2 .\" Copyright (c) 1995-2005 The FreeBSD Project
3 .\" All rights reserved.
5 .\" Redistribution and use in source and binary forms, with or without
6 .\" modification, are permitted provided that the following conditions
8 .\" 1. Redistributions of source code must retain the above copyright
9 .\" notice, this list of conditions and the following disclaimer.
10 .\" 2. Redistributions in binary form must reproduce the above copyright
11 .\" notice, this list of conditions and the following disclaimer in the
12 .\" documentation and/or other materials provided with the distribution.
14 .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15 .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 .\" ARE DISCLAIMED. IN NO EVENT SHALL [your name] OR CONTRIBUTORS BE LIABLE
18 .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26 .\" From: @(#)style 1.14 (Berkeley) 4/28/95
34 .Nd "kernel source file style guide"
36 This file specifies the preferred style for kernel source files in the
39 It is also a guide for the preferred userland code style.
40 Many of the style rules are implicit in the examples.
41 Be careful to check the examples before assuming that
43 is silent on an issue.
46 * Style guide for FreeBSD. Based on the CSRG's KNF (Kernel Normal Form).
48 * @(#)style 1.14 (Berkeley) 4/28/95
53 * VERY important single-line comments look like this.
56 /* Most single-line comments look like this. */
59 * Multi-line comments look like this. Make them real sentences. Fill
60 * them so they look like real paragraphs.
64 The copyright header should be a multi-line comment, with the first
65 line of the comment having a dash after the star like so:
68 * Copyright (c) 1984-2025 John Q. Public
69 * All rights reserved.
71 * Long, boring license goes here, but trimmed for brevity
75 An automatic script collects license information from the tree for
76 all comments that start in the first column with
80 to not reformat a comment that starts in the first column which is not a
81 license or copyright notice, change the dash to a star for those
83 Comments starting in columns other than the first are never
84 considered license statements.
86 After any copyright header, there is a blank line, and the
88 for non C/C++ language source files.
89 Version control system ID tags should only exist once in a file
91 Non-C/C++ source files follow the example above, while C/C++ source files
93 All VCS (version control system) revision identification in files obtained
94 from elsewhere should be maintained, including, where applicable, multiple IDs
95 showing a file's history.
96 In general, do not edit foreign IDs or their infrastructure.
97 Unless otherwise wrapped (such as
98 .Dq Li "#if defined(LIBC_SCCS)" ) ,
100 .Dq Li "#if 0 ... #endif"
101 to hide any uncompilable bits
102 and to keep the IDs out of object files.
105 in front of foreign VCS IDs if the file is renamed.
109 static char sccsid[] = "@(#)style 1.14 (Berkeley) 4/28/95";
110 #endif /* not lint */
113 #include <sys/cdefs.h>
114 __FBSDID("$FreeBSD$");
117 Leave another blank line before the header files.
119 Kernel include files (i.e.\&
121 come first; normally, include
129 and it is okay to depend on that.
131 #include <sys/types.h> /* Non-local includes in angle brackets. */
134 For a network program, put the network include files next.
137 #include <net/if_dl.h>
138 #include <net/route.h>
139 #include <netinet/in.h>
140 #include <protocols/rwhod.h>
145 for files in the kernel.
147 Leave a blank line before the next group, the
150 which should be sorted alphabetically by name.
155 Global pathnames are defined in
160 in the local directory.
165 Leave another blank line before the user include files.
167 #include "pathnames.h" /* Local includes in double quotes. */
172 or declare names in the implementation namespace except
173 for implementing application interfaces.
177 macros (ones that have side effects), and the names of macros for
178 manifest constants, are all in uppercase.
179 The expansions of expression-like macros are either a single token
180 or have outer parentheses.
181 Put a single tab character between the
184 If a macro is an inline expansion of a function, the function name is
185 all in lowercase and the macro has the same name all in uppercase.
186 .\" XXX the above conflicts with ANSI style where the names are the
187 .\" same and you #undef the macro (if any) to get the function.
188 .\" It is not followed for MALLOC(), and not very common if inline
189 .\" functions are used.
191 backslashes; it makes it easier to read.
192 If the macro encapsulates a compound statement, enclose it in a
195 so that it can safely be used in
198 Any final statement-terminating semicolon should be
199 supplied by the macro invocation rather than the macro, to make parsing easier
200 for pretty-printers and editors.
202 #define MACRO(x, y) do { \e
203 variable = (x) + (y); \e
208 When code is conditionally compiled using
212 a comment may be added following the matching
216 to permit the reader to easily discern where conditionally compiled code
218 This comment should be used only for (subjectively) long regions, regions
219 greater than 20 lines, or where a series of nested
221 may be confusing to the reader.
222 Exceptions may be made for cases where code is conditionally not compiled for
225 even though the uncompiled region may be small.
226 The comment should be separated from the
231 For short conditionally compiled regions, a closing comment should not be
236 should match the expression used in the corresponding
244 should match the inverse of the expression(s) used in the preceding
249 In the comments, the subexpression
253 For the purposes of comments,
254 .Dq Ic #ifndef Li FOO
256 .Dq Ic #if Li !defined(FOO) .
259 #include <sys/ktrace.h>
263 /* A large region here, or other conditional code. */
264 #else /* !COMPAT_43 */
266 #endif /* COMPAT_43 */
269 /* Yet another large region here, or other conditional code. */
270 #else /* COMPAT_43 */
272 #endif /* !COMPAT_43 */
275 The project is slowly moving to use the
277 unsigned integer identifiers of the form
279 in preference to the older
281 integer identifiers of the form
283 New code should use the former, and old code should be converted to
284 the new form if other major work is being done in that area and
285 there is no overriding reason to prefer the older
287 Like white-space commits, care should be taken in making
291 Similarly, the project is slowly moving to use the
294 in preference to the older
300 and old code may be converted if it is
302 Literal values are named
306 These are preferred to the old spellings
310 Userspace code should include
312 while kernel code should include
315 Enumeration values are all uppercase.
317 enum enumtype { ONE, TWO } et;
320 The use of internal_underscores in identifiers is preferred over
321 camelCase or TitleCase.
323 In declarations, do not put any whitespace between asterisks and
324 adjacent tokens, except for tokens that are identifiers related to
326 (These identifiers are the names of basic types, type
328 .Ic typedef Ns -names
329 other than the one being declared.)
330 Separate these identifiers from asterisks using a single space.
332 When declaring variables in structures, declare them sorted by use, then
333 by size (largest to smallest), and then in alphabetical order.
334 The first category normally does not apply, but there are exceptions.
335 Each one gets its own line.
336 Try to make the structure
337 readable by aligning the member names using either one or two tabs
338 depending upon your judgment.
339 You should use one tab only if it suffices to align at least 90% of
341 Names following extremely long types
342 should be separated by a single space.
344 Major structures should be declared at the top of the file in which they
345 are used, or in separate header files if they are used in multiple
347 Use of the structures should be by separate declarations
350 if they are declared in a header file.
353 struct foo *next; /* List of active foo. */
354 struct mumble amumble; /* Comment for mumble. */
355 int bar; /* Try to align the comments. */
356 struct verylongtypename *baz; /* Won't fit in 2 tabs. */
358 struct foo *foohead; /* Head of global foo list. */
363 macros rather than rolling your own lists, whenever possible.
365 the previous example would be better written:
367 #include <sys/queue.h>
370 LIST_ENTRY(foo) link; /* Use queue macros for foo lists. */
371 struct mumble amumble; /* Comment for mumble. */
372 int bar; /* Try to align the comments. */
373 struct verylongtypename *baz; /* Won't fit in 2 tabs. */
375 LIST_HEAD(, foo) foohead; /* Head of global foo list. */
378 Avoid using typedefs for structure types.
379 Typedefs are problematic because they do not properly hide their
380 underlying type; for example you need to know if the typedef is
381 the structure itself or a pointer to the structure.
382 In addition they must be declared exactly once, whereas an
383 incomplete structure type can be mentioned as many times as
385 Typedefs are difficult to use in stand-alone header files:
386 the header that defines the typedef must be included
387 before the header that uses it, or by the header that uses
388 it (which causes namespace pollution), or there must be a
389 back-door mechanism for obtaining the typedef.
391 When convention requires a
393 make its name match the struct tag.
394 Avoid typedefs ending in
396 except as specified in Standard C or by
399 /* Make the structure name match the typedef. */
403 typedef int foo; /* This is foo. */
404 typedef const long baz; /* This is baz. */
407 All functions are prototyped somewhere.
409 Function prototypes for private functions (i.e., functions not used
410 elsewhere) go at the top of the first source module.
412 local to one source module should be declared
415 Functions used from other parts of the kernel are prototyped in the
416 relevant include file.
417 Function prototypes should be listed in a logical order, preferably
418 alphabetical unless there is a compelling reason to use a different
421 Functions that are used locally in more than one module go into a
422 separate header file, e.g.\&
429 In general code can be considered
431 when it makes up about 50% or more of the file(s) involved.
433 to break precedents in the existing code and use the current
437 The kernel has a name associated with parameter types, e.g., in the kernel
440 void function(int fd);
443 In header files visible to userland applications, prototypes that are
444 visible must use either
446 names (ones beginning with an underscore)
447 or no names with the types.
448 It is preferable to use protected names.
456 void function(int _fd);
459 Prototypes may have an extra space after a tab to enable function names
462 static char *function(int _arg, const char *_arg2, struct foo *_arg3,
464 static void usage(void);
467 * All major routines should have a comment briefly describing what
468 * they do. The comment before the "main" routine should describe
469 * what the program does.
472 main(int argc, char *argv[])
481 should be used to parse options.
483 should be sorted in the
493 statement that cascade should have a
496 Numerical arguments should be checked for accuracy.
497 Code which is unreachable for non-obvious reasons may be marked /*
501 while ((ch = getopt(argc, argv, "abNn:")) != -1)
502 switch (ch) { /* Indent the switch. */
503 case 'a': /* Don't indent the case. */
504 aflag = 1; /* Indent case body one tab. */
513 num = strtol(optarg, &ep, 10);
514 if (num <= 0 || *ep != '\e0') {
515 warnx("illegal number, -n argument -- %s",
529 .Pq Ic if , while , for , return , switch .
534 are allowed for single line statements.
535 Either they are used for all single statements, or
536 they are used only where needed for clarity.
537 Usage within a function should be consistent.
538 Forever loops are done with
543 for (p = buf; *p != '\e0'; ++p)
548 z = a + really + long + statement + that + needs +
549 two + lines + gets + indented + four + spaces +
550 on + the + second + and + subsequent + lines;
557 val = realloc(val, newsize);
562 loop may be left empty.
563 Do not put declarations
564 inside blocks unless the routine is unusually complicated.
566 for (; cnt < 15; cnt++) {
572 Indentation is an 8 character tab.
573 Second level indents are four spaces.
574 If you have to wrap a long statement, put the operator at the end of the
577 while (cnt < 20 && this_variable_name_is_too_long &&
579 z = a + really + long + statement + that + needs +
580 two + lines + gets + indented + four + spaces +
581 on + the + second + and + subsequent + lines;
584 Do not add whitespace at the end of a line, and only use tabs
586 to form the indentation.
587 Do not use more spaces than a tab will produce
588 and do not use spaces in front of tabs.
590 Closing and opening braces go on the same line as the
592 Braces that are not necessary may be left out.
603 No spaces after function names.
604 Commas have a space after them.
616 error = function(a1, a2);
621 Unary operators do not require spaces, binary operators do.
622 Do not use parentheses unless they are required for precedence or unless the
623 statement is confusing without them.
624 Remember that other people may
625 confuse easier than you.
626 Do YOU understand the following?
628 a = b->c[0] + ~d == (e || f) || g && h ? i : j >> 1;
632 Exits should be 0 on success, or 1 on failure.
635 * Avoid obvious comments such as
636 * "Exit 0 on success."
641 The function type should be on a line by itself
642 preceding the function.
643 The opening brace of the function body should be
647 function(int a1, int a2, float fl, int a4)
651 When declaring variables in functions declare them sorted by size,
652 then in alphabetical order; multiple ones per line are okay.
653 If a line overflows reuse the type keyword.
655 Be careful to not obfuscate the code by initializing variables in
657 Use this feature only thoughtfully.
658 DO NOT use function calls in initializers.
660 struct foo one, *two;
663 char *six, seven, eight, nine, ten, eleven, twelve;
668 Do not declare functions inside other functions; ANSI C says that
669 such declarations have file scope regardless of the nesting of the
671 Hiding file declarations in what appears to be a local
672 scope is undesirable and will elicit complaints from a good compiler.
676 are not followed by a space.
679 does not understand this rule.
681 are written with parenthesis always.
682 The redundant parenthesis rules do not apply to
687 is the preferred null pointer constant.
691 .Vt ( "type *" ) Ns 0
693 .Vt ( "type *" ) Ns Dv NULL
694 in contexts where the compiler knows the
695 type, e.g., in assignments.
697 .Vt ( "type *" ) Ns Dv NULL
699 in particular for all function args.
700 (Casting is essential for
701 variadic args and is necessary for other args if the function prototype
702 might not be in scope.)
703 Test pointers against
717 for tests unless it is a boolean, e.g.\& use:
729 should not have their return values cast
734 statements should be enclosed in parentheses.
740 do not roll your own.
742 if ((four = malloc(sizeof(struct foo))) == NULL)
743 err(1, (char *)NULL);
744 if ((six = (int *)overflow()) == NULL)
745 errx(1, "number overflowed");
750 Old-style function declarations look like this:
753 function(a1, a2, fl, a4)
754 int a1, a2; /* Declare ints, too, don't default them. */
755 float fl; /* Beware double vs. float prototype differences. */
756 int a4; /* List in order declared. */
760 Use ANSI function declarations unless you explicitly need K&R compatibility.
761 Long parameter lists are wrapped with a normal four space indent.
763 Variable numbers of arguments should look like this:
768 vaf(const char *fmt, ...)
775 /* No return needed for void functions. */
781 /* Insert an empty line if the function has no local variables. */
790 whatever; it is faster and usually cleaner, not
791 to mention avoiding stupid bugs.
793 Usage statements should look like the manual pages
795 The usage statement should be structured in the following order:
798 Options without operands come first,
799 in alphabetical order,
800 inside a single set of brackets
805 Options with operands come next,
806 also in alphabetical order,
807 with each option and its argument inside its own pair of brackets.
812 listed in the order they should be specified on the command line.
815 any optional arguments should be listed,
816 listed in the order they should be specified,
817 and all inside brackets.
825 and multiple options/arguments which are specified together are
826 placed in a single set of brackets.
827 .Bd -literal -offset 4n
828 "usage: f [-aDde] [-b b_arg] [-m m_arg] req1 req2 [opt1 [opt2]]\en"
829 "usage: f [-a | -b] [-c [-dEe] [-n number]]\en"
832 (void)fprintf(stderr, "usage: f [-ab]\en");
837 Note that the manual page options description should list the options in
838 pure alphabetical order.
839 That is, without regard to whether an option takes arguments or not.
840 The alphabetical ordering should take into account the case ordering
843 New core kernel code should be reasonably compliant with the
846 The guidelines for third-party maintained modules and device drivers are more
847 relaxed but at a minimum should be internally consistent with their style.
849 Stylistic changes (including whitespace changes) are hard on the source
850 repository and are to be avoided without good reason.
851 Code that is approximately
855 compliant in the repository must not diverge from compliance.
857 Whenever possible, code should be run through a code checker
862 and produce minimal warnings.
870 This manual page is largely based on the
871 .Pa src/admin/style/style
874 release, with occasional updates to reflect the current practice and
878 .Pa src/admin/style/style
879 is a codification by the CSRG of the programming style of Ken Thompson and