FreeBSD Manual Pages
STYLE(9) Kernel Developer's Manual STYLE(9) NAME style -- kernel source file style guide DESCRIPTION This file specifies the preferred style for kernel source files in the FreeBSD source tree. It is also a guide for the preferred userland code style. The preferred line width is 80 characters, but some exceptions are made when a slightly longer line is clearer or easier to read. Anything that is frequently grepped for, such as diagnostic, error, or panic mes- sages, should not be broken up over multiple lines despite this rule. Many of the style rules are implicit in the examples. Be careful to check the examples before assuming that style is silent on an issue. /* * Style guide for FreeBSD. Based on the CSRG's KNF (Kernel Normal Form). */ /* * VERY important single-line comments look like this. */ /* Most single-line comments look like this. */ // Although they may look like this. /* * Multi-line comments look like this. Make them real sentences. Fill * them so they look like real paragraphs. */ C++ comments may be used in C and C++ code. Single-line comments should be consistently either C or C++ within a file. Multi-line comments should also be consistently either C or C++, but may differ from single-line com- ments. The copyright header should be a multi-line comment like so: /* * Copyright (c) 1984-2025 John Q. Public * * SPDX-License-Identifier: BSD-2-Clause */ Comments starting in columns other than the first are never considered li- cense statements. Write the copyright lines before the appropriate SPDX- License-Identifier. If the copyright assertion contains the phrase "All Rights Reserved" that should be on the same line as the word "Copyright". You should not insert a new copyright line between an old copyright line and this phrase. Instead, you should insert a new copyright phrase after a pre-existing "All Rights Reserved" line. When making changes, it is ac- ceptable to fold an "All Rights Reserved" line with each of the "Copyright" lines. For files that have the "All Rights Reserved" line on the same line(s) as the word "Copyright", new copyright assertions should be added last. New "Copyright" lines should only be added when making substantial changes to the file, not for trivial changes. After any copyright and license comment, there is a blank line. Non-C/C++ source files follow the example above, while C/C++ source files follow the one below. Version control system ID tags should only exist once in a file (unlike in this one). All VCS (version control system) revision identifi- cation 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 "#if defined(LIBC_SCCS)"), enclose both in "#if 0 ... #endif" to hide any uncompilable bits and to keep the IDs out of object files. Only add "From: " in front of foreign VCS IDs if the file is renamed. Add "From: " and the FreeBSD git hash with full path name if the file was de- rived from another FreeBSD file and include relevant copyright info from the original file. Leave one blank line before the header files. Kernel include files (sys/*.h) come first. If either <sys/types.h> or <sys/param.h> is needed, include it before other include files. (<sys/param.h> includes <sys/types.h>; do not include both.) Next, include <sys/systm.h>, if needed. The remaining kernel headers should be sorted alphabetically. #include <sys/types.h> /* Non-local includes in angle brackets. */ #include <sys/systm.h> #include <sys/endian.h> #include <sys/lock.h> #include <sys/queue.h> For a network program, put the network include files next. #include <net/if.h> #include <net/if_dl.h> #include <net/route.h> #include <netinet/in.h> #include <protocols/rwhod.h> Do not include files from /usr/include in the kernel. Leave a blank line before the next group, the /usr/include files, which should be sorted alphabetically by name. #include <stdio.h> Global pathnames are defined in <paths.h>. Pathnames local to the program go in "pathnames.h" in the local directory. #include <paths.h> Leave another blank line before the local include files. #include "pathnames.h" /* Local includes in double quotes. */ Do not #define or declare names in the implementation namespace except for implementing application interfaces. The names of "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 space or tab character between the #define and the macro name, but be consistent within a file. 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. Right-justify the backslashes; it makes it easier to read. If the macro encapsulates a compound statement, enclose it in a do loop, so that it can safely be used in if statements. Any final state- ment-terminating semicolon should be supplied by the macro invocation rather than the macro, to make parsing easier for pretty-printers and edi- tors. #define MACRO(x, y) do { \ variable = (x) + (y); \ (y) += 2; \ } while (0) When code is conditionally compiled using #ifdef or #if, a comment may be added following the matching #endif or #else to permit the reader to easily discern where conditionally compiled code regions end. This comment should be used only for (subjectively) long regions, regions greater than 20 lines, or where a series of nested #ifdef 's may be confusing to the reader. The comment should be separated from the #endif or #else by a sin- gle space. For short conditionally compiled regions, a closing comment should not be used. The comment for #endif should match the expression used in the correspond- ing #if or #ifdef. The comment for #else and #elif should match the in- verse of the expression(s) used in the preceding #if and/or #elif state- ments. In the comments, the subexpression "defined(FOO)" is abbreviated as "FOO". For the purposes of comments, "#ifndef FOO" is treated as "#if !defined(FOO)". #ifdef KTRACE #include <sys/ktrace.h> #endif #ifdef COMPAT_43 /* A large region here, or other conditional code. */ #else /* !COMPAT_43 */ /* Or here. */ #endif /* COMPAT_43 */ #ifndef COMPAT_43 /* Yet another large region here, or other conditional code. */ #else /* COMPAT_43 */ /* Or here. */ #endif /* !COMPAT_43 */ The project prefers the use of ISO/IEC 9899:1999 ("ISO C99") unsigned inte- ger identifiers of the form uintXX_t rather than the older BSD-style inte- ger identifiers of the form u_intXX_t. New code should use the former, and old code should be converted to the new form if other major work is being done in that area and there is no overriding reason to prefer the older BSD-style. Like white-space commits, care should be taken in making uintXX_t only commits. Similarly, the project prefers the use of ISO C99 bool rather than the older int or boolean_t. New code should use bool, and old code may be con- verted if it is reasonable to do so. Literal values are named true and false. These are preferred to the old spellings TRUE and FALSE. Userspace code should include <stdbool.h>, while kernel code should include <sys/types.h>. Likewise, the project prefers ISO C99 designated initializers when it makes sense to do so. Enumeration values are all uppercase. enum enumtype { ONE, TWO } et; The use of internal_underscores in identifiers is preferred over camelCase or TitleCase. 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 identi- fiers from asterisks using a single space. When declaring variables in structures, declare them sorted by use, then by size (largest to smallest), and then in alphabetical order. The first cat- egory 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. 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 extern if they are declared in a header file. 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; /* Does not fit in 2 tabs. */ }; struct foo *foohead; /* Head of global foo list. */ Use queue(3) macros rather than rolling your own lists, whenever possible. Thus, the previous example would be better written: #include <sys/queue.h> 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; /* Does not fit in 2 tabs. */ }; LIST_HEAD(, foo) foohead; /* Head of global foo list. */ Avoid using typedefs for structure types. Typedefs are problematic because they do not properly hide their underlying type; for example you need to know if the typedef is the structure itself or a pointer to the structure. In addition they must be declared exactly once, whereas an incomplete structure type can be mentioned as many times as necessary. Typedefs are difficult to use in stand-alone header files: the header that defines the typedef must be included before the header that uses it, or by the header that uses it (which causes namespace pollution), or there must be a back- door mechanism for obtaining the typedef. When convention requires a typedef, make its name match the struct tag. Avoid typedefs ending in "_t", except as specified in Standard C or by POSIX. /* 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. */ All functions are prototyped somewhere. Function prototypes for private functions (i.e., functions not used else- where) go at the top of the first source module. Functions local to one source module should be declared static. Functions used from other parts of the kernel are prototyped in the rele- vant include file. Function prototypes should be listed in a logical or- der, preferably alphabetical unless there is a compelling reason to use a different ordering. Functions that are used locally in more than one module go into a separate header file, e.g., "extern.h". The kernel has a name associated with parameter types, e.g., in the kernel use: void function(int fd); In header files visible to userland applications, prototypes that are visi- ble must use either "protected" names (ones beginning with an underscore) or no names with the types. It is preferable to use protected names. E.g., use: void function(int); or: void function(int _fd); Prototypes may have an extra space after a tab to enable function names to line up: 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; For consistency, getopt(3) should be used to parse options. Options should be sorted in the getopt(3) call and the switch statement, unless parts of the switch cascade. Elements in a switch statement that execute some code and then cascade to the next case should have a FALLTHROUGH comment. Nu- merical arguments should be checked for accuracy. Code which is unreach- able for non-obvious reasons may be marked /* NOTREACHED */. while ((ch = getopt(argc, argv, "abNn:")) != -1) switch (ch) { /* Indent the switch. */ case 'a': /* Do not indent the case. */ aflag = 1; /* Indent case body one tab. */ /* FALLTHROUGH */ case 'b': bflag = 1; break; case 'N': Nflag = 1; break; case 'n': num = strtol(optarg, &ep, 10); if (num <= 0 || *ep != '\0') { warnx("illegal number, -n argument -- %s", optarg); usage(); } break; case '?': default: usage(); } argc -= optind; argv += optind; Space after keywords (if, while, for, return, switch). Two styles of braces (`{' and `}') are allowed for single line statements. Either they are used for all single statements, or they are used only where needed for clarity. Usage within a function should be consistent. Forever loops are done with for's, not while's. for (p = buf; *p != '\0'; ++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); Parts of a for loop may be left empty. for (; cnt < 15; cnt++) { stmt1; stmt2; } A for loop may declare and initialize its counting variable. for (int i = 0; i < 15; i++) { stmt1; } 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. 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; 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. Closing and opening braces go on the same line as the else. Braces that are not necessary may be left out. if (test) stmt; else if (bar) { stmt; stmt; } else stmt; No spaces after function names. Commas have a space after them. No spaces after `(' or `[' or preceding `]' or `)' characters. error = function(a1, a2); if (error != 0) exit(error); 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? a = b->c[0] + ~d == (e || f) || g && h ? i : j >> 1; k = !(l & FLAGS); Exits should be 0 on success, or 1 on failure. exit(0); /* * Avoid obvious comments such as * "Exit 0 on success." */ } 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. static char * function(int a1, int a2, float fl, int a4, struct bar *bar) { 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. Variables may be initialized where declared espe- cially when they are constant for the rest of the scope. Declarations may be in any block, but must be placed before statements. Calls to compli- cated functions should be avoided when initializing variables. struct foo one, *two; struct baz *three = bar_get_baz(bar); double four; int *five, six; char *seven, eight, nine, ten, eleven, twelve; four = my_complicated_function(a1, f1, a4); Do not declare functions inside other functions; ANSI C says that such dec- larations 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. Casts and sizeof's are not followed by a space. sizeof's are written with parenthesis always. The redundant parenthesis rules do not apply to sizeof(var) instances. NULL is the preferred null pointer constant. Use NULL instead of (type *)0 or (type *)NULL in contexts where the compiler knows the type, e.g., in as- signments. Use (type *)NULL in other contexts, in particular for all func- tion 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 NULL, e.g., use: (p = f()) == NULL not: !(p = f()) Do not test without a comparison, or with unary ! (except for booleans). For example, use: if (*p == '\0') not: if (!*p) Prefer: if (count != 0) over: if (count) Routines returning void * should not have their return values cast to any pointer type. Values in return statements should be enclosed in parentheses. Use err(3) or warn(3), do not roll your own. if ((four = malloc(sizeof(struct foo))) == NULL) err(1, (char *)NULL); if ((six = (int *)overflow()) == NULL) errx(1, "number overflowed"); return (eight); } Do not use K&R style declarations or definitions, they are obsolete and are forbidden in C23. Compilers warn of their use and some treat them as an error by default. When converting K&R style definitions to ANSI style, preserve any comments about parameters. Long parameter lists are wrapped with a normal four space indent. Variable numbers of arguments should look like this: #include <stdarg.h> 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) { Functions should have local variable declarations first, followed by one blank line, followed by the first statement. If no local variables are de- clared, the first line should be a statement. Older versions of this style document required a blank line before code. Such lines should be removed when significant changes are made to the code. Use printf(3), not fputs(3), puts(3), putchar(3), whatever; it is faster and usually cleaner, not to mention avoiding stupid bugs. Usage statements should look like the manual pages "SYNOPSIS". The usage statement should be structured in the following order: 1. Options without operands come first, in alphabetical order, inside a single set of brackets (`[' and `]'). 2. Options with operands come next, also in alphabetical order, with each option and its argument inside its own pair of brackets. 3. Required arguments (if any) are next, listed in the order they should be specified on the command line. 4. Finally, any optional arguments should be listed, listed in the order they should be specified, and all inside brackets. A bar (`|') separates "either-or" options/arguments, and multiple op- tions/arguments which are specified together are placed in a single set of brackets. "usage: f [-aDde] [-b b_arg] [-m m_arg] req1 req2 [opt1 [opt2]]\n" "usage: f [-a | -b] [-c [-dEe] [-n number]]\n" (void)fprintf(stderr, "usage: f [-ab]\n"); exit(1); } 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. Whenever possible, code should be run through a code checker (e.g., various static analyzers or cc -Wall) and produce minimal warnings. New code should use _Static_assert() instead of the older CTASSERT(). __predict_true() and __predict_false() should only be used in frequently executed code when it makes the code measurably faster. It is wasteful to make predictions for infrequently run code, like subsystem initialization. When using branch prediction hints, atypical error conditions should use __predict_false() (document the exceptions). Operations that almost always succeed use __predict_true(). Only use the annotation for the entire if statement, rather than individual clauses. Do not add these annotations without empirical evidence of the likelihood of the branch. New core kernel code should be compliant with the style guides. The guide- lines for third-party maintained modules and device drivers are more re- laxed. Their code is expected to at least be internally consistent with their style. Stylistic changes, including whitespace ones, complicate the work of down- stream consumers and may impair developers' ability to trace the history of some changes. Such standalone must be avoided, and should not span unre- lated directories as this increases the chances of conflicts when merging to stable and release branches (MFCs). On the other hand, when a signifi- cant portion, usually about a half, of some logical unit of code, be it a function, group of functions, file or group of files, is going to be modi- fied, developers are encouraged to amend the style of the whole unit as de- scribed in this document. In this case, style changes to otherwise unmodi- fied code should be committed separately. Style-only commits should be added to the file .git-blame-ignore-revs at the top of the source reposi- tory to hide them from `git blame'. Comments in this file indicate how to use it. Code that is approximately FreeBSD KNF style compliant in the repository must not diverge from compliance. C++ KNF style was originally defined as a style for C. C++ introduces several new idioms which do not have an existing corollary in KNF C such as inline function definitions in classes. C++ is also not always compatible with some KNF guidelines such as enclosing return values in parentheses. For C++ code, FreeBSD aims to follow broadly accepted C++ practices while also following the general shape of KNF. This section enumerates C++ specific guidelines that differ from KNF C. The preferred suffixes for C++ source files are ".cc" and ".hh". Header files should always use a suffix, unlike headers from the C++ standard li- brary. Return values should not be enclosed in parentheses. When converting ex- isting C code to C++, existing return values may remain in parentheses. The opening curly brace for namespace declarations should be on the first line similar to structure and class definitions. Nested namespaces should be declared using a single namespace declaration. namespace foo::bar { } Member function declarations should follow the same style used for stand- alone function prototypes except that a space should be used between a function's return type and name. Function definitions at the top level should use a newline after the func- tion type similar to C function definitions. Nested member function definitions inside of a class, structure, or union should not use a newline after the function type. Instead, these should follow the style of member function declarations. This is more common C++ style and is more compact for small methods such as getters and setters. Inline functions whose body consists of a single statement may use a single line for the function body. Inline functions with an empty body should al- ways use a single line. struct widget { int foo() { return 4; } int bar(); }; int widget::bar() { return 6; } Default and deleted methods should be declared as a single line. class box { ~box() = default; }; In template declarations, the template keyword and list of template parame- ters should be followed by a newline before the templated declaration. template <typename T> class box { T data; }; The & for reference variables should be placed on the variable name rather than the type similar to the style used with * for pointers. int x; int &xp = x; Variables may be declared at any point within a function, not just at the start of blocks. Standard library containers should be used in preference to queue(3) or tree(3) macros. nullptr should be used instead of NULL or 0. Use standard library types for managing strings such as std::string and std::string_view rather than char * and const char *. C types may be used when interfacing with C code. The auto keyword can be used in various contexts which improve readability. Examples include iterators, non-trivial types of ranged-for values, and re- turn values of obvious types, such as static_cast or std::make_unique(). Place any qualifiers before auto, for example: const auto. Use the std::unique_ptr and std::shared_ptr smart pointers to manage the lifetime of dynamically allocated objects instead of new and delete. Con- struct smart pointers with std::make_unique() or std::make_shared(). Do not use malloc(3) except when necessary to interface with C code. Do not import any namespaces with using at global scope in header files. Namespaces other than the std namespace (for example, std::literals) may be imported in source files and in function scope in header files. Define type aliases using using instead of typedef. FILES /usr/src/tools/build/checkstyle9.pl A script to check for violations of style in a source file. /usr/src/tools/tools/editing/freebsd.el An Emacs plugin to follow the FreeBSD style indentation rules. /usr/src/tools/tools/editing/freebsd.vim A Vim plugin to follow the FreeBSD style indentation rules. SEE ALSO indent(1), err(3), warn(3), style.Makefile(5), style.mdoc(5), style.lua(9) HISTORY This manual page was originally based on the src/admin/style/style file from the 4.4BSD-Lite2 release, with frequent updates to reflect the current practice and desire of the FreeBSD project. src/admin/style/style is a codification by the CSRG of the programming style of Ken Thompson and Den- nis Ritchie in Version 6 AT&T UNIX. FreeBSD ports 15.quarterly July 28, 2025 STYLE(9)
NAME | DESCRIPTION | FILES | SEE ALSO | HISTORY
Want to link to this manual page? Use this URL:
<https://man.freebsd.org/cgi/man.cgi?query=style&sektion=9&manpath=FreeBSD+15.1-RELEASE+and+Ports.quarterly>