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.
10 Limit lines to 79 characters.
12 Use form feeds (control+L) to divide long source files into logical
13 pieces. A form feed should appear as the only character on a line.
15 Do not use tabs for indentation.
17 Avoid trailing spaces on lines.
22 Use names that explain the purpose of a function or object.
24 Use underscores to separate words in an identifier: multi_word_name.
26 Use lowercase for most names. Use uppercase for macros, macro
27 parameters, and members of enumerations.
29 Give arrays names that are plural.
31 Pick a unique name prefix (ending with an underscore) for each
32 module, and apply that prefix to all of that module's externally
33 visible names. Names of macro parameters, struct and union members,
34 and parameters in function prototypes are not considered externally
35 visible for this purpose.
37 Do not use names that begin with _. If you need a name for
38 "internal use only", use __ as a suffix instead of a prefix.
40 Avoid negative names: "found" is a better name than "not_found".
42 In names, a "size" is a count of bytes, a "length" is a count of
43 characters. A buffer has size, but a string has length. The length
44 of a string does not include the null terminator, but the size of the
45 buffer that contains the string does.
50 Comments should be written as full sentences that start with a
51 capital letter and end with a period. Put two spaces between
54 Write block comments as shown below. You may put the /* and */ on
55 the same line as comment text if you prefer.
58 * We redirect stderr to /dev/null because we often want to remove all
59 * traffic control configuration on a port so its in a known state. If
60 * this done when there is no such configuration, tc complains, so we just
64 Each function and each variable declared outside a function, and
65 each struct, union, and typedef declaration should be preceded by a
66 comment. See FUNCTION DEFINITIONS below for function comment
69 Each struct and union member should each have an inline comment that
70 explains its meaning. structs and unions with many members should be
71 additionally divided into logical groups of members by block comments,
74 /* An event that will wake the following call to poll_block(). */
76 /* Set when the waiter is created. */
77 struct list node; /* Element in global waiters list. */
78 int fd; /* File descriptor. */
79 short int events; /* Events to wait for (POLLIN, POLLOUT). */
80 poll_fd_func *function; /* Callback function, if any, or null. */
81 void *aux; /* Argument to callback function. */
82 struct backtrace *backtrace; /* Event that created waiter, or null. */
84 /* Set only when poll_block() is called. */
85 struct pollfd *pollfd; /* Pointer to element of the pollfds array
86 (null if added from a callback). */
89 Use XXX or FIXME comments to mark code that needs work.
91 Don't use // comments.
93 Don't comment out or #if 0 out code. Just remove it. The code that
94 was there will still be in version control history.
99 Put the return type, function name, and the braces that surround the
100 function's code on separate lines, all starting in column 0.
102 Before each function definition, write a comment that describes the
103 function's purpose, including each parameter, the return value, and
104 side effects. References to argument names should be given in
105 single-quotes, e.g. 'arg'. The comment should not include the
106 function name, nor need it follow any formal structure. The comment
107 does not need to describe how a function does its work, unless this
108 information is needed to use the function correctly (this is often
109 better done with comments *inside* the function).
111 Simple static functions do not need a comment.
113 Within a file, non-static functions should come first, in the order
114 that they are declared in the header file, followed by static
115 functions. Static functions should be in one or more separate pages
116 (separated by form feed characters) in logical groups. A commonly
117 useful way to divide groups is by "level", with high-level functions
118 first, followed by groups of progressively lower-level functions.
119 This makes it easy for the program's reader to see the top-down
120 structure by reading from top to bottom.
122 All function declarations and definitions should include a
123 prototype. Empty parentheses, e.g. "int foo();", do not include a
124 prototype (they state that the function's parameters are unknown);
125 write "void" in parentheses instead, e.g. "int foo(void);".
127 Prototypes for static functions should either all go at the top of
128 the file, separated into groups by blank lines, or they should appear
129 at the top of each page of functions. Don't comment individual
130 prototypes, but a comment on each group of prototypes is often
133 In the absence of good reasons for another order, the following
134 parameter order is preferred. One notable exception is that data
135 parameters and their corresponding size parameters should be paired.
137 1. The primary object being manipulated, if any (equivalent to the
138 "this" pointer in C++).
139 2. Input-only parameters.
140 3. Input/output parameters.
141 4. Output-only parameters.
146 /* Stores the features supported by 'netdev' into each of '*current',
147 * '*advertised', '*supported', and '*peer' that are non-null. Each value
148 * is a bitmap of "enum ofp_port_features" bits, in host byte order.
149 * Returns 0 if successful, otherwise a positive errno value. On failure,
150 * all of the passed-in values are set to 0. */
152 netdev_get_features(struct netdev *netdev,
153 uint32_t *current, uint32_t *advertised,
154 uint32_t *supported, uint32_t *peer)
159 Functions that destroy an instance of a dynamically-allocated type
160 should accept and ignore a null pointer argument. Code that calls
161 such a function (including the C standard library function free())
162 should omit a null-pointer check. We find that this usually makes
165 Functions in .c files should not normally be marked "inline", because
166 it does not usually help code generation and it does suppress
167 compilers warnings about unused functions. (Functions defined in .h
168 usually should be marked inline.)
173 Put the return type and function name on the same line in a function
176 static const struct option_class *get_option_class(int code);
179 Omit parameter names from function prototypes when the names do not
180 give useful information, e.g.:
182 int netdev_get_mtu(const struct netdev *, int *mtup);
187 Indent each level of code with 4 spaces. Use BSD-style brace
195 Put a space between "if", "while", "for", etc. and the expressions
198 Enclose single statements in braces:
206 Use comments and blank lines to divide long functions into logical
207 groups of statements.
209 Avoid assignments inside "if" and "while" conditions.
211 Do not put gratuitous parentheses around the expression in a return
212 statement, that is, write "return 0;" and not "return(0);"
214 Write only one statement per line.
216 Indent "switch" statements like this:
218 switch (conn->state) {
220 error = run_connection_input(conn);
228 error = run_connection_output(conn);
235 "switch" statements with very short, uniform cases may use an
239 case 200: return "OK";
240 case 201: return "Created";
241 case 202: return "Accepted";
242 case 204: return "No Content";
243 default: return "Unknown";
246 Use "for (;;)" to write an infinite loop.
248 In an if/else construct where one branch is the "normal" or "common"
249 case and the other branch is the "uncommon" or "error" case, put the
250 common case after the "if", not the "else". This is a form of
251 documentation. It also places the most important code in sequential
252 order without forcing the reader to visually skip past less important
253 details. (Some compilers also assume that the "if" branch is the more
254 common case, so this can be a real form of optimization as well.)
259 For functions that return a success or failure indication, prefer
260 one of the following return value conventions:
262 * An "int" where 0 indicates success and a positive errno value
263 indicates a reason for failure.
265 * A "bool" where true indicates success and false indicates
271 Don't define an object-like macro if an enum can be used instead.
273 Don't define a function-like macro if a "static inline" function
276 If a macro's definition contains multiple statements, enclose them
277 with "do { ... } while (0)" to allow them to work properly in all
278 syntactic circumstances.
280 Do use macros to eliminate the need to update different parts of a
281 single file in parallel, e.g. a list of enums and an array that gives
282 the name of each enum. For example:
284 /* Logging importance levels. */
285 #define VLOG_LEVELS \
286 VLOG_LEVEL(EMER, LOG_ALERT) \
287 VLOG_LEVEL(ERR, LOG_ERR) \
288 VLOG_LEVEL(WARN, LOG_WARNING) \
289 VLOG_LEVEL(INFO, LOG_NOTICE) \
290 VLOG_LEVEL(DBG, LOG_DEBUG)
292 #define VLOG_LEVEL(NAME, SYSLOG_LEVEL) VLL_##NAME,
298 /* Name for each logging level. */
299 static const char *level_names[VLL_N_LEVELS] = {
300 #define VLOG_LEVEL(NAME, SYSLOG_LEVEL) #NAME,
306 THREAD SAFETY ANNOTATIONS
308 Use the macros in lib/compiler.h to annotate locking requirements.
311 static struct ovs_mutex mutex = OVS_MUTEX_INITIALIZER;
312 static struct ovs_rwlock rwlock = OVS_RWLOCK_INITIALIZER;
314 void function_require_plain_mutex(void) OVS_REQUIRES(mutex);
315 void function_require_rwlock(void) OVS_REQ_RDLOCK(rwlock);
317 Pass lock objects, not their addresses, to the annotation macros.
318 (Thus we have OVS_REQUIRES(mutex) above, not OVS_REQUIRES(&mutex).)
323 Each source file should state its license in a comment at the very
324 top, followed by a comment explaining the purpose of the code that is
325 in that file. The comment should explain how the code in the file
326 relates to code in other files. The goal is to allow a programmer to
327 quickly figure out where a given module fits into the larger system.
329 The first non-comment line in a .c source file should be:
333 #include directives should appear in the following order:
335 1. #include <config.h>
337 2. The module's own headers, if any. Including this before any
338 other header (besides <config.h>) ensures that the module's
339 header file is self-contained (see HEADER FILES) below.
341 3. Standard C library headers and other system headers, preferably
342 in alphabetical order. (Occasionally one encounters a set of
343 system headers that must be included in a particular order, in
344 which case that order must take precedence.)
346 4. Open vSwitch headers, in alphabetical order. Use "", not <>,
347 to specify Open vSwitch header names.
352 Each header file should start with its license, as described under
353 SOURCE FILES above, followed by a "header guard" to make the header
354 file idempotent, like so:
361 #endif /* netdev.h */
363 Header files should be self-contained; that is, they should #include
364 whatever additional headers are required, without requiring the client
365 to #include them for it.
367 Don't define the members of a struct or union in a header file,
368 unless client code is actually intended to access them directly or if
369 the definition is otherwise actually needed (e.g. inline functions
370 defined in the header need them).
372 Similarly, don't #include a header file just for the declaration of
373 a struct or union tag (e.g. just for "struct <name>;"). Just declare
374 the tag yourself. This reduces the number of header file
380 Use typedefs sparingly. Code is clearer if the actual type is
381 visible at the point of declaration. Do not, in general, declare a
382 typedef for a struct, union, or enum. Do not declare a typedef for a
383 pointer type, because this can be very confusing to the reader.
385 A function type is a good use for a typedef because it can clarify
386 code. The type should be a function type, not a pointer-to-function
387 type. That way, the typedef name can be used to declare function
388 prototypes. (It cannot be used for function definitions, because that
389 is explicitly prohibited by C89 and C99.)
391 You may assume that "char" is exactly 8 bits and that "int" and
392 "long" are at least 32 bits.
394 Don't assume that "long" is big enough to hold a pointer. If you
395 need to cast a pointer to an integer, use "intptr_t" or "uintptr_t"
398 Use the int<N>_t and uint<N>_t types from <stdint.h> for exact-width
399 integer types. Use the PRId<N>, PRIu<N>, and PRIx<N> macros from
400 <inttypes.h> for formatting them with printf() and related functions.
402 For compatibility with antique printf() implementations:
404 - Instead of "%zu", use "%"PRIuSIZE.
406 - Instead of "%td", use "%"PRIdPTR.
408 - Instead of "%ju", use "%"PRIuMAX.
410 Other variants exist for different radixes. For example, use
411 "%"PRIxSIZE instead of "%zx" or "%x" instead of "%hhx".
413 Also, instead of "%hhd", use "%d". Be cautious substituting "%u",
414 "%x", and "%o" for the corresponding versions with "hh": cast the
415 argument to unsigned char if necessary, because printf("%hhu", -1)
416 prints 255 but printf("%u", -1) prints 4294967295.
418 Use bit-fields sparingly. Do not use bit-fields for layout of
419 network protocol fields or in other circumstances where the exact
422 Declare bit-fields to be type "unsigned int" or "signed int". Do
423 *not* declare bit-fields of type "int": C89 allows these to be either
424 signed or unsigned according to the compiler's whim. (A 1-bit
425 bit-field of type "int" may have a range of -1...0!) Do not declare
426 bit-fields of type _Bool or enum or any other type, because these are
429 Try to order structure members such that they pack well on a system
430 with 2-byte "short", 4-byte "int", and 4- or 8-byte "long" and pointer
431 types. Prefer clear organization over size optimization unless you
432 are convinced there is a size or speed benefit.
434 Pointer declarators bind to the variable name, not the type name.
435 Write "int *x", not "int* x" and definitely not "int * x".
440 Put one space on each side of infix binary and ternary operators:
453 = += -= *= /= %= &= ^= |= <<= >>=
455 Avoid comma operators.
457 Do not put any white space around postfix, prefix, or grouping
463 Exception 1: Put a space after (but not before) the "sizeof" keyword.
464 Exception 2: Put a space between the () used in a cast and the
465 expression whose type is cast: (void *) 0.
467 Break long lines before the ternary operators ? and :, rather than
470 return (out_port != VIGP_CONTROL_PATH
471 ? alpheus_output_port(dp, skb, out_port)
472 : alpheus_output_control(dp, skb, fwd_save_skb(skb),
476 Do not parenthesize the operands of && and || unless operator
477 precedence makes it necessary, or unless the operands are themselves
478 expressions that use && and ||. Thus:
480 if (!isdigit((unsigned char)s[0])
481 || !isdigit((unsigned char)s[1])
482 || !isdigit((unsigned char)s[2])) {
483 printf("string %s does not start with 3-digit code\n", s);
488 if (rule && (!best || rule->priority > best->priority)) {
492 Do parenthesize a subexpression that must be split across more than
495 *idxp = ((l1_idx << PORT_ARRAY_L1_SHIFT)
496 | (l2_idx << PORT_ARRAY_L2_SHIFT)
497 | (l3_idx << PORT_ARRAY_L3_SHIFT));
499 Try to avoid casts. Don't cast the return value of malloc().
501 The "sizeof" operator is unique among C operators in that it accepts
502 two very different kinds of operands: an expression or a type. In
503 general, prefer to specify an expression, e.g. "int *x =
504 xmalloc(sizeof *x);". When the operand of sizeof is an expression,
505 there is no need to parenthesize that operand, and please don't.
507 Use the ARRAY_SIZE macro from lib/util.h to calculate the number of
508 elements in an array.
510 When using a relational operator like "<" or "==", put an expression
511 or variable argument on the left and a constant argument on the
512 right, e.g. "x == 0", *not* "0 == x".
517 Put one blank line between top-level definitions of functions and
523 Most C99 features are OK because they are widely implemented:
525 * Flexible array members (e.g. struct { int foo[]; }).
527 * "static inline" functions (but no other forms of "inline", for
528 which GCC and C99 have differing interpretations).
532 * <stdint.h> and <inttypes.h>.
534 * bool and <stdbool.h>, but don't assume that bool or _Bool can
535 only take on the values 0 or 1, because this behavior can't be
536 simulated on C89 compilers.
538 * Designated initializers (e.g. "struct foo foo = {.a = 1};" and
539 "int a[] = {[2] = 5};").
541 * Mixing of declarations and code within a block. Please use this
542 judiciously; keep declarations nicely grouped together in the
543 beginning of a block if possible.
545 * Use of declarations in iteration statements (e.g.
546 "for (int i = 0; i < 10; i++)").
548 * Use of a trailing comma in an enum declaration (e.g.
551 As a matter of style, avoid // comments.
553 Avoid using GCC or Clang extensions unless you also add a fallback
554 for other compilers. You can, however, use C99 features or GCC
555 extensions also supported by Clang in code that compiles only on
556 GNU/Linux (such as lib/netdev-linux.c), because GCC is the system