.\" Copyright (c) 1995 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. .\" .\" $OpenBSD: style.9,v 1.35 2003/10/04 11:35:31 jmc Exp $ .\" .Dd June 18, 2001 .Dt STYLE 9 .Os .Sh NAME .Nm style .Nd "Kernel source file style guide (KNF)" .Sh DESCRIPTION This file specifies the preferred style for kernel source files in the .Ox source tree. It is also a guide for preferred user land code style. These guidelines should be followed for all new code. 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 style guidelines. .Bd -literal -offset 0i /* * Style guide for the OpenBSD KNF (Kernel Normal Form). */ /* * 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 Kernel include files (i.e., .Aq Pa sys/*.h ) come first; normally, you'll need .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's okay to depend on that. .Bd -literal -offset 0i #include /* Non-local includes in brackets. */ .Ed .Pp If it's a network program, put the network include files next. .Bd -literal -offset 0i #include #include #include #include #include .Ed .Pp Then there's a blank line, followed by the .Pa /usr/include files. The .Pa /usr/include files should be sorted! .Bd -literal -offset 0i #include .Ed .Pp Global pathnames are defined in .Pa /usr/include/paths.h . Pathnames local to the program go in .Pa pathnames.h in the local directory. .Bd -literal -offset 0i #include .Ed .Pp Then, there's a blank line, and the user include files. .Bd -literal -offset 0i #include "pathnames.h" /* Local includes in double quotes. */ .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. In user land, functions local to one source module should be declared .Ql static . This should not be done in kernel land since it makes it impossible to use the kernel debugger. .Pp Functions used from other parts of the kernel are prototyped in the relevant include file. .Pp Functions that are used locally in more than one module go into a separate header file, e.g., .Pa extern.h . .Pp Use of the .Li __P macro has been deprecated. It is allowed in code imported from other sources but should not be used in native .Ox code. Prototypes should not have variable names associated with the types; i.e.: .Bd -literal -compact -offset 0i void function(int); .Ed .Pp not: .Bd -literal -compact -offset 0i void function(int a); .Ed .Pp Prototypes may have an extra space after a tab to enable function names to line up: .Bd -literal -offset 0i static char *function(int, const char *); static void usage(void); .Ed .Pp There should be no space between the function name and the argument list. .Pp Use .Li __dead from .Aq Pa sys/cdefs.h for functions that don't return, i.e., .Bd -literal -offset 0i __dead void abort(void); .Ed .Pp In header files, put function prototypes within .Dv __BEGIN_DECLS / __END_DECLS matching pairs. This makes the header file usable from C++. .Pp Macros are capitalized and parenthesized, and should avoid side-effects. If they are an inline expansion of a function, the function is defined all in lowercase; the macro has the same name all in uppercase. If the macro needs more than a single line, use braces. Right-justify the backslashes, as the resulting definition is easier to read. If the macro encapsulates a compound statement, enclose it in a .Dq Li do loop, so that it can safely be used in .Dq Li 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 -offset 0i #define MACRO(x, y) do { \e variable = (x) + (y); \e (y) += 2; \e } while (0) .Ed .Pp Enumeration values are all uppercase. .Bd -literal -offset 0i enum enumtype { ONE, TWO } et; .Ed .Pp When declaring variables in structures, declare them sorted by use, then by size (largest to smallest), then by alphabetical order. The first category normally doesn't apply, but there are exceptions. Each one gets its own line. Put a tab after the first word, i.e., use .Ql int^Ix; and .Ql struct^Ifoo *x; . .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 .Dq Li extern if they are declared in a header file. .Bd -literal -offset 0i struct foo { struct foo *next; /* List of active foo */ struct mumble amumble; /* Comment for mumble */ int bar; }; 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 -offset 0i #include struct foo { LIST_ENTRY(foo) link; /* Queue macro glue for foo lists */ struct mumble amumble; /* Comment for mumble */ int bar; }; 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 typedef, make its name match the struct tag. Avoid typedefs ending in .Dq Li \&_t , except as specified in Standard C or by .Tn POSIX . Don't use the same name for a struct tag and a typedef, as this makes the code unusable from C++. .Bd -literal -offset 0i /* Make the structure name match the typedef. */ typedef struct _bar { int level; } BAR; .Ed .Pp .Bd -literal -offset 0i /* * 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[]) { long num; int ch; char *ep; .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 switch statement, unless parts of the switch cascade. Elements in a switch statement that cascade should have a FALLTHROUGH comment. Numerical arguments should be checked for accuracy. Code that cannot be reached should have a NOTREACHED comment. .Bd -literal -offset 0i 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 Use space after keywords .Pf ( Li if , .Li while , .Li for , .Li return , .Li switch ) . No braces 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 .Li for , not .Li while . .Bd -literal -offset 0i 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; } .Ed .Pp Parts of a for loop may be left empty. Don't put declarations inside blocks unless the routine is unusually complicated. .Bd -literal -offset 0i for (; cnt < 15; cnt++) { stmt1; stmt2; } .Ed .Pp Indentation is an 8 character tab. Second level indents are four spaces. .Bd -literal -offset 0i while (cnt < 20) 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 else. Braces that aren't necessary may be left out, unless they cause a compiler warning. .Bd -literal -offset 0i if (test) stmt; else if (bar) { stmt; stmt; } else stmt; .Ed .Pp Do not use spaces after function names. Commas have a space after them. Do not use spaces after .Sq \&( or .Sq \&[ or preceding .Sq \&] or .Sq \&) characters. .Bd -literal -offset 0i if ((error = function(a1, a2))) exit(error); .Ed .Pp Unary operators don't require spaces; binary operators do. Don't use parentheses unless they're required for precedence, the statement is confusing without them, or the compiler generates a warning without them. Remember that other people may be confused more easily than you. Do YOU understand the following? .Bd -literal -offset 0i a = b->c[0] + ~d == (e || f) || g && h ? i : j >> 1; k = !(l & FLAGS); .Ed .Pp Exits should be 0 on success, or non-zero for errors. .Bd -literal -offset 0i exit(0); /* * Avoid obvious comments such as * "Exit 0 on success." */ } .Ed .Pp The function type should be on a line by itself preceding the function. .Bd -literal -offset 0i static char * function(int a1, int a2, float fl, int a4) { .Ed .Pp When declaring variables in functions declare them sorted by size (largest to smallest), then in alphabetical order; multiple ones per line are okay. Old style function declarations should be avoided. ANSI style function declarations should go in an include file such as .Dq Pa extern.h . If a line overflows reuse the type keyword. .Pp Be careful not to obfuscate the code by initializing variables in the declarations. Use this feature only thoughtfully. DO NOT use function calls in initializers! .Bd -literal -offset 0i 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, such as .Dq Li gcc -Wtraditional . .Pp Casts and .Fn sizeof calls are not followed by a space. Note that .Xr indent 1 does not understand this rule. .Pp Use of the .Dq register specifier is discouraged in new code. Optimizing compilers such as gcc can generally do a better job of choosing which variables to place in registers to improve code performance. The exception to this is in functions containing assembly code where the .Dq register specifier is required for proper code generation in the absence of compiler optimization. .Pp When using .Fn longjmp or .Fn vfork in a program, the .Fl W or .Fl Wall flag should be used to verify that the compiler does not generate warnings such as .Bd -literal -offset 0i warning: variable `foo' might be clobbered by `longjmp' or `vfork'. .Ed .Pp If any warnings of this type occur you must apply the .Dq volatile specifier to the variable in question. Failure to do so may result in improper code generation when optimization is enabled. Note that for pointers, the location of .Dq volatile specifies if the type-qualifier applies to the pointer, or the thing being pointed to. A volatile pointer is declared with .Dq volatile to the right of the .Dq * . Example: .Bd -literal -offset 0i char *volatile foo; .Ed .Pp says that .Dq foo is volatile, but .Dq *foo is not. To make .Dq *foo volatile use the syntax .Bd -literal -offset 0i volatile char *foo; .Ed .Pp If both the pointer and the thing pointed to are volatile use .Bd -literal -offset 0i volatile char *volatile foo; .Ed .Pp .Dq const is also a type-qualifier and the same rules apply. The description of a read-only hardware register might look something like: .Bd -literal -offset 0i const volatile char *reg; .Ed .Pp Global flags set inside signal handlers should be of type .Dq volatile sig_atomic_t if possible. This guarantees that the variable may be accessed as an atomic entity, even when a signal has been delivered. Global variables of other types (such as structures) are not guaranteed to have consistent values when accessed via a signal handler. .Pp .Dv NULL is the preferred null pointer constant. Use .Dv NULL instead of (type\ *)0 or (type\ *)NULL in contexts where the compiler knows the type, e.g., in assignments. Use (type\ *)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 , i.e., use: .Bd -literal -offset 0i (p = f()) == NULL .Ed .Pp not: .Bd -literal -offset 0i !(p = f()) .Ed .Pp Don't use .Ql \&! for tests unless it's a boolean, i.e., use .Bd -literal -offset 0i if (*p == '\e0') .Ed .Pp not .Bd -literal -offset 0i if (!*p) .Ed .Pp Routines returning .Li void * should not have their return values cast to any pointer type. .Pp Use .Xr err 3 or .Xr warn 3 , don't roll your own! .Bd -literal -offset 0i 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 -offset 0i 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 -offset 0i #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(void) { extern char *__progname; /* from crt0.o */ .Ed .Pp Usage statements should look like the manual pages. Options without operands come first, in alphabetical order inside a single set of braces, followed by options with operands, in alphabetical order, each in braces, followed by required arguments in the order they are specified, followed by optional arguments in the order they are specified. .Pp A bar .Pq Sq \&| separates either-or options/arguments, and multiple options/arguments which are specified together are placed in a single set of braces. .Pp If numbers are used as options, they should be placed first, as shown in the example below. Uppercase letters take precedence over lowercase. Note that the options list in manual pages should be purely alphabetical. That is, with no regard to whether an option takes an argument. .Bd -ragged -offset 0.3i "usage: f [-12aDde] [-b b_arg] [-m m_arg] req1 req2 [opt1 [opt2]]\en" "usage: f [-a | -b] [-c [-de] [-n number]]\en" .Ed .Pp The .Li __progname string, should be used instead of hard-coding the program name. .Bd -literal -offset 0i (void)fprintf(stderr, "usage: %s [-ab]\en", __progname); exit(1); } .Ed .Pp New core kernel code should be reasonably compliant with the style 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 KNF compliant in the repository must not diverge from compliance. .Pp Whenever possible, code should be run through a code checker (e.g., .Dq Li gcc -Wall -W -Wtraditional -Wpointer-arith -Wbad-function-cast ... , .Xr lint 1 or lclint from the ports tree) and produce minimal warnings. .Sh SEE ALSO .Xr indent 1 , .Xr err 3 , .Xr queue 3 , .Xr warn 3 .Sh HISTORY This man page is largely based on the src/admin/style/style file from the .Tn BSD 4.4-Lite2 release, with updates to reflect the current practice and desire of the .Ox project.