1 Open vSwitch Coding Style
2 =========================
4 This file describes the coding style used in most C files in the Open
5 vSwitch distribution. However, Linux kernel code datapath directory
6 follows the Linux kernel's established coding conventions. For the
7 Windows kernel datapath code, use the coding style described in
8 datapath-windows/CodingStyle.
10 The following GNU indent options approximate this style:
12 -npro -bad -bap -bbb -br -blf -brs -cdw -ce -fca -cli0 -npcs -i4 -l79 \
13 -lc79 -nbfda -nut -saf -sai -saw -sbi4 -sc -sob -st -ncdb -pi4 -cs -bs \
19 Limit lines to 79 characters.
21 Use form feeds (control+L) to divide long source files into logical
22 pieces. A form feed should appear as the only character on a line.
24 Do not use tabs for indentation.
26 Avoid trailing spaces on lines.
31 Use names that explain the purpose of a function or object.
33 Use underscores to separate words in an identifier: multi_word_name.
35 Use lowercase for most names. Use uppercase for macros, macro
36 parameters, and members of enumerations.
38 Give arrays names that are plural.
40 Pick a unique name prefix (ending with an underscore) for each
41 module, and apply that prefix to all of that module's externally
42 visible names. Names of macro parameters, struct and union members,
43 and parameters in function prototypes are not considered externally
44 visible for this purpose.
46 Do not use names that begin with _. If you need a name for
47 "internal use only", use __ as a suffix instead of a prefix.
49 Avoid negative names: "found" is a better name than "not_found".
51 In names, a "size" is a count of bytes, a "length" is a count of
52 characters. A buffer has size, but a string has length. The length
53 of a string does not include the null terminator, but the size of the
54 buffer that contains the string does.
59 Comments should be written as full sentences that start with a
60 capital letter and end with a period. Put two spaces between
63 Write block comments as shown below. You may put the /* and */ on
64 the same line as comment text if you prefer.
67 * We redirect stderr to /dev/null because we often want to remove all
68 * traffic control configuration on a port so its in a known state. If
69 * this done when there is no such configuration, tc complains, so we just
73 Each function and each variable declared outside a function, and
74 each struct, union, and typedef declaration should be preceded by a
75 comment. See FUNCTION DEFINITIONS below for function comment
78 Each struct and union member should each have an inline comment that
79 explains its meaning. structs and unions with many members should be
80 additionally divided into logical groups of members by block comments,
83 /* An event that will wake the following call to poll_block(). */
85 /* Set when the waiter is created. */
86 struct list node; /* Element in global waiters list. */
87 int fd; /* File descriptor. */
88 short int events; /* Events to wait for (POLLIN, POLLOUT). */
89 poll_fd_func *function; /* Callback function, if any, or null. */
90 void *aux; /* Argument to callback function. */
91 struct backtrace *backtrace; /* Event that created waiter, or null. */
93 /* Set only when poll_block() is called. */
94 struct pollfd *pollfd; /* Pointer to element of the pollfds array
95 (null if added from a callback). */
98 Use XXX or FIXME comments to mark code that needs work.
100 Don't use // comments.
102 Don't comment out or #if 0 out code. Just remove it. The code that
103 was there will still be in version control history.
108 Put the return type, function name, and the braces that surround the
109 function's code on separate lines, all starting in column 0.
111 Before each function definition, write a comment that describes the
112 function's purpose, including each parameter, the return value, and
113 side effects. References to argument names should be given in
114 single-quotes, e.g. 'arg'. The comment should not include the
115 function name, nor need it follow any formal structure. The comment
116 does not need to describe how a function does its work, unless this
117 information is needed to use the function correctly (this is often
118 better done with comments *inside* the function).
120 Simple static functions do not need a comment.
122 Within a file, non-static functions should come first, in the order
123 that they are declared in the header file, followed by static
124 functions. Static functions should be in one or more separate pages
125 (separated by form feed characters) in logical groups. A commonly
126 useful way to divide groups is by "level", with high-level functions
127 first, followed by groups of progressively lower-level functions.
128 This makes it easy for the program's reader to see the top-down
129 structure by reading from top to bottom.
131 All function declarations and definitions should include a
132 prototype. Empty parentheses, e.g. "int foo();", do not include a
133 prototype (they state that the function's parameters are unknown);
134 write "void" in parentheses instead, e.g. "int foo(void);".
136 Prototypes for static functions should either all go at the top of
137 the file, separated into groups by blank lines, or they should appear
138 at the top of each page of functions. Don't comment individual
139 prototypes, but a comment on each group of prototypes is often
142 In the absence of good reasons for another order, the following
143 parameter order is preferred. One notable exception is that data
144 parameters and their corresponding size parameters should be paired.
146 1. The primary object being manipulated, if any (equivalent to the
147 "this" pointer in C++).
148 2. Input-only parameters.
149 3. Input/output parameters.
150 4. Output-only parameters.
155 /* Stores the features supported by 'netdev' into each of '*current',
156 * '*advertised', '*supported', and '*peer' that are non-null. Each value
157 * is a bitmap of "enum ofp_port_features" bits, in host byte order.
158 * Returns 0 if successful, otherwise a positive errno value. On failure,
159 * all of the passed-in values are set to 0. */
161 netdev_get_features(struct netdev *netdev,
162 uint32_t *current, uint32_t *advertised,
163 uint32_t *supported, uint32_t *peer)
168 Functions that destroy an instance of a dynamically-allocated type
169 should accept and ignore a null pointer argument. Code that calls
170 such a function (including the C standard library function free())
171 should omit a null-pointer check. We find that this usually makes
174 Functions in .c files should not normally be marked "inline", because
175 it does not usually help code generation and it does suppress
176 compilers warnings about unused functions. (Functions defined in .h
177 usually should be marked inline.)
182 Put the return type and function name on the same line in a function
185 static const struct option_class *get_option_class(int code);
188 Omit parameter names from function prototypes when the names do not
189 give useful information, e.g.:
191 int netdev_get_mtu(const struct netdev *, int *mtup);
196 Indent each level of code with 4 spaces. Use BSD-style brace
204 Put a space between "if", "while", "for", etc. and the expressions
207 Enclose single statements in braces:
215 Use comments and blank lines to divide long functions into logical
216 groups of statements.
218 Avoid assignments inside "if" and "while" conditions.
220 Do not put gratuitous parentheses around the expression in a return
221 statement, that is, write "return 0;" and not "return(0);"
223 Write only one statement per line.
225 Indent "switch" statements like this:
227 switch (conn->state) {
229 error = run_connection_input(conn);
237 error = run_connection_output(conn);
244 "switch" statements with very short, uniform cases may use an
248 case 200: return "OK";
249 case 201: return "Created";
250 case 202: return "Accepted";
251 case 204: return "No Content";
252 default: return "Unknown";
255 Use "for (;;)" to write an infinite loop.
257 In an if/else construct where one branch is the "normal" or "common"
258 case and the other branch is the "uncommon" or "error" case, put the
259 common case after the "if", not the "else". This is a form of
260 documentation. It also places the most important code in sequential
261 order without forcing the reader to visually skip past less important
262 details. (Some compilers also assume that the "if" branch is the more
263 common case, so this can be a real form of optimization as well.)
268 For functions that return a success or failure indication, prefer
269 one of the following return value conventions:
271 * An "int" where 0 indicates success and a positive errno value
272 indicates a reason for failure.
274 * A "bool" where true indicates success and false indicates
280 Don't define an object-like macro if an enum can be used instead.
282 Don't define a function-like macro if a "static inline" function
285 If a macro's definition contains multiple statements, enclose them
286 with "do { ... } while (0)" to allow them to work properly in all
287 syntactic circumstances.
289 Do use macros to eliminate the need to update different parts of a
290 single file in parallel, e.g. a list of enums and an array that gives
291 the name of each enum. For example:
293 /* Logging importance levels. */
294 #define VLOG_LEVELS \
295 VLOG_LEVEL(EMER, LOG_ALERT) \
296 VLOG_LEVEL(ERR, LOG_ERR) \
297 VLOG_LEVEL(WARN, LOG_WARNING) \
298 VLOG_LEVEL(INFO, LOG_NOTICE) \
299 VLOG_LEVEL(DBG, LOG_DEBUG)
301 #define VLOG_LEVEL(NAME, SYSLOG_LEVEL) VLL_##NAME,
307 /* Name for each logging level. */
308 static const char *level_names[VLL_N_LEVELS] = {
309 #define VLOG_LEVEL(NAME, SYSLOG_LEVEL) #NAME,
315 THREAD SAFETY ANNOTATIONS
317 Use the macros in lib/compiler.h to annotate locking requirements.
320 static struct ovs_mutex mutex = OVS_MUTEX_INITIALIZER;
321 static struct ovs_rwlock rwlock = OVS_RWLOCK_INITIALIZER;
323 void function_require_plain_mutex(void) OVS_REQUIRES(mutex);
324 void function_require_rwlock(void) OVS_REQ_RDLOCK(rwlock);
326 Pass lock objects, not their addresses, to the annotation macros.
327 (Thus we have OVS_REQUIRES(mutex) above, not OVS_REQUIRES(&mutex).)
332 Each source file should state its license in a comment at the very
333 top, followed by a comment explaining the purpose of the code that is
334 in that file. The comment should explain how the code in the file
335 relates to code in other files. The goal is to allow a programmer to
336 quickly figure out where a given module fits into the larger system.
338 The first non-comment line in a .c source file should be:
342 #include directives should appear in the following order:
344 1. #include <config.h>
346 2. The module's own headers, if any. Including this before any
347 other header (besides <config.h>) ensures that the module's
348 header file is self-contained (see HEADER FILES) below.
350 3. Standard C library headers and other system headers, preferably
351 in alphabetical order. (Occasionally one encounters a set of
352 system headers that must be included in a particular order, in
353 which case that order must take precedence.)
355 4. Open vSwitch headers, in alphabetical order. Use "", not <>,
356 to specify Open vSwitch header names.
361 Each header file should start with its license, as described under
362 SOURCE FILES above, followed by a "header guard" to make the header
363 file idempotent, like so:
370 #endif /* netdev.h */
372 Header files should be self-contained; that is, they should #include
373 whatever additional headers are required, without requiring the client
374 to #include them for it.
376 Don't define the members of a struct or union in a header file,
377 unless client code is actually intended to access them directly or if
378 the definition is otherwise actually needed (e.g. inline functions
379 defined in the header need them).
381 Similarly, don't #include a header file just for the declaration of
382 a struct or union tag (e.g. just for "struct <name>;"). Just declare
383 the tag yourself. This reduces the number of header file
389 Use typedefs sparingly. Code is clearer if the actual type is
390 visible at the point of declaration. Do not, in general, declare a
391 typedef for a struct, union, or enum. Do not declare a typedef for a
392 pointer type, because this can be very confusing to the reader.
394 A function type is a good use for a typedef because it can clarify
395 code. The type should be a function type, not a pointer-to-function
396 type. That way, the typedef name can be used to declare function
397 prototypes. (It cannot be used for function definitions, because that
398 is explicitly prohibited by C89 and C99.)
400 You may assume that "char" is exactly 8 bits and that "int" and
401 "long" are at least 32 bits.
403 Don't assume that "long" is big enough to hold a pointer. If you
404 need to cast a pointer to an integer, use "intptr_t" or "uintptr_t"
407 Use the int<N>_t and uint<N>_t types from <stdint.h> for exact-width
408 integer types. Use the PRId<N>, PRIu<N>, and PRIx<N> macros from
409 <inttypes.h> for formatting them with printf() and related functions.
411 For compatibility with antique printf() implementations:
413 - Instead of "%zu", use "%"PRIuSIZE.
415 - Instead of "%td", use "%"PRIdPTR.
417 - Instead of "%ju", use "%"PRIuMAX.
419 Other variants exist for different radixes. For example, use
420 "%"PRIxSIZE instead of "%zx" or "%x" instead of "%hhx".
422 Also, instead of "%hhd", use "%d". Be cautious substituting "%u",
423 "%x", and "%o" for the corresponding versions with "hh": cast the
424 argument to unsigned char if necessary, because printf("%hhu", -1)
425 prints 255 but printf("%u", -1) prints 4294967295.
427 Use bit-fields sparingly. Do not use bit-fields for layout of
428 network protocol fields or in other circumstances where the exact
431 Declare bit-fields to be type "unsigned int" or "signed int". Do
432 *not* declare bit-fields of type "int": C89 allows these to be either
433 signed or unsigned according to the compiler's whim. (A 1-bit
434 bit-field of type "int" may have a range of -1...0!) Do not declare
435 bit-fields of type _Bool or enum or any other type, because these are
438 Try to order structure members such that they pack well on a system
439 with 2-byte "short", 4-byte "int", and 4- or 8-byte "long" and pointer
440 types. Prefer clear organization over size optimization unless you
441 are convinced there is a size or speed benefit.
443 Pointer declarators bind to the variable name, not the type name.
444 Write "int *x", not "int* x" and definitely not "int * x".
449 Put one space on each side of infix binary and ternary operators:
462 = += -= *= /= %= &= ^= |= <<= >>=
464 Avoid comma operators.
466 Do not put any white space around postfix, prefix, or grouping
472 Exception 1: Put a space after (but not before) the "sizeof" keyword.
473 Exception 2: Put a space between the () used in a cast and the
474 expression whose type is cast: (void *) 0.
476 Break long lines before the ternary operators ? and :, rather than
479 return (out_port != VIGP_CONTROL_PATH
480 ? alpheus_output_port(dp, skb, out_port)
481 : alpheus_output_control(dp, skb, fwd_save_skb(skb),
485 Do not parenthesize the operands of && and || unless operator
486 precedence makes it necessary, or unless the operands are themselves
487 expressions that use && and ||. Thus:
489 if (!isdigit((unsigned char)s[0])
490 || !isdigit((unsigned char)s[1])
491 || !isdigit((unsigned char)s[2])) {
492 printf("string %s does not start with 3-digit code\n", s);
497 if (rule && (!best || rule->priority > best->priority)) {
501 Do parenthesize a subexpression that must be split across more than
504 *idxp = ((l1_idx << PORT_ARRAY_L1_SHIFT)
505 | (l2_idx << PORT_ARRAY_L2_SHIFT)
506 | (l3_idx << PORT_ARRAY_L3_SHIFT));
508 Try to avoid casts. Don't cast the return value of malloc().
510 The "sizeof" operator is unique among C operators in that it accepts
511 two very different kinds of operands: an expression or a type. In
512 general, prefer to specify an expression, e.g. "int *x =
513 xmalloc(sizeof *x);". When the operand of sizeof is an expression,
514 there is no need to parenthesize that operand, and please don't.
516 Use the ARRAY_SIZE macro from lib/util.h to calculate the number of
517 elements in an array.
519 When using a relational operator like "<" or "==", put an expression
520 or variable argument on the left and a constant argument on the
521 right, e.g. "x == 0", *not* "0 == x".
526 Put one blank line between top-level definitions of functions and
532 Most C99 features are OK because they are widely implemented:
534 * Flexible array members (e.g. struct { int foo[]; }).
536 * "static inline" functions (but no other forms of "inline", for
537 which GCC and C99 have differing interpretations).
541 * <stdint.h> and <inttypes.h>.
543 * bool and <stdbool.h>, but don't assume that bool or _Bool can
544 only take on the values 0 or 1, because this behavior can't be
545 simulated on C89 compilers.
547 * Designated initializers (e.g. "struct foo foo = {.a = 1};" and
548 "int a[] = {[2] = 5};").
550 * Mixing of declarations and code within a block. Please use this
551 judiciously; keep declarations nicely grouped together in the
552 beginning of a block if possible.
554 * Use of declarations in iteration statements (e.g.
555 "for (int i = 0; i < 10; i++)").
557 * Use of a trailing comma in an enum declaration (e.g.
560 As a matter of style, avoid // comments.
562 Avoid using GCC or Clang extensions unless you also add a fallback
563 for other compilers. You can, however, use C99 features or GCC
564 extensions also supported by Clang in code that compiles only on
565 GNU/Linux (such as lib/netdev-linux.c), because GCC is the system