.\" Copyright (c) 1995-2001 FreeBSD Inc. .\" All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL [your name] OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" .Dd December 7, 2001 .Dt STYLE 9 .Os .Sh NAME .Nm style .Nd "kernel source file style guide" .Sh DESCRIPTION This file specifies the preferred style for kernel source files in the .Fx source tree. It is also a guide for the preferred userland code style. Many of the style rules are implicit in the examples. Be careful to check the examples before assuming that .Nm is silent on an issue. .Bd -literal /* * Style guide for FreeBSD. Based on the CSRG's KNF (Kernel Normal Form). * * @(#)style 1.14 (Berkeley) 4/28/95 * $FreeBSD$ */ /* * VERY important single-line comments look like this. */ /* Most single-line comments look like this. */ /* * Multi-line comments look like this. Make them real sentences. Fill * them so they look like real paragraphs. */ .Ed .Pp After any copyright header, there is a blank line, and the .Va rcsid for source files. Version control system ID tags should only exist once in a file (unlike in this one). Non-C/C++ source files follow the example above, while C/C++ source files follow the one below. All VCS (version control system) revision identification in files obtained from elsewhere should be maintained, including, where applicable, multiple IDs showing a file's history. In general, do not edit foreign IDs or their infrastructure. Unless otherwise wrapped (such as .Dq Li "#if defined(LIBC_SCCS)" ) , enclose both in .Dq Li "#if 0 ... #endif" to hide any uncompilable bits and to keep the IDs out of object files. Only add .Dq Li "From: " in front of foreign VCS IDs if the file is renamed. .Bd -literal #if 0 #ifndef lint static char sccsid[] = "@(#)style 1.14 (Berkeley) 4/28/95"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); .Ed .Pp Leave another blank line before the header files. .Pp Kernel include files (i.e.\& .Pa sys/*.h ) come first; normally, include .Aq Pa sys/types.h OR .Aq Pa sys/param.h , but not both. .Aq Pa sys/types.h includes .Aq Pa sys/cdefs.h , and it is okay to depend on that. .Bd -literal #include /* Non-local includes in angle brackets. */ .Ed .Pp For a network program, put the network include files next. .Bd -literal #include #include #include #include #include .Ed .Pp Do not use files in .Pa /usr/include for files in the kernel. .Pp Leave a blank line before the next group, the .Pa /usr include files, which should be sorted alphabetically by name. .Bd -literal #include .Ed .Pp Global pathnames are defined in .Aq Pa paths.h . Pathnames local to the program go in .Qq Pa pathnames.h in the local directory. .Bd -literal #include .Ed .Pp Leave another blank line before the user include files. .Bd -literal #include "pathnames.h" /* Local includes in double quotes. */ .Ed .Pp Do not .Ic #define or declare names in the implementation namespace except for implementing application interfaces. .Pp The names of .Dq unsafe macros (ones that have side effects), and the names of macros for manifest constants, are all in uppercase. The expansions of expression-like macros are either a single token or have outer parentheses. Put a single tab character between the .Ic #define and the macro name. If a macro is an inline expansion of a function, the function name is all in lowercase and the macro has the same name all in uppercase. .\" XXX the above conflicts with ANSI style where the names are the .\" same and you #undef the macro (if any) to get the function. .\" It is not followed for MALLOC(), and not very common if inline .\" functions are used. Right-justify the backslashes; it makes it easier to read. If the macro encapsulates a compound statement, enclose it in a .Ic do loop, so that it can safely be used in .Ic if statements. Any final statement-terminating semicolon should be supplied by the macro invocation rather than the macro, to make parsing easier for pretty-printers and editors. .Bd -literal #define MACRO(x, y) do { \e variable = (x) + (y); \e (y) += 2; \e } while (0) .Ed .Pp Enumeration values are all uppercase. .Bd -literal enum enumtype { ONE, TWO } et; .Ed .Pp In declarations, do not put any whitespace between asterisks and adjacent tokens, except for tokens that are identifiers related to types. (These identifiers are the names of basic types, type qualifiers, and typedef-names other than the one being declared.) Separate these identifers from asterisks using a single space. .Pp When declaring variables in structures, declare them sorted by use, then by size, and then in alphabetical order. The first category normally does not apply, but there are exceptions. Each one gets its own line. Try to make the structure readable by aligning the member names using either one or two tabs depending upon your judgment. You should use one tab only if it suffices to align at least 90% of the member names. Names following extremely long types should be separated by a single space. .Pp Major structures should be declared at the top of the file in which they are used, or in separate header files if they are used in multiple source files. Use of the structures should be by separate declarations and should be .Ic extern if they are declared in a header file. .Bd -literal struct foo { struct foo *next; /* List of active foo. */ struct mumble amumble; /* Comment for mumble. */ int bar; /* Try to align the comments. */ struct verylongtypename *baz; /* Won't fit in 2 tabs. */ }; struct foo *foohead; /* Head of global foo list. */ .Ed .Pp Use .Xr queue 3 macros rather than rolling your own lists, whenever possible. Thus, the previous example would be better written: .Bd -literal #include struct foo { LIST_ENTRY(foo) link; /* Use queue macros for foo lists. */ struct mumble amumble; /* Comment for mumble. */ int bar; /* Try to align the comments. */ struct verylongtypename *baz; /* Won't fit in 2 tabs. */ }; LIST_HEAD(, foo) foohead; /* Head of global foo list. */ .Ed .Pp Avoid using typedefs for structure types. This makes it impossible for applications to use pointers to such a structure opaquely, which is both possible and beneficial when using an ordinary struct tag. When convention requires a .Ic typedef , make its name match the struct tag. Avoid typedefs ending in .Dq Li _t , except as specified in Standard C or by \*[Px]. .Bd -literal /* Make the structure name match the typedef. */ typedef struct bar { int level; } BAR; typedef int foo; /* This is foo. */ typedef const long baz; /* This is baz. */ .Ed .Pp All functions are prototyped somewhere. .Pp Function prototypes for private functions (i.e. functions not used elsewhere) go at the top of the first source module. Functions local to one source module should be declared .Ic static . .Pp Functions used from other parts of the kernel are prototyped in the relevant include file. Function prototypes should be listed in a logical order, preferably alphabetical unless there is a compelling reason to use a different ordering. .Pp Functions that are used locally in more than one module go into a separate header file, e.g.\& .Qq Pa extern.h . .Pp Do not use the .Dv __P macro. .Pp In general code can be considered .Dq "new code" when it makes up about 50% or more of the file(s) involved. This is enough to break precedents in the existing code and use the current .Nm guidelines. .Pp The kernel has a name associated with parameter types, e.g., in the kernel use: .Bd -literal void function(int fd); .Ed .Pp In header files visible to userland applications, prototypes that are visible must use either .Dq protected names (ones beginning with an underscore) or no names with the types. It is preferable to use protected names. E.g., use: .Bd -literal void function(int); .Ed .Pp or: .Bd -literal void function(int _fd); .Ed .Pp Prototypes may have an extra space after a tab to enable function names to line up: .Bd -literal static char *function(int _arg, const char *_arg2, struct foo *_arg3, struct bar *_arg4); static void usage(void); /* * All major routines should have a comment briefly describing what * they do. The comment before the "main" routine should describe * what the program does. */ int main(int argc, char *argv[]) { char *ep; long num; int ch; .Ed .Pp For consistency, .Xr getopt 3 should be used to parse options. Options should be sorted in the .Xr getopt 3 call and the .Ic switch statement, unless parts of the .Ic switch cascade. Elements in a .Ic switch statement that cascade should have a .Li FALLTHROUGH comment. Numerical arguments should be checked for accuracy. Code that cannot be reached should have a .Li NOTREACHED comment. .Bd -literal while ((ch = getopt(argc, argv, "abn:")) != -1) switch (ch) { /* Indent the switch. */ case 'a': /* Don't indent the case. */ aflag = 1; /* FALLTHROUGH */ case 'b': bflag = 1; break; case 'n': num = strtol(optarg, &ep, 10); if (num <= 0 || *ep != '\e0') { warnx("illegal number, -n argument -- %s", optarg); usage(); } break; case '?': default: usage(); /* NOTREACHED */ } argc -= optind; argv += optind; .Ed .Pp Space after keywords .Pq Ic if , while , for , return , switch . No braces .Ql ( \&{ and .Ql \&} ) are used for control statements with zero or only a single statement unless that statement is more than a single line in which case they are permitted. Forever loops are done with .Ic for Ns 's , not .Ic while Ns 's . .Bd -literal for (p = buf; *p != '\e0'; ++p) ; /* nothing */ for (;;) stmt; for (;;) { z = a + really + long + statement + that + needs + two + lines + gets + indented + four + spaces + on + the + second + and + subsequent + lines; } for (;;) { if (cond) stmt; } if (val != NULL) val = realloc(val, newsize); .Ed .Pp Parts of a .Ic for loop may be left empty. Do not put declarations inside blocks unless the routine is unusually complicated. .Bd -literal for (; cnt < 15; cnt++) { stmt1; stmt2; } .Ed .Pp Indentation is an 8 character tab. Second level indents are four spaces. If you have to wrap a long statement, put the operator at the end of the line. .Bd -literal while (cnt < 20 && this_variable_name_is_too_long && ep != NULL) z = a + really + long + statement + that + needs + two + lines + gets + indented + four + spaces + on + the + second + and + subsequent + lines; .Ed .Pp Do not add whitespace at the end of a line, and only use tabs followed by spaces to form the indentation. Do not use more spaces than a tab will produce and do not use spaces in front of tabs. .Pp Closing and opening braces go on the same line as the .Ic else . Braces that are not necessary may be left out. .Bd -literal if (test) stmt; else if (bar) { stmt; stmt; } else stmt; .Ed .Pp No spaces after function names. Commas have a space after them. No spaces after .Ql \&( or .Ql \&[ or preceding .Ql \&] or .Ql \&) characters. .Bd -literal error = function(a1, a2); if (error != 0) exit(error); .Ed .Pp Unary operators do not require spaces, binary operators do. Do not use parentheses unless they are required for precedence or unless the statement is confusing without them. Remember that other people may confuse easier than you. Do YOU understand the following? .Bd -literal a = b->c[0] + ~d == (e || f) || g && h ? i : j >> 1; k = !(l & FLAGS); .Ed .Pp Exits should be 0 on success, or according to the predefined values in .Xr sysexits 3 . .Bd -literal exit(EX_OK); /* * Avoid obvious comments such as * "Exit 0 on success." */ } .Ed .Pp The function type should be on a line by itself preceding the function. The opening brace of the function body should be on a line by itself. .Bd -literal static char * function(int a1, int a2, float fl, int a4) { .Ed .Pp When declaring variables in functions declare them sorted by size, then in alphabetical order; multiple ones per line are okay. If a line overflows reuse the type keyword. .Pp Be careful to not obfuscate the code by initializing variables in the declarations. Use this feature only thoughtfully. DO NOT use function calls in initializers. .Bd -literal struct foo one, *two; double three; int *four, five; char *six, seven, eight, nine, ten, eleven, twelve; four = myfunction(); .Ed .Pp Do not declare functions inside other functions; ANSI C says that such declarations have file scope regardless of the nesting of the declaration. Hiding file declarations in what appears to be a local scope is undesirable and will elicit complaints from a good compiler. .Pp Casts and .Ic sizeof Ns 's are not followed by a space. Note that .Xr indent 1 does not understand this rule. .Ic sizeof Ns 's are written with parenthesis always. The redundant parenthesis rules do not apply to .Fn sizeof var instances. .Pp .Dv NULL is the preferred null pointer constant. Use .Dv NULL instead of .Vt ( "type *" ) Ns 0 or .Vt ( "type *" ) Ns Dv NULL in contexts where the compiler knows the type, e.g., in assignments. Use .Vt ( "type *" ) Ns Dv NULL in other contexts, in particular for all function args. (Casting is essential for variadic args and is necessary for other args if the function prototype might not be in scope.) Test pointers against .Dv NULL , e.g., use: .Pp .Bd -literal (p = f()) == NULL .Ed .Pp not: .Bd -literal !(p = f()) .Ed .Pp Do not use .Ic \&! for tests unless it is a boolean, e.g. use .Bd -literal if (*p == '\e0') .Ed .Pp not .Bd -literal if (!*p) .Ed .Pp Routines returning .Vt "void *" should not have their return values cast to any pointer type. .Pp Values in .Ic return statements should be enclosed in parentheses. .Pp Use .Xr err 3 or .Xr warn 3 , do not roll your own. .Bd -literal if ((four = malloc(sizeof(struct foo))) == NULL) err(1, (char *)NULL); if ((six = (int *)overflow()) == NULL) errx(1, "number overflowed"); return (eight); } .Ed .Pp Old-style function declarations look like this: .Bd -literal static char * function(a1, a2, fl, a4) int a1, a2; /* Declare ints, too, don't default them. */ float fl; /* Beware double vs. float prototype differences. */ int a4; /* List in order declared. */ { .Ed .Pp Use ANSI function declarations unless you explicitly need K&R compatibility. Long parameter lists are wrapped with a normal four space indent. .Pp Variable numbers of arguments should look like this. .Bd -literal #include void vaf(const char *fmt, ...) { va_list ap; va_start(ap, fmt); STUFF; va_end(ap); /* No return needed for void functions. */ } static void usage() { /* Insert an empty line if the function has no local variables. */ .Ed .Pp Use .Xr printf 3 , not .Xr fputs 3 , .Xr puts 3 , .Xr putchar 3 , whatever; it is faster and usually cleaner, not to mention avoiding stupid bugs. .Pp Usage statements should look like the manual pages .Sx SYNOPSIS . The usage statement should be structured in the following order: .Bl -enum .It Options without operands come first, in alphabetical order, inside a single set of brackets .Ql ( \&[ and .Ql \&] ) . .It Options with operands come next, also in alphabetical order, with each option and its argument inside its own pair of brackets. .It Required arguments (if any) are next, listed in the order they should be specified on the command line. .It Finally, any optional arguments should be listed, listed in the order they should be specified, and all inside brackets. .El .Pp A bar .Pq Ql \&| separates .Dq either-or options/arguments, and multiple options/arguments which are specified together are placed in a single set of brackets. .Bd -literal -offset 4n "usage: f [-aDde] [-b b_arg] [-m m_arg] req1 req2 [opt1 [opt2]]\en" "usage: f [-a | -b] [-c [-dEe] [-n number]]\en" .Ed .Bd -literal (void)fprintf(stderr, "usage: f [-ab]\en"); exit(EX_USAGE); } .Ed .Pp Note that the manual page options description should list the options in pure alphabetical order. That is, without regard to whether an option takes arguments or not. The alphabetical ordering should take into account the case ordering shown above. .Pp New core kernel code should be reasonably compliant with the .Nm guides. The guidelines for third-party maintained modules and device drivers are more relaxed but at a minimum should be internally consistent with their style. .Pp Stylistic changes (including whitespace changes) are hard on the source repository and are to be avoided without good reason. Code that is approximately .Fx KNF .Nm compliant in the repository must not diverge from compliance. .Pp Whenever possible, code should be run through a code checker (e.g., .Xr lint 1 or .Nm gcc Fl Wall ) and produce minimal warnings. .Sh SEE ALSO .Xr indent 1 , .Xr lint 1 , .Xr err 3 , .Xr sysexits 3 , .Xr warn 3 .Sh HISTORY This man page is largely based on the .Pa src/admin/style/style file from the .Bx 4.4 Lite2 release, with occasional updates to reflect the current practice and desire of the .Fx project.