would succeed if either of the names @code{__vax__} or
3376 @code{__ns16000__} is defined as a macro.
3378 Conditionals written like this:
3381 #if defined BUFSIZE && BUFSIZE >= 1024
3385 can generally be simplified to just @code{@w{#if BUFSIZE >= 1024}},
3386 since if @code{BUFSIZE} is not defined, it will be interpreted as having
3389 If the @code{defined} operator appears as a result of a macro expansion,
3390 the C standard says the behavior is undefined. GNU cpp treats it as a
3391 genuine @code{defined} operator and evaluates it normally. It will warn
3392 wherever your code uses this feature if you use the command-line option
3393 @option{-Wpedantic}, since other compilers may handle it differently. The
3394 warning is also enabled by @option{-Wextra}, and can also be enabled
3395 individually with @option{-Wexpansion-to-defined}.
3401 The @samp{#else} directive can be added to a conditional to provide
3402 alternative text to be used if the condition fails. This is what it
3407 #if @var{expression}
3409 #else /* Not @var{expression} */
3411 #endif /* Not @var{expression} */
3416 If @var{expression} is nonzero, the @var{text-if-true} is included and
3417 the @var{text-if-false} is skipped. If @var{expression} is zero, the
3420 You can use @samp{#else} with @samp{#ifdef} and @samp{#ifndef}, too.
3426 One common case of nested conditionals is used to check for more than two
3427 possible alternatives. For example, you might have
3441 Another conditional directive, @samp{#elif}, allows this to be
3442 abbreviated as follows:
3449 #else /* X != 2 and X != 1*/
3451 #endif /* X != 2 and X != 1*/
3454 @samp{#elif} stands for ``else if''. Like @samp{#else}, it goes in the
3455 middle of a conditional group and subdivides it; it does not require a
3456 matching @samp{#endif} of its own. Like @samp{#if}, the @samp{#elif}
3457 directive includes an expression to be tested. The text following the
3458 @samp{#elif} is processed only if the original @samp{#if}-condition
3459 failed and the @samp{#elif} condition succeeds.
3461 More than one @samp{#elif} can go in the same conditional group. Then
3462 the text after each @samp{#elif} is processed only if the @samp{#elif}
3463 condition succeeds after the original @samp{#if} and all previous
3464 @samp{#elif} directives within it have failed.
3466 @samp{#else} is allowed after any number of @samp{#elif} directives, but
3467 @samp{#elif} may not follow @samp{#else}.
3469 @node @code{__has_attribute}
3470 @subsection @code{__has_attribute}
3471 @cindex @code{__has_attribute}
3473 The special operator @code{__has_attribute (@var{operand})} may be used
3474 in @samp{#if} and @samp{#elif} expressions to test whether the attribute
3475 referenced by its @var{operand} is recognized by GCC. Using the operator
3476 in other contexts is not valid. In C code, if compiling for strict
3477 conformance to standards before C23, @var{operand} must be
3478 a valid identifier. Otherwise, @var{operand} may be optionally
3479 introduced by the @code{@var{attribute-scope}::} prefix.
3480 The @var{attribute-scope} prefix identifies the ``namespace'' within
3481 which the attribute is recognized. The scope of GCC attributes is
3482 @samp{gnu} or @samp{__gnu__}. The @code{__has_attribute} operator by
3483 itself, without any @var{operand} or parentheses, acts as a predefined
3484 macro so that support for it can be tested in portable code. Thus,
3485 the recommended use of the operator is as follows:
3488 #if defined __has_attribute
3489 # if __has_attribute (nonnull)
3490 # define ATTR_NONNULL __attribute__ ((nonnull))
3495 The first @samp{#if} test succeeds only when the operator is supported
3496 by the version of GCC (or another compiler) being used. Only when that
3497 test succeeds is it valid to use @code{__has_attribute} as a preprocessor
3498 operator. As a result, combining the two tests into a single expression as
3499 shown below would only be valid with a compiler that supports the operator
3500 but not with others that don't.
3503 #if defined __has_attribute && __has_attribute (nonnull) /* not portable */
3508 @node @code{__has_cpp_attribute}
3509 @subsection @code{__has_cpp_attribute}
3510 @cindex @code{__has_cpp_attribute}
3512 The special operator @code{__has_cpp_attribute (@var{operand})} may be used
3513 in @samp{#if} and @samp{#elif} expressions in C++ code to test whether
3514 the attribute referenced by its @var{operand} is recognized by GCC.
3515 @code{__has_cpp_attribute (@var{operand})} is equivalent to
3516 @code{__has_attribute (@var{operand})} except that when @var{operand}
3517 designates a supported standard attribute it evaluates to an integer
3518 constant of the form @code{YYYYMM} indicating the year and month when
3519 the attribute was first introduced into the C++ standard. For additional
3520 information including the dates of the introduction of current standard
3521 attributes, see @w{@uref{https://isocpp.org/std/standing-documents/sd-6-sg10-feature-test-recommendations/,
3522 SD-6: SG10 Feature Test Recommendations}}.
3524 @node @code{__has_c_attribute}
3525 @subsection @code{__has_c_attribute}
3526 @cindex @code{__has_c_attribute}
3528 The special operator @code{__has_c_attribute (@var{operand})} may be
3529 used in @samp{#if} and @samp{#elif} expressions in C code to test
3530 whether the attribute referenced by its @var{operand} is recognized by
3531 GCC in attributes using the @samp{[[]]} syntax. GNU attributes must
3532 be specified with the scope @samp{gnu} or @samp{__gnu__} with
3533 @code{__has_c_attribute}. When @var{operand} designates a supported
3534 standard attribute it evaluates to an integer constant of the form
3535 @code{YYYYMM} indicating the year and month when the attribute was
3536 first introduced into the C standard, or when the syntax of operands
3537 to the attribute was extended in the C standard.
3539 @node @code{__has_builtin}
3540 @subsection @code{__has_builtin}
3541 @cindex @code{__has_builtin}
3543 The special operator @code{__has_builtin (@var{operand})} may be used in
3544 constant integer contexts and in preprocessor @samp{#if} and @samp{#elif}
3545 expressions to test whether the symbol named by its @var{operand} is
3546 recognized as a built-in function by GCC in the current language and
3547 conformance mode. It evaluates to a constant integer with a nonzero
3548 value if the argument refers to such a function, and to zero otherwise.
3549 The operator may also be used in preprocessor @samp{#if} and @samp{#elif}
3550 expressions. The @code{__has_builtin} operator by itself, without any
3551 @var{operand} or parentheses, acts as a predefined macro so that support
3552 for it can be tested in portable code. Thus, the recommended use of
3553 the operator is as follows:
3556 #if defined __has_builtin
3557 # if __has_builtin (__builtin_object_size)
3558 # define builtin_object_size(ptr) __builtin_object_size (ptr, 2)
3561 #ifndef builtin_object_size
3562 # define builtin_object_size(ptr) ((size_t)-1)
3566 @node @code{__has_feature}
3567 @subsection @code{__has_feature}
3568 @cindex @code{__has_feature}
3570 The special operator @code{__has_feature (@var{operand})} may be used in
3571 constant integer contexts and in preprocessor @samp{#if} and @samp{#elif}
3572 expressions to test whether the identifier given in @var{operand} is recognized
3573 as a feature supported by GCC given the current options and, in the case of
3574 standard language features, whether the feature is available in the chosen
3575 version of the language standard.
3577 Note that @code{__has_feature} and @code{__has_extension} are not recommended
3578 for use in new code, and are only provided for compatibility with Clang. For
3579 details of which identifiers are accepted by these function-like macros, see
3580 @w{@uref{https://clang.llvm.org/docs/LanguageExtensions.html#has-feature-and-has-extension,
3581 the Clang documentation}}.
3583 @node @code{__has_extension}
3584 @subsection @code{__has_extension}
3585 @cindex @code{__has_extension}
3587 The special operator @code{__has_extension (@var{operand})} may be used in
3588 constant integer contexts and in preprocessor @samp{#if} and @samp{#elif}
3589 expressions to test whether the identifier given in @var{operand} is recognized
3590 as an extension supported by GCC given the current options. In any given
3591 context, the features accepted by @code{__has_extension} are a strict superset
3592 of those accepted by @code{__has_feature}. Unlike @code{__has_feature},
3593 @code{__has_extension} tests whether a given feature is available regardless of
3594 strict language standards conformance.
3596 If the @option{-pedantic-errors} flag is given, @code{__has_extension} is
3597 equivalent to @code{__has_feature}.
3599 Note that @code{__has_feature} and @code{__has_extension} are not recommended
3600 for use in new code, and are only provided for compatibility with Clang. For
3601 details of which identifiers are accepted by these function-like macros, see
3602 @w{@uref{https://clang.llvm.org/docs/LanguageExtensions.html#has-feature-and-has-extension,
3603 the Clang documentation}}.
3605 @node @code{__has_include}
3606 @subsection @code{__has_include}
3607 @cindex @code{__has_include}
3609 The special operator @code{__has_include (@var{operand})} may be used in
3610 @samp{#if} and @samp{#elif} expressions to test whether the header referenced
3611 by its @var{operand} can be included using the @samp{#include} directive. Using
3612 the operator in other contexts is not valid. The @var{operand} takes
3613 the same form as the file in the @samp{#include} directive (@pxref{Include
3614 Syntax}) and evaluates to a nonzero value if the header can be included and
3615 to zero otherwise. Note that that the ability to include a header doesn't
3616 imply that the header doesn't contain invalid constructs or @samp{#error}
3617 directives that would cause the preprocessor to fail.
3619 The @code{__has_include} operator by itself, without any @var{operand} or
3620 parentheses, acts as a predefined macro so that support for it can be tested
3621 in portable code. Thus, the recommended use of the operator is as follows:
3624 #if defined __has_include
3625 # if __has_include (<stdatomic.h>)
3626 # include <stdatomic.h>
3631 The first @samp{#if} test succeeds only when the operator is supported
3632 by the version of GCC (or another compiler) being used. Only when that
3633 test succeeds is it valid to use @code{__has_include} as a preprocessor
3634 operator. As a result, combining the two tests into a single expression
3635 as shown below would only be valid with a compiler that supports the operator
3636 but not with others that don't.
3639 #if defined __has_include && __has_include ("header.h") /* not portable */
3645 @section Deleted Code
3646 @cindex commenting out code
3648 If you replace or delete a part of the program but want to keep the old
3649 code around for future reference, you often cannot simply comment it
3650 out. Block comments do not nest, so the first comment inside the old
3651 code will end the commenting-out. The probable result is a flood of
3654 One way to avoid this problem is to use an always-false conditional
3655 instead. For instance, put @code{#if 0} before the deleted code and
3656 @code{#endif} after it. This works even if the code being turned
3657 off contains conditionals, but they must be entire conditionals
3658 (balanced @samp{#if} and @samp{#endif}).
3660 Some people use @code{#ifdef notdef} instead. This is risky, because
3661 @code{notdef} might be accidentally defined as a macro, and then the
3662 conditional would succeed. @code{#if 0} can be counted on to fail.
3664 Do not use @code{#if 0} for comments which are not C code. Use a real
3665 comment, instead. The interior of @code{#if 0} must consist of complete
3666 tokens; in particular, single-quote characters must balance. Comments
3667 often contain unbalanced single-quote characters (known in English as
3668 apostrophes). These confuse @code{#if 0}. They don't confuse
3672 @chapter Diagnostics
3674 @cindex reporting errors
3675 @cindex reporting warnings
3678 The directive @samp{#error} causes the preprocessor to report a fatal
3679 error. The tokens forming the rest of the line following @samp{#error}
3680 are used as the error message.
3682 You would use @samp{#error} inside of a conditional that detects a
3683 combination of parameters which you know the program does not properly
3684 support. For example, if you know that the program will not run
3685 properly on a VAX, you might write
3690 #error "Won't work on VAXen. See comments at get_last_object."
3695 If you have several configuration parameters that must be set up by
3696 the installation in a consistent way, you can use conditionals to detect
3697 an inconsistency and report it with @samp{#error}. For example,
3700 #if !defined(FOO) && defined(BAR)
3701 #error "BAR requires FOO."
3706 The directive @samp{#warning} is like @samp{#error}, but causes the
3707 preprocessor to issue a warning and continue preprocessing. The tokens
3708 following @samp{#warning} are used as the warning message.
3710 You might use @samp{#warning} in obsolete header files, with a message
3711 directing the user to the header file which should be used instead.
3713 Neither @samp{#error} nor @samp{#warning} macro-expands its argument.
3714 Internal whitespace sequences are each replaced with a single space.
3715 The line must consist of complete tokens. It is wisest to make the
3716 argument of these directives be a single string constant; this avoids
3717 problems with apostrophes and the like.
3720 @chapter Line Control
3721 @cindex line control
3723 The C preprocessor informs the C compiler of the location in your source
3724 code where each token came from. Presently, this is just the file name
3725 and line number. All the tokens resulting from macro expansion are
3726 reported as having appeared on the line of the source file where the
3727 outermost macro was used. We intend to be more accurate in the future.
3729 If you write a program which generates source code, such as the
3730 @command{bison} parser generator, you may want to adjust the preprocessor's
3731 notion of the current file name and line number by hand. Parts of the
3732 output from @command{bison} are generated from scratch, other parts come
3733 from a standard parser file. The rest are copied verbatim from
3734 @command{bison}'s input. You would like compiler error messages and
3735 symbolic debuggers to be able to refer to @code{bison}'s input file.
3738 @command{bison} or any such program can arrange this by writing
3739 @samp{#line} directives into the output file. @samp{#line} is a
3740 directive that specifies the original line number and source file name
3741 for subsequent input in the current preprocessor input file.
3742 @samp{#line} has three variants:
3745 @item #line @var{linenum}
3746 @var{linenum} is a non-negative decimal integer constant. It specifies
3747 the line number which should be reported for the following line of
3748 input. Subsequent lines are counted from @var{linenum}.
3750 @item #line @var{linenum} @var{filename}
3751 @var{linenum} is the same as for the first form, and has the same
3752 effect. In addition, @var{filename} is a string constant. The
3753 following line and all subsequent lines are reported to come from the
3754 file it specifies, until something else happens to change that.
3755 @var{filename} is interpreted according to the normal rules for a string
3756 constant: backslash escapes are interpreted. This is different from
3759 @item #line @var{anything else}
3760 @var{anything else} is checked for macro calls, which are expanded.
3761 The result should match one of the above two forms.
3764 @samp{#line} directives alter the results of the @code{__FILE__} and
3765 @code{__LINE__} predefined macros from that point on. @xref{Standard
3766 Predefined Macros}. They do not have any effect on @samp{#include}'s
3767 idea of the directory containing the current file.
3772 @cindex pragma directive
3774 The @samp{#pragma} directive is the method specified by the C standard
3775 for providing additional information to the compiler, beyond what is
3776 conveyed in the language itself. The forms of this directive
3777 (commonly known as @dfn{pragmas}) specified by C standard are prefixed with
3778 @code{STDC}. A C compiler is free to attach any meaning it likes to other
3779 pragmas. Most GNU-defined, supported pragmas have been given a
3782 @cindex @code{_Pragma}
3783 C99 introduced the @code{@w{_Pragma}} operator. This feature addresses a
3784 major problem with @samp{#pragma}: being a directive, it cannot be
3785 produced as the result of macro expansion. @code{@w{_Pragma}} is an
3786 operator, much like @code{sizeof} or @code{defined}, and can be embedded
3789 Its syntax is @code{@w{_Pragma (@var{string-literal})}}, where
3790 @var{string-literal} can be either a normal or wide-character string
3791 literal. It is destringized, by replacing all @samp{\\} with a single
3792 @samp{\} and all @samp{\"} with a @samp{"}. The result is then
3793 processed as if it had appeared as the right hand side of a
3794 @samp{#pragma} directive. For example,
3797 _Pragma ("GCC dependency \"parse.y\"")
3801 has the same effect as @code{#pragma GCC dependency "parse.y"}. The
3802 same effect could be achieved using macros, for example
3805 #define DO_PRAGMA(x) _Pragma (#x)
3806 DO_PRAGMA (GCC dependency "parse.y")
3809 The standard is unclear on where a @code{_Pragma} operator can appear.
3810 The preprocessor does not accept it within a preprocessing conditional
3811 directive like @samp{#if}. To be safe, you are probably best keeping it
3812 out of directives other than @samp{#define}, and putting it on a line of
3815 This manual documents the pragmas which are meaningful to the
3816 preprocessor itself. Other pragmas are meaningful to the C or C++
3817 compilers. They are documented in the GCC manual.
3819 GCC plugins may provide their own pragmas.
3822 @item #pragma GCC dependency
3823 @code{#pragma GCC dependency} allows you to check the relative dates of
3824 the current file and another file. If the other file is more recent than
3825 the current file, a warning is issued. This is useful if the current
3826 file is derived from the other file, and should be regenerated. The
3827 other file is searched for using the normal include search path.
3828 Optional trailing text can be used to give more information in the
3832 #pragma GCC dependency "parse.y"
3833 #pragma GCC dependency "/usr/include/time.h" rerun fixincludes
3836 @item #pragma GCC poison
3837 Sometimes, there is an identifier that you want to remove completely
3838 from your program, and make sure that it never creeps back in. To
3839 enforce this, you can @dfn{poison} the identifier with this pragma.
3840 @code{#pragma GCC poison} is followed by a list of identifiers to
3841 poison. If any of those identifiers appears anywhere in the source
3842 after the directive, it is a hard error. For example,
3845 #pragma GCC poison printf sprintf fprintf
3846 sprintf(some_string, "hello");
3850 will produce an error.
3852 If a poisoned identifier appears as part of the expansion of a macro
3853 which was defined before the identifier was poisoned, it will @emph{not}
3854 cause an error. This lets you poison an identifier without worrying
3855 about system headers defining macros that use it.
3860 #define strrchr rindex
3861 #pragma GCC poison rindex
3862 strrchr(some_string, 'h');
3866 will not produce an error.
3868 @item #pragma GCC system_header
3869 This pragma takes no arguments. It causes the rest of the code in the
3870 current file to be treated as if it came from a system header.
3871 @xref{System Headers}.
3873 @item #pragma GCC warning
3874 @itemx #pragma GCC error
3875 @code{#pragma GCC warning "message"} causes the preprocessor to issue
3876 a warning diagnostic with the text @samp{message}. The message
3877 contained in the pragma must be a single string literal. Similarly,
3878 @code{#pragma GCC error "message"} issues an error message. Unlike
3879 the @samp{#warning} and @samp{#error} directives, these pragmas can be
3880 embedded in preprocessor macros using @samp{_Pragma}.
3883 If @code{#pragma once} is seen when scanning a header file, that
3884 file will never be read again, no matter what. It is a less-portable
3885 alternative to using @samp{#ifndef} to guard the contents of header files
3886 against multiple inclusions.
3888 @item #pragma region @{tokens@}...
3889 @itemx #pragma endregion @{tokens@}...
3890 These pragmas are accepted, but have no effect.
3894 @node Other Directives
3895 @chapter Other Directives
3899 The @samp{#ident} directive takes one argument, a string constant. On
3900 some systems, that string constant is copied into a special segment of
3901 the object file. On other systems, the directive is ignored. The
3902 @samp{#sccs} directive is a synonym for @samp{#ident}.
3904 These directives are not part of the C standard, but they are not
3905 official GNU extensions either. What historical information we have
3906 been able to find, suggests they originated with System V@.
3908 @cindex null directive
3909 The @dfn{null directive} consists of a @samp{#} followed by a newline,
3910 with only whitespace (including comments) in between. A null directive
3911 is understood as a preprocessing directive but has no effect on the
3912 preprocessor output. The primary significance of the existence of the
3913 null directive is that an input line consisting of just a @samp{#} will
3914 produce no output, rather than a line of output containing just a
3915 @samp{#}. Supposedly some old C programs contain such lines.
3917 @node Preprocessor Output
3918 @chapter Preprocessor Output
3920 When the C preprocessor is used with the C, C++, or Objective-C
3921 compilers, it is integrated into the compiler and communicates a stream
3922 of binary tokens directly to the compiler's parser. However, it can
3923 also be used in the more conventional standalone mode, where it produces
3925 @c FIXME: Document the library interface.
3927 @cindex output format
3928 The output from the C preprocessor looks much like the input, except
3929 that all preprocessing directive lines have been replaced with blank
3930 lines and all comments with spaces. Long runs of blank lines are
3933 The ISO standard specifies that it is implementation defined whether a
3934 preprocessor preserves whitespace between tokens, or replaces it with
3935 e.g.@: a single space. In GNU CPP, whitespace between tokens is collapsed
3936 to become a single space, with the exception that the first token on a
3937 non-directive line is preceded with sufficient spaces that it appears in
3938 the same column in the preprocessed output that it appeared in the
3939 original source file. This is so the output is easy to read.
3940 CPP does not insert any
3941 whitespace where there was none in the original source, except where
3942 necessary to prevent an accidental token paste.
3945 Source file name and line number information is conveyed by lines
3949 # @var{linenum} @var{filename} @var{flags}
3953 These are called @dfn{linemarkers}. They are inserted as needed into
3954 the output (but never within a string or character constant). They mean
3955 that the following line originated in file @var{filename} at line
3956 @var{linenum}. @var{filename} will never contain any non-printing
3957 characters; they are replaced with octal escape sequences.
3959 After the file name comes zero or more flags, which are @samp{1},
3960 @samp{2}, @samp{3}, or @samp{4}. If there are multiple flags, spaces
3961 separate them. Here is what the flags mean:
3965 This indicates the start of a new file.
3967 This indicates returning to a file (after having included another file).
3969 This indicates that the following text comes from a system header file,
3970 so certain warnings should be suppressed.
3972 This indicates that the following text should be treated as being
3973 wrapped in an implicit @code{extern "C"} block.
3974 @c maybe cross reference SYSTEM_IMPLICIT_EXTERN_C
3977 As an extension, the preprocessor accepts linemarkers in non-assembler
3978 input files. They are treated like the corresponding @samp{#line}
3979 directive, (@pxref{Line Control}), except that trailing flags are
3980 permitted, and are interpreted with the meanings described above. If
3981 multiple flags are given, they must be in ascending order.
3983 Some directives may be duplicated in the output of the preprocessor.
3984 These are @samp{#ident} (always), @samp{#pragma} (only if the
3985 preprocessor does not handle the pragma itself), and @samp{#define} and
3986 @samp{#undef} (with certain debugging options). If this happens, the
3987 @samp{#} of the directive will always be in the first column, and there
3988 will be no space between the @samp{#} and the directive name. If macro
3989 expansion happens to generate tokens which might be mistaken for a
3990 duplicated directive, a space will be inserted between the @samp{#} and
3993 @node Traditional Mode
3994 @chapter Traditional Mode
3996 Traditional (pre-standard) C preprocessing is rather different from
3997 the preprocessing specified by the standard. When the preprocessor
3999 @option{-traditional-cpp} option, it attempts to emulate a traditional
4002 This mode is not useful for compiling C code with GCC,
4003 but is intended for use with non-C preprocessing applications. Thus
4004 traditional mode semantics are supported only when invoking
4005 the preprocessor explicitly, and not in the compiler front ends.
4007 The implementation does not correspond precisely to the behavior of
4008 early pre-standard versions of GCC, nor to any true traditional preprocessor.
4009 After all, inconsistencies among traditional implementations were a
4010 major motivation for C standardization. However, we intend that it
4011 should be compatible with true traditional preprocessors in all ways
4012 that actually matter.
4015 * Traditional lexical analysis::
4016 * Traditional macros::
4017 * Traditional miscellany::
4018 * Traditional warnings::
4021 @node Traditional lexical analysis
4022 @section Traditional lexical analysis
4024 The traditional preprocessor does not decompose its input into tokens
4025 the same way a standards-conforming preprocessor does. The input is
4026 simply treated as a stream of text with minimal internal form.
4028 This implementation does not treat trigraphs (@pxref{trigraphs})
4029 specially since they were an invention of the standards committee. It
4030 handles arbitrarily-positioned escaped newlines properly and splices
4031 the lines as you would expect; many traditional preprocessors did not
4034 The form of horizontal whitespace in the input file is preserved in
4035 the output. In particular, hard tabs remain hard tabs. This can be
4036 useful if, for example, you are preprocessing a Makefile.
4038 Traditional CPP only recognizes C-style block comments, and treats the
4039 @samp{/*} sequence as introducing a comment only if it lies outside
4040 quoted text. Quoted text is introduced by the usual single and double
4041 quotes, and also by an initial @samp{<} in a @code{#include}
4044 Traditionally, comments are completely removed and are not replaced
4045 with a space. Since a traditional compiler does its own tokenization
4046 of the output of the preprocessor, this means that comments can
4047 effectively be used as token paste operators. However, comments
4048 behave like separators for text handled by the preprocessor itself,
4049 since it doesn't re-lex its input. For example, in
4056 @samp{foo} and @samp{bar} are distinct identifiers and expanded
4057 separately if they happen to be macros. In other words, this
4058 directive is equivalent to
4071 Generally speaking, in traditional mode an opening quote need not have
4072 a matching closing quote. In particular, a macro may be defined with
4073 replacement text that contains an unmatched quote. Of course, if you
4074 attempt to compile preprocessed output containing an unmatched quote
4075 you will get a syntax error.
4077 However, all preprocessing directives other than @code{#define}
4078 require matching quotes. For example:
4081 #define m This macro's fine and has an unmatched quote
4082 "/* This is not a comment. */
4083 /* @r{This is a comment. The following #include directive
4088 Just as for the ISO preprocessor, what would be a closing quote can be
4089 escaped with a backslash to prevent the quoted text from closing.
4091 @node Traditional macros
4092 @section Traditional macros
4094 The major difference between traditional and ISO macros is that the
4095 former expand to text rather than to a token sequence. CPP removes
4096 all leading and trailing horizontal whitespace from a macro's
4097 replacement text before storing it, but preserves the form of internal
4100 One consequence is that it is legitimate for the replacement text to
4101 contain an unmatched quote (@pxref{Traditional lexical analysis}). An
4102 unclosed string or character constant continues into the text
4103 following the macro call. Similarly, the text at the end of a macro's
4104 expansion can run together with the text after the macro invocation to
4105 produce a single token.
4107 Normally comments are removed from the replacement text after the
4108 macro is expanded, but if the @option{-CC} option is passed on the
4109 command-line comments are preserved. (In fact, the current
4110 implementation removes comments even before saving the macro
4111 replacement text, but it careful to do it in such a way that the
4112 observed effect is identical even in the function-like macro case.)
4114 The ISO stringizing operator @samp{#} and token paste operator
4115 @samp{##} have no special meaning. As explained later, an effect
4116 similar to these operators can be obtained in a different way. Macro
4117 names that are embedded in quotes, either from the main file or after
4118 macro replacement, do not expand.
4120 CPP replaces an unquoted object-like macro name with its replacement
4121 text, and then rescans it for further macros to replace. Unlike
4122 standard macro expansion, traditional macro expansion has no provision
4123 to prevent recursion. If an object-like macro appears unquoted in its
4124 replacement text, it will be replaced again during the rescan pass,
4125 and so on @emph{ad infinitum}. GCC detects when it is expanding
4126 recursive macros, emits an error message, and continues after the
4127 offending macro invocation.
4131 #define INC(x) PLUS+x
4136 Function-like macros are similar in form but quite different in
4137 behavior to their ISO counterparts. Their arguments are contained
4138 within parentheses, are comma-separated, and can cross physical lines.
4139 Commas within nested parentheses are not treated as argument
4140 separators. Similarly, a quote in an argument cannot be left
4141 unclosed; a following comma or parenthesis that comes before the
4142 closing quote is treated like any other character. There is no
4143 facility for handling variadic macros.
4145 This implementation removes all comments from macro arguments, unless
4146 the @option{-C} option is given. The form of all other horizontal
4147 whitespace in arguments is preserved, including leading and trailing
4148 whitespace. In particular
4155 is treated as an invocation of the macro @samp{f} with a single
4156 argument consisting of a single space. If you want to invoke a
4157 function-like macro that takes no arguments, you must not leave any
4158 whitespace between the parentheses.
4160 If a macro argument crosses a new line, the new line is replaced with
4161 a space when forming the argument. If the previous line contained an
4162 unterminated quote, the following line inherits the quoted state.
4164 Traditional preprocessors replace parameters in the replacement text
4165 with their arguments regardless of whether the parameters are within
4166 quotes or not. This provides a way to stringize arguments. For
4171 str(/* @r{A comment} */some text )
4172 @expansion{} "some text "
4176 Note that the comment is removed, but that the trailing space is
4177 preserved. Here is an example of using a comment to effect token
4181 #define suffix(x) foo_/**/x
4183 @expansion{} foo_bar
4186 @node Traditional miscellany
4187 @section Traditional miscellany
4189 Here are some things to be aware of when using the traditional
4194 Preprocessing directives are recognized only when their leading
4195 @samp{#} appears in the first column. There can be no whitespace
4196 between the beginning of the line and the @samp{#}, but whitespace can
4197 follow the @samp{#}.
4200 A true traditional C preprocessor does not recognize @samp{#error} or
4201 @samp{#pragma}, and may not recognize @samp{#elif}. CPP supports all
4202 the directives in traditional mode that it supports in ISO mode,
4203 including extensions, with the exception that the effects of
4204 @samp{#pragma GCC poison} are undefined.
4207 __STDC__ is not defined.
4210 If you use digraphs the behavior is undefined.
4213 If a line that looks like a directive appears within macro arguments,
4214 the behavior is undefined.
4218 @node Traditional warnings
4219 @section Traditional warnings
4220 You can request warnings about features that did not exist, or worked
4221 differently, in traditional C with the @option{-Wtraditional} option.
4222 GCC does not warn about features of ISO C which you must use when you
4223 are using a conforming compiler, such as the @samp{#} and @samp{##}
4226 Presently @option{-Wtraditional} warns about:
4230 Macro parameters that appear within string literals in the macro body.
4231 In traditional C macro replacement takes place within string literals,
4232 but does not in ISO C@.
4235 In traditional C, some preprocessor directives did not exist.
4236 Traditional preprocessors would only consider a line to be a directive
4237 if the @samp{#} appeared in column 1 on the line. Therefore
4238 @option{-Wtraditional} warns about directives that traditional C
4239 understands but would ignore because the @samp{#} does not appear as the
4240 first character on the line. It also suggests you hide directives like
4241 @samp{#pragma} not understood by traditional C by indenting them. Some
4242 traditional implementations would not recognize @samp{#elif}, so it
4243 suggests avoiding it altogether.
4246 A function-like macro that appears without an argument list. In some
4247 traditional preprocessors this was an error. In ISO C it merely means
4248 that the macro is not expanded.
4251 The unary plus operator. This did not exist in traditional C@.
4254 The @samp{U} and @samp{LL} integer constant suffixes, which were not
4255 available in traditional C@. (Traditional C does support the @samp{L}
4256 suffix for simple long integer constants.) You are not warned about
4257 uses of these suffixes in macros defined in system headers. For
4258 instance, @code{UINT_MAX} may well be defined as @code{4294967295U}, but
4259 you will not be warned if you use @code{UINT_MAX}.
4261 You can usually avoid the warning, and the related warning about
4262 constants which are so large that they are unsigned, by writing the
4263 integer constant in question in hexadecimal, with no U suffix. Take
4264 care, though, because this gives the wrong result in exotic cases.
4267 @node Implementation Details
4268 @chapter Implementation Details
4270 Here we document details of how the preprocessor's implementation
4271 affects its user-visible behavior. You should try to avoid undue
4272 reliance on behavior described here, as it is possible that it will
4273 change subtly in future implementations.
4275 Also documented here are obsolete features still supported by CPP@.
4278 * Implementation-defined behavior::
4279 * Implementation limits::
4280 * Obsolete Features::
4283 @node Implementation-defined behavior
4284 @section Implementation-defined behavior
4285 @cindex implementation-defined behavior
4287 This is how CPP behaves in all the cases which the C standard
4288 describes as @dfn{implementation-defined}. This term means that the
4289 implementation is free to do what it likes, but must document its choice
4291 @c FIXME: Check the C++ standard for more implementation-defined stuff.
4295 @item The mapping of physical source file multi-byte characters to the
4296 execution character set.
4298 The input character set can be specified using the
4299 @option{-finput-charset} option, while the execution character set may
4300 be controlled using the @option{-fexec-charset} and
4301 @option{-fwide-exec-charset} options.
4303 @item Identifier characters.
4304 @anchor{Identifier characters}
4306 The C and C++ standards allow identifiers to be composed of @samp{_}
4307 and the alphanumeric characters. C++ also allows universal character
4308 names. C99 and later C standards permit both universal character
4309 names and implementation-defined characters. In both C and C++ modes,
4310 GCC accepts in identifiers exactly those extended characters that
4311 correspond to universal character names permitted by the chosen
4314 GCC allows the @samp{$} character in identifiers as an extension for
4315 most targets. This is true regardless of the @option{std=} switch,
4316 since this extension cannot conflict with standards-conforming
4317 programs. When preprocessing assembler, however, dollars are not
4318 identifier characters by default.
4320 Currently the targets that by default do not permit @samp{$} are AVR,
4321 IP2K, MMIX, MIPS Irix 3, ARM aout, and PowerPC targets for the AIX
4324 You can override the default with @option{-fdollars-in-identifiers} or
4325 @option{-fno-dollars-in-identifiers}. @xref{fdollars-in-identifiers}.
4327 @item Non-empty sequences of whitespace characters.
4329 In textual output, each whitespace sequence is collapsed to a single
4330 space. For aesthetic reasons, the first token on each non-directive
4331 line of output is preceded with sufficient spaces that it appears in the
4332 same column as it did in the original source file.
4334 @item The numeric value of character constants in preprocessor expressions.
4336 The preprocessor and compiler interpret character constants in the
4337 same way; i.e.@: escape sequences such as @samp{\a} are given the
4338 values they would have on the target machine.
4340 The compiler evaluates a multi-character character constant a character
4341 at a time, shifting the previous value left by the number of bits per
4342 target character, and then or-ing in the bit-pattern of the new
4343 character truncated to the width of a target character. The final
4344 bit-pattern is given type @code{int}, and is therefore signed,
4345 regardless of whether single characters are signed or not.
4347 characters in the constant than would fit in the target @code{int} the
4348 compiler issues a warning, and the excess leading characters are
4351 For example, @code{'ab'} for a target with an 8-bit @code{char} would be
4352 interpreted as @w{@samp{(int) ((unsigned char) 'a' * 256 + (unsigned char)
4353 'b')}}, and @code{'\234a'} as @w{@samp{(int) ((unsigned char) '\234' *
4354 256 + (unsigned char) 'a')}}.
4356 @item Source file inclusion.
4358 For a discussion on how the preprocessor locates header files,
4359 @ref{Include Operation}.
4361 @item Interpretation of the filename resulting from a macro-expanded
4362 @samp{#include} directive.
4364 @xref{Computed Includes}.
4366 @item Treatment of a @samp{#pragma} directive that after macro-expansion
4367 results in a standard pragma.
4369 No macro expansion occurs on any @samp{#pragma} directive line, so the
4370 question does not arise.
4372 Note that GCC does not yet implement any of the standard
4377 @node Implementation limits
4378 @section Implementation limits
4379 @cindex implementation limits
4381 CPP has a small number of internal limits. This section lists the
4382 limits which the C standard requires to be no lower than some minimum,
4383 and all the others known. It is intended that there should be as few limits
4384 as possible. If you encounter an undocumented or inconvenient limit,
4385 please report that as a bug. @xref{Bugs, , Reporting Bugs, gcc, Using
4386 the GNU Compiler Collection (GCC)}.
4388 Where we say something is limited @dfn{only by available memory}, that
4389 means that internal data structures impose no intrinsic limit, and space
4390 is allocated with @code{malloc} or equivalent. The actual limit will
4391 therefore depend on many things, such as the size of other things
4392 allocated by the compiler at the same time, the amount of memory
4393 consumed by other processes on the same computer, etc.
4397 @item Nesting levels of @samp{#include} files.
4399 We impose an arbitrary limit of 200 levels, to avoid runaway recursion.
4400 The standard requires at least 15 levels.
4402 @item Nesting levels of conditional inclusion.
4404 The C standard mandates this be at least 63. CPP is limited only by
4407 @item Levels of parenthesized expressions within a full expression.
4409 The C standard requires this to be at least 63. In preprocessor
4410 conditional expressions, it is limited only by available memory.
4412 @item Significant initial characters in an identifier or macro name.
4414 The preprocessor treats all characters as significant. The C standard
4415 requires only that the first 63 be significant.
4417 @item Number of macros simultaneously defined in a single translation unit.
4419 The standard requires at least 4095 be possible. CPP is limited only
4420 by available memory.
4422 @item Number of parameters in a macro definition and arguments in a macro call.
4424 We allow @code{USHRT_MAX}, which is no smaller than 65,535. The minimum
4425 required by the standard is 127.
4427 @item Number of characters on a logical source line.
4429 The C standard requires a minimum of 4096 be permitted. CPP places
4430 no limits on this, but you may get incorrect column numbers reported in
4431 diagnostics for lines longer than 65,535 characters.
4433 @item Maximum size of a source file.
4435 The standard does not specify any lower limit on the maximum size of a
4436 source file. GNU cpp maps files into memory, so it is limited by the
4437 available address space. This is generally at least two gigabytes.
4438 Depending on the operating system, the size of physical memory may or
4439 may not be a limitation.
4443 @node Obsolete Features
4444 @section Obsolete Features
4446 CPP has some features which are present mainly for compatibility with
4447 older programs. We discourage their use in new code. In some cases,
4448 we plan to remove the feature in a future version of GCC@.
4450 @subsection Assertions
4453 @dfn{Assertions} are a deprecated alternative to macros in writing
4454 conditionals to test what sort of computer or system the compiled
4455 program will run on. Assertions are usually predefined, but you can
4456 define them with preprocessing directives or command-line options.
4458 Assertions were intended to provide a more systematic way to describe
4459 the compiler's target system and we added them for compatibility with
4460 existing compilers. In practice they are just as unpredictable as the
4461 system-specific predefined macros. In addition, they are not part of
4462 any standard, and only a few compilers support them.
4463 Therefore, the use of assertions is @strong{less} portable than the use
4464 of system-specific predefined macros. We recommend you do not use them at
4468 An assertion looks like this:
4471 #@var{predicate} (@var{answer})
4475 @var{predicate} must be a single identifier. @var{answer} can be any
4476 sequence of tokens; all characters are significant except for leading
4477 and trailing whitespace, and differences in internal whitespace
4478 sequences are ignored. (This is similar to the rules governing macro
4479 redefinition.) Thus, @code{(x + y)} is different from @code{(x+y)} but
4480 equivalent to @code{@w{( x + y )}}. Parentheses do not nest inside an
4483 @cindex testing predicates
4484 To test an assertion, you write it in an @samp{#if}. For example, this
4485 conditional succeeds if either @code{vax} or @code{ns16000} has been
4486 asserted as an answer for @code{machine}.
4489 #if #machine (vax) || #machine (ns16000)
4493 You can test whether @emph{any} answer is asserted for a predicate by
4494 omitting the answer in the conditional:
4501 Assertions are made with the @samp{#assert} directive. Its sole
4502 argument is the assertion to make, without the leading @samp{#} that
4503 identifies assertions in conditionals.
4506 #assert @var{predicate} (@var{answer})
4510 You may make several assertions with the same predicate and different
4511 answers. Subsequent assertions do not override previous ones for the
4512 same predicate. All the answers for any given predicate are
4513 simultaneously true.
4515 @cindex assertions, canceling
4517 Assertions can be canceled with the @samp{#unassert} directive. It
4518 has the same syntax as @samp{#assert}. In that form it cancels only the
4519 answer which was specified on the @samp{#unassert} line; other answers
4520 for that predicate remain true. You can cancel an entire predicate by
4521 leaving out the answer:
4524 #unassert @var{predicate}
4528 In either form, if no such assertion has been made, @samp{#unassert} has
4531 You can also make or cancel assertions using command-line options.
4537 @cindex command line
4539 Most often when you use the C preprocessor you do not have to invoke it
4540 explicitly: the C compiler does so automatically. However, the
4541 preprocessor is sometimes useful on its own. You can invoke the
4542 preprocessor either with the @command{cpp} command, or via @command{gcc -E}.
4543 In GCC, the preprocessor is actually integrated with the compiler
4544 rather than a separate program, and both of these commands invoke
4545 GCC and tell it to stop after the preprocessing phase.
4547 The @command{cpp} options listed here are also accepted by
4548 @command{gcc} and have the same meaning. Likewise the @command{cpp}
4549 command accepts all the usual @command{gcc} driver options, although those
4550 pertaining to compilation phases after preprocessing are ignored.
4552 Only options specific to preprocessing behavior are documented here.
4553 Refer to the GCC manual for full documentation of other driver options.
4556 @c man begin SYNOPSIS
4557 cpp [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
4558 [@option{-I}@var{dir}@dots{}] [@option{-iquote}@var{dir}@dots{}]
4559 [@option{-M}|@option{-MM}] [@option{-MG}] [@option{-MF} @var{filename}]
4560 [@option{-MP}] [@option{-MQ} @var{target}@dots{}]
4561 [@option{-MT} @var{target}@dots{}]
4562 @var{infile} [[@option{-o}] @var{outfile}]
4564 Only the most useful options are given above; see below for a more
4565 complete list of preprocessor-specific options.
4566 In addition, @command{cpp} accepts most @command{gcc} driver options, which
4567 are not listed here. Refer to the GCC documentation for details.
4569 @c man begin SEEALSO
4570 gpl(7), gfdl(7), fsf-funding(7),
4571 gcc(1), and the Info entries for @file{cpp} and @file{gcc}.
4575 @c man begin OPTIONS
4576 The @command{cpp} command expects two file names as arguments, @var{infile} and
4577 @var{outfile}. The preprocessor reads @var{infile} together with any
4578 other files it specifies with @samp{#include}. All the output generated
4579 by the combined input files is written in @var{outfile}.
4581 Either @var{infile} or @var{outfile} may be @option{-}, which as
4582 @var{infile} means to read from standard input and as @var{outfile}
4583 means to write to standard output. If either file is omitted, it
4584 means the same as if @option{-} had been specified for that file.
4585 You can also use the @option{-o @var{outfile}} option to specify the
4588 Unless otherwise noted, or the option ends in @samp{=}, all options
4589 which take an argument may have that argument appear either immediately
4590 after the option, or with a space between option and argument:
4591 @option{-Ifoo} and @option{-I foo} have the same effect.
4593 @cindex grouping options
4594 @cindex options, grouping
4595 Many options have multi-letter names; therefore multiple single-letter
4596 options may @emph{not} be grouped: @option{-dM} is very different from
4602 @include cppopts.texi
4603 @include cppdiropts.texi
4604 @include cppwarnopts.texi
4608 @node Environment Variables
4609 @chapter Environment Variables
4610 @cindex environment variables
4611 @c man begin ENVIRONMENT
4613 This section describes the environment variables that affect how CPP
4614 operates. You can use them to specify directories or prefixes to use
4615 when searching for include files, or to control dependency output.
4617 Note that you can also specify places to search using options such as
4618 @option{-I}, and control dependency output with options like
4619 @option{-M} (@pxref{Invocation}). These take precedence over
4620 environment variables, which in turn take precedence over the
4621 configuration of GCC@.
4623 @include cppenv.texi
4630 @node Index of Directives
4631 @unnumbered Index of Directives
4635 @unnumbered Option Index
4637 CPP's command-line options and environment variables are indexed here
4638 without any initial @samp{-} or @samp{--}.
4643 @unnumbered Concept Index