1 .. _codingstyle:
2
3 Linux kernel coding style
4 =========================
5
6 This is a short document describing the preferred coding style for the
7 linux kernel. Coding style is very personal, and I won't **force** my
8 views on anybody, but this is what goes for anything that I have to be
9 able to maintain, and I'd prefer it for most other things too. Please
10 at least consider the points made here.
11
12 First off, I'd suggest printing out a copy of the GNU coding standards,
13 and NOT read it. Burn them, it's a great symbolic gesture.
14
15 Anyway, here goes:
16
17
18 1) Indentation
19 --------------
20
21 Tabs are 8 characters, and thus indentations are also 8 characters.
22 There are heretic movements that try to make indentations 4 (or even 2!)
23 characters deep, and that is akin to trying to define the value of PI to
24 be 3.
25
26 Rationale: The whole idea behind indentation is to clearly define where
27 a block of control starts and ends. Especially when you've been looking
28 at your screen for 20 straight hours, you'll find it a lot easier to see
29 how the indentation works if you have large indentations.
30
31 Now, some people will claim that having 8-character indentations makes
32 the code move too far to the right, and makes it hard to read on a
33 80-character terminal screen. The answer to that is that if you need
34 more than 3 levels of indentation, you're screwed anyway, and should fix
35 your program.
36
37 In short, 8-char indents make things easier to read, and have the added
38 benefit of warning you when you're nesting your functions too deep.
39 Heed that warning.
40
41 The preferred way to ease multiple indentation levels in a switch statement is
42 to align the ``switch`` and its subordinate ``case`` labels in the same column
43 instead of ``double-indenting`` the ``case`` labels. E.g.:
44
45 .. code-block:: c
46
47 switch (suffix) {
48 case 'G':
49 case 'g':
50 mem <<= 30;
51 break;
52 case 'M':
53 case 'm':
54 mem <<= 20;
55 break;
56 case 'K':
57 case 'k':
58 mem <<= 10;
59 fallthrough;
60 default:
61 break;
62 }
63
64 Don't put multiple statements on a single line unless you have
65 something to hide:
66
67 .. code-block:: c
68
69 if (condition) do_this;
70 do_something_everytime;
71
72 Don't use commas to avoid using braces:
73
74 .. code-block:: c
75
76 if (condition)
77 do_this(), do_that();
78
79 Always uses braces for multiple statements:
80
81 .. code-block:: c
82
83 if (condition) {
84 do_this();
85 do_that();
86 }
87
88 Don't put multiple assignments on a single line either. Kernel coding style
89 is super simple. Avoid tricky expressions.
90
91
92 Outside of comments, documentation and except in Kconfig, spaces are never
93 used for indentation, and the above example is deliberately broken.
94
95 Get a decent editor and don't leave whitespace at the end of lines.
96
97
98 2) Breaking long lines and strings
99 ----------------------------------
100
101 Coding style is all about readability and maintainability using commonly
102 available tools.
103
104 The preferred limit on the length of a single line is 80 columns.
105
106 Statements longer than 80 columns should be broken into sensible chunks,
107 unless exceeding 80 columns significantly increases readability and does
108 not hide information.
109
110 Descendants are always substantially shorter than the parent and
111 are placed substantially to the right. A very commonly used style
112 is to align descendants to a function open parenthesis.
113
114 These same rules are applied to function headers with a long argument list.
115
116 However, never break user-visible strings such as printk messages because
117 that breaks the ability to grep for them.
118
119
120 3) Placing Braces and Spaces
121 ----------------------------
122
123 The other issue that always comes up in C styling is the placement of
124 braces. Unlike the indent size, there are few technical reasons to
125 choose one placement strategy over the other, but the preferred way, as
126 shown to us by the prophets Kernighan and Ritchie, is to put the opening
127 brace last on the line, and put the closing brace first, thusly:
128
129 .. code-block:: c
130
131 if (x is true) {
132 we do y
133 }
134
135 This applies to all non-function statement blocks (if, switch, for,
136 while, do). E.g.:
137
138 .. code-block:: c
139
140 switch (action) {
141 case KOBJ_ADD:
142 return "add";
143 case KOBJ_REMOVE:
144 return "remove";
145 case KOBJ_CHANGE:
146 return "change";
147 default:
148 return NULL;
149 }
150
151 However, there is one special case, namely functions: they have the
152 opening brace at the beginning of the next line, thus:
153
154 .. code-block:: c
155
156 int function(int x)
157 {
158 body of function
159 }
160
161 Heretic people all over the world have claimed that this inconsistency
162 is ... well ... inconsistent, but all right-thinking people know that
163 (a) K&R are **right** and (b) K&R are right. Besides, functions are
164 special anyway (you can't nest them in C).
165
166 Note that the closing brace is empty on a line of its own, **except** in
167 the cases where it is followed by a continuation of the same statement,
168 ie a ``while`` in a do-statement or an ``else`` in an if-statement, like
169 this:
170
171 .. code-block:: c
172
173 do {
174 body of do-loop
175 } while (condition);
176
177 and
178
179 .. code-block:: c
180
181 if (x == y) {
182 ..
183 } else if (x > y) {
184 ...
185 } else {
186 ....
187 }
188
189 Rationale: K&R.
190
191 Also, note that this brace-placement also minimizes the number of empty
192 (or almost empty) lines, without any loss of readability. Thus, as the
193 supply of new-lines on your screen is not a renewable resource (think
194 25-line terminal screens here), you have more empty lines to put
195 comments on.
196
197 Do not unnecessarily use braces where a single statement will do.
198
199 .. code-block:: c
200
201 if (condition)
202 action();
203
204 and
205
206 .. code-block:: c
207
208 if (condition)
209 do_this();
210 else
211 do_that();
212
213 This does not apply if only one branch of a conditional statement is a single
214 statement; in the latter case use braces in both branches:
215
216 .. code-block:: c
217
218 if (condition) {
219 do_this();
220 do_that();
221 } else {
222 otherwise();
223 }
224
225 Also, use braces when a loop contains more than a single simple statement:
226
227 .. code-block:: c
228
229 while (condition) {
230 if (test)
231 do_something();
232 }
233
234 3.1) Spaces
235 ***********
236
237 Linux kernel style for use of spaces depends (mostly) on
238 function-versus-keyword usage. Use a space after (most) keywords. The
239 notable exceptions are sizeof, typeof, alignof, and __attribute__, which look
240 somewhat like functions (and are usually used with parentheses in Linux,
241 although they are not required in the language, as in: ``sizeof info`` after
242 ``struct fileinfo info;`` is declared).
243
244 So use a space after these keywords::
245
246 if, switch, case, for, do, while
247
248 but not with sizeof, typeof, alignof, or __attribute__. E.g.,
249
250 .. code-block:: c
251
252
253 s = sizeof(struct file);
254
255 Do not add spaces around (inside) parenthesized expressions. This example is
256 **bad**:
257
258 .. code-block:: c
259
260
261 s = sizeof( struct file );
262
263 When declaring pointer data or a function that returns a pointer type, the
264 preferred use of ``*`` is adjacent to the data name or function name and not
265 adjacent to the type name. Examples:
266
267 .. code-block:: c
268
269
270 char *linux_banner;
271 unsigned long long memparse(char *ptr, char **retptr);
272 char *match_strdup(substring_t *s);
273
274 Use one space around (on each side of) most binary and ternary operators,
275 such as any of these::
276
277 = + - < > * / % | & ^ <= >= == != ? :
278
279 but no space after unary operators::
280
281 & * + - ~ ! sizeof typeof alignof __attribute__ defined
282
283 no space before the postfix increment & decrement unary operators::
284
285 ++ --
286
287 no space after the prefix increment & decrement unary operators::
288
289 ++ --
290
291 and no space around the ``.`` and ``->`` structure member operators.
292
293 Do not leave trailing whitespace at the ends of lines. Some editors with
294 ``smart`` indentation will insert whitespace at the beginning of new lines as
295 appropriate, so you can start typing the next line of code right away.
296 However, some such editors do not remove the whitespace if you end up not
297 putting a line of code there, such as if you leave a blank line. As a result,
298 you end up with lines containing trailing whitespace.
299
300 Git will warn you about patches that introduce trailing whitespace, and can
301 optionally strip the trailing whitespace for you; however, if applying a series
302 of patches, this may make later patches in the series fail by changing their
303 context lines.
304
305
306 4) Naming
307 ---------
308
309 C is a Spartan language, and your naming conventions should follow suit.
310 Unlike Modula-2 and Pascal programmers, C programmers do not use cute
311 names like ThisVariableIsATemporaryCounter. A C programmer would call that
312 variable ``tmp``, which is much easier to write, and not the least more
313 difficult to understand.
314
315 HOWEVER, while mixed-case names are frowned upon, descriptive names for
316 global variables are a must. To call a global function ``foo`` is a
317 shooting offense.
318
319 GLOBAL variables (to be used only if you **really** need them) need to
320 have descriptive names, as do global functions. If you have a function
321 that counts the number of active users, you should call that
322 ``count_active_users()`` or similar, you should **not** call it ``cntusr()``.
323
324 Encoding the type of a function into the name (so-called Hungarian
325 notation) is asinine - the compiler knows the types anyway and can check
326 those, and it only confuses the programmer.
327
328 LOCAL variable names should be short, and to the point. If you have
329 some random integer loop counter, it should probably be called ``i``.
330 Calling it ``loop_counter`` is non-productive, if there is no chance of it
331 being mis-understood. Similarly, ``tmp`` can be just about any type of
332 variable that is used to hold a temporary value.
333
334 If you are afraid to mix up your local variable names, you have another
335 problem, which is called the function-growth-hormone-imbalance syndrome.
336 See chapter 6 (Functions).
337
338 For symbol names and documentation, avoid introducing new usage of
339 'master / slave' (or 'slave' independent of 'master') and 'blacklist /
340 whitelist'.
341
342 Recommended replacements for 'master / slave' are:
343 '{primary,main} / {secondary,replica,subordinate}'
344 '{initiator,requester} / {target,responder}'
345 '{controller,host} / {device,worker,proxy}'
346 'leader / follower'
347 'director / performer'
348
349 Recommended replacements for 'blacklist/whitelist' are:
350 'denylist / allowlist'
351 'blocklist / passlist'
352
353 Exceptions for introducing new usage is to maintain a userspace ABI/API,
354 or when updating code for an existing (as of 2020) hardware or protocol
355 specification that mandates those terms. For new specifications
356 translate specification usage of the terminology to the kernel coding
357 standard where possible.
358
359 5) Typedefs
360 -----------
361
362 Please don't use things like ``vps_t``.
363 It's a **mistake** to use typedef for structures and pointers. When you see a
364
365 .. code-block:: c
366
367
368 vps_t a;
369
370 in the source, what does it mean?
371 In contrast, if it says
372
373 .. code-block:: c
374
375 struct virtual_container *a;
376
377 you can actually tell what ``a`` is.
378
379 Lots of people think that typedefs ``help readability``. Not so. They are
380 useful only for:
381
382 (a) totally opaque objects (where the typedef is actively used to **hide**
383 what the object is).
384
385 Example: ``pte_t`` etc. opaque objects that you can only access using
386 the proper accessor functions.
387
388 .. note::
389
390 Opaqueness and ``accessor functions`` are not good in themselves.
391 The reason we have them for things like pte_t etc. is that there
392 really is absolutely **zero** portably accessible information there.
393
394 (b) Clear integer types, where the abstraction **helps** avoid confusion
395 whether it is ``int`` or ``long``.
396
397 u8/u16/u32 are perfectly fine typedefs, although they fit into
398 category (d) better than here.
399
400 .. note::
401
402 Again - there needs to be a **reason** for this. If something is
403 ``unsigned long``, then there's no reason to do
404
405 typedef unsigned long myflags_t;
406
407 but if there is a clear reason for why it under certain circumstances
408 might be an ``unsigned int`` and under other configurations might be
409 ``unsigned long``, then by all means go ahead and use a typedef.
410
411 (c) when you use sparse to literally create a **new** type for
412 type-checking.
413
414 (d) New types which are identical to standard C99 types, in certain
415 exceptional circumstances.
416
417 Although it would only take a short amount of time for the eyes and
418 brain to become accustomed to the standard types like ``uint32_t``,
419 some people object to their use anyway.
420
421 Therefore, the Linux-specific ``u8/u16/u32/u64`` types and their
422 signed equivalents which are identical to standard types are
423 permitted -- although they are not mandatory in new code of your
424 own.
425
426 When editing existing code which already uses one or the other set
427 of types, you should conform to the existing choices in that code.
428
429 (e) Types safe for use in userspace.
430
431 In certain structures which are visible to userspace, we cannot
432 require C99 types and cannot use the ``u32`` form above. Thus, we
433 use __u32 and similar types in all structures which are shared
434 with userspace.
435
436 Maybe there are other cases too, but the rule should basically be to NEVER
437 EVER use a typedef unless you can clearly match one of those rules.
438
439 In general, a pointer, or a struct that has elements that can reasonably
440 be directly accessed should **never** be a typedef.
441
442
443 6) Functions
444 ------------
445
446 Functions should be short and sweet, and do just one thing. They should
447 fit on one or two screenfuls of text (the ISO/ANSI screen size is 80x24,
448 as we all know), and do one thing and do that well.
449
450 The maximum length of a function is inversely proportional to the
451 complexity and indentation level of that function. So, if you have a
452 conceptually simple function that is just one long (but simple)
453 case-statement, where you have to do lots of small things for a lot of
454 different cases, it's OK to have a longer function.
455
456 However, if you have a complex function, and you suspect that a
457 less-than-gifted first-year high-school student might not even
458 understand what the function is all about, you should adhere to the
459 maximum limits all the more closely. Use helper functions with
460 descriptive names (you can ask the compiler to in-line them if you think
461 it's performance-critical, and it will probably do a better job of it
462 than you would have done).
463
464 Another measure of the function is the number of local variables. They
465 shouldn't exceed 5-10, or you're doing something wrong. Re-think the
466 function, and split it into smaller pieces. A human brain can
467 generally easily keep track of about 7 different things, anything more
468 and it gets confused. You know you're brilliant, but maybe you'd like
469 to understand what you did 2 weeks from now.
470
471 In source files, separate functions with one blank line. If the function is
472 exported, the **EXPORT** macro for it should follow immediately after the
473 closing function brace line. E.g.:
474
475 .. code-block:: c
476
477 int system_is_up(void)
478 {
479 return system_state == SYSTEM_RUNNING;
480 }
481 EXPORT_SYMBOL(system_is_up);
482
483 6.1) Function prototypes
484 ************************
485
486 In function prototypes, include parameter names with their data types.
487 Although this is not required by the C language, it is preferred in Linux
488 because it is a simple way to add valuable information for the reader.
489
490 Do not use the ``extern`` keyword with function declarations as this makes
491 lines longer and isn't strictly necessary.
492
493 When writing function prototypes, please keep the `order of elements regular
494 <https://lore.kernel.org/mm-commits/CAHk-=wiOCLRny5aifWNhr621kYrJwhfURsa0vFPeUEm8mF0ufg@mail.gmail.com/">https://lore.kernel.org/mm-commits/CAHk-=wiOCLRny5aifWNhr621kYrJwhfURsa0vFPeUEm8mF0ufg@mail.gmail.com/>`_.
495 For example, using this function declaration example::
496
497 __init void * __must_check action(enum magic value, size_t size, u8 count,
498 char *fmt, ...) __printf(4, 5) __malloc;
499
500 The preferred order of elements for a function prototype is:
501
502 - storage class (below, ``static __always_inline``, noting that ``__always_inline``
503 is technically an attribute but is treated like ``inline``)
504 - storage class attributes (here, ``__init`` -- i.e. section declarations, but also
505 things like ``__cold``)
506 - return type (here, ``void *``)
507 - return type attributes (here, ``__must_check``)
508 - function name (here, ``action``)
509 - function parameters (here, ``(enum magic value, size_t size, u8 count, char *fmt, ...)``,
510 noting that parameter names should always be included)
511 - function parameter attributes (here, ``__printf(4, 5)``)
512 - function behavior attributes (here, ``__malloc``)
513
514 Note that for a function **definition** (i.e. the actual function body),
515 the compiler does not allow function parameter attributes after the
516 function parameters. In these cases, they should go after the storage
517 class attributes (e.g. note the changed position of ``__printf(4, 5)``
518 below, compared to the **declaration** example above)::
519
520 static __always_inline __init __printf(4, 5) void * __must_check action(enum magic value,
521 size_t size, u8 count, char *fmt, ...) __malloc
522 {
523 ...
524 }
525
526 7) Centralized exiting of functions
527 -----------------------------------
528
529 Albeit deprecated by some people, the equivalent of the goto statement is
530 used frequently by compilers in form of the unconditional jump instruction.
531
532 The goto statement comes in handy when a function exits from multiple
533 locations and some common work such as cleanup has to be done. If there is no
534 cleanup needed then just return directly.
535
536 Choose label names which say what the goto does or why the goto exists. An
537 example of a good name could be ``out_free_buffer:`` if the goto frees ``buffer``.
538 Avoid using GW-BASIC names like ``err1:`` and ``err2:``, as you would have to
539 renumber them if you ever add or remove exit paths, and they make correctness
540 difficult to verify anyway.
541
542 The rationale for using gotos is:
543
544 - unconditional statements are easier to understand and follow
545 - nesting is reduced
546 - errors by not updating individual exit points when making
547 modifications are prevented
548 - saves the compiler work to optimize redundant code away ;)
549
550 .. code-block:: c
551
552 int fun(int a)
553 {
554 int result = 0;
555 char *buffer;
556
557 buffer = kmalloc(SIZE, GFP_KERNEL);
558 if (!buffer)
559 return -ENOMEM;
560
561 if (condition1) {
562 while (loop1) {
563 ...
564 }
565 result = 1;
566 goto out_free_buffer;
567 }
568 ...
569 out_free_buffer:
570 kfree(buffer);
571 return result;
572 }
573
574 A common type of bug to be aware of is ``one err bugs`` which look like this:
575
576 .. code-block:: c
577
578 err:
579 kfree(foo->bar);
580 kfree(foo);
581 return ret;
582
583 The bug in this code is that on some exit paths ``foo`` is NULL. Normally the
584 fix for this is to split it up into two error labels ``err_free_bar:`` and
585 ``err_free_foo:``:
586
587 .. code-block:: c
588
589 err_free_bar:
590 kfree(foo->bar);
591 err_free_foo:
592 kfree(foo);
593 return ret;
594
595 Ideally you should simulate errors to test all exit paths.
596
597
598 8) Commenting
599 -------------
600
601 Comments are good, but there is also a danger of over-commenting. NEVER
602 try to explain HOW your code works in a comment: it's much better to
603 write the code so that the **working** is obvious, and it's a waste of
604 time to explain badly written code.
605
606 Generally, you want your comments to tell WHAT your code does, not HOW.
607 Also, try to avoid putting comments inside a function body: if the
608 function is so complex that you need to separately comment parts of it,
609 you should probably go back to chapter 6 for a while. You can make
610 small comments to note or warn about something particularly clever (or
611 ugly), but try to avoid excess. Instead, put the comments at the head
612 of the function, telling people what it does, and possibly WHY it does
613 it.
614
615 When commenting the kernel API functions, please use the kernel-doc format.
616 See the files at :ref:`Documentation/doc-guide/ <doc_guide>` and
617 ``scripts/kernel-doc`` for details.
618
619 The preferred style for long (multi-line) comments is:
620
621 .. code-block:: c
622
623 /*
624 * This is the preferred style for multi-line
625 * comments in the Linux kernel source code.
626 * Please use it consistently.
627 *
628 * Description: A column of asterisks on the left side,
629 * with beginning and ending almost-blank lines.
630 */
631
632 It's also important to comment data, whether they are basic types or derived
633 types. To this end, use just one data declaration per line (no commas for
634 multiple data declarations). This leaves you room for a small comment on each
635 item, explaining its use.
636
637
638 9) You've made a mess of it
639 ---------------------------
640
641 That's OK, we all do. You've probably been told by your long-time Unix
642 user helper that ``GNU emacs`` automatically formats the C sources for
643 you, and you've noticed that yes, it does do that, but the defaults it
644 uses are less than desirable (in fact, they are worse than random
645 typing - an infinite number of monkeys typing into GNU emacs would never
646 make a good program).
647
648 So, you can either get rid of GNU emacs, or change it to use saner
649 values. To do the latter, you can stick the following in your .emacs file:
650
651 .. code-block:: elisp
652
653 (defun c-lineup-arglist-tabs-only (ignored)
654 "Line up argument lists by tabs, not spaces"
655 (let* ((anchor (c-langelem-pos c-syntactic-element))
656 (column (c-langelem-2nd-pos c-syntactic-element))
657 (offset (- (1+ column) anchor))
658 (steps (floor offset c-basic-offset)))
659 (* (max steps 1)
660 c-basic-offset)))
661
662 (dir-locals-set-class-variables
663 'linux-kernel
664 '((c-mode . (
665 (c-basic-offset . 8)
666 (c-label-minimum-indentation . 0)
667 (c-offsets-alist . (
668 (arglist-close . c-lineup-arglist-tabs-only)
669 (arglist-cont-nonempty .
670 (c-lineup-gcc-asm-reg c-lineup-arglist-tabs-only))
671 (arglist-intro . +)
672 (brace-list-intro . +)
673 (c . c-lineup-C-comments)
674 (case-label . 0)
675 (comment-intro . c-lineup-comment)
676 (cpp-define-intro . +)
677 (cpp-macro . -1000)
678 (cpp-macro-cont . +)
679 (defun-block-intro . +)
680 (else-clause . 0)
681 (func-decl-cont . +)
682 (inclass . +)
683 (inher-cont . c-lineup-multi-inher)
684 (knr-argdecl-intro . 0)
685 (label . -1000)
686 (statement . 0)
687 (statement-block-intro . +)
688 (statement-case-intro . +)
689 (statement-cont . +)
690 (substatement . +)
691 ))
692 (indent-tabs-mode . t)
693 (show-trailing-whitespace . t)
694 ))))
695
696 (dir-locals-set-directory-class
697 (expand-file-name "~/src/linux-trees")
698 'linux-kernel)
699
700 This will make emacs go better with the kernel coding style for C
701 files below ``~/src/linux-trees``.
702
703 But even if you fail in getting emacs to do sane formatting, not
704 everything is lost: use ``indent``.
705
706 Now, again, GNU indent has the same brain-dead settings that GNU emacs
707 has, which is why you need to give it a few command line options.
708 However, that's not too bad, because even the makers of GNU indent
709 recognize the authority of K&R (the GNU people aren't evil, they are
710 just severely misguided in this matter), so you just give indent the
711 options ``-kr -i8`` (stands for ``K&R, 8 character indents``), or use
712 ``scripts/Lindent``, which indents in the latest style.
713
714 ``indent`` has a lot of options, and especially when it comes to comment
715 re-formatting you may want to take a look at the man page. But
716 remember: ``indent`` is not a fix for bad programming.
717
718 Note that you can also use the ``clang-format`` tool to help you with
719 these rules, to quickly re-format parts of your code automatically,
720 and to review full files in order to spot coding style mistakes,
721 typos and possible improvements. It is also handy for sorting ``#includes``,
722 for aligning variables/macros, for reflowing text and other similar tasks.
723 See the file :ref:`Documentation/dev-tools/clang-format.rst <clangformat>`
724 for more details.
725
726 Some basic editor settings, such as indentation and line endings, will be
727 set automatically if you are using an editor that is compatible with
728 EditorConfig. See the official EditorConfig website for more information:
729 https://editorconfig.org/
730
731 10) Kconfig configuration files
732 -------------------------------
733
734 For all of the Kconfig* configuration files throughout the source tree,
735 the indentation is somewhat different. Lines under a ``config`` definition
736 are indented with one tab, while help text is indented an additional two
737 spaces. Example::
738
739 config AUDIT
740 bool "Auditing support"
741 depends on NET
742 help
743 Enable auditing infrastructure that can be used with another
744 kernel subsystem, such as SELinux (which requires this for
745 logging of avc messages output). Does not do system-call
746 auditing without CONFIG_AUDITSYSCALL.
747
748 Seriously dangerous features (such as write support for certain
749 filesystems) should advertise this prominently in their prompt string::
750
751 config ADFS_FS_RW
752 bool "ADFS write support (DANGEROUS)"
753 depends on ADFS_FS
754 ...
755
756 For full documentation on the configuration files, see the file
757 Documentation/kbuild/kconfig-language.rst.
758
759
760 11) Data structures
761 -------------------
762
763 Data structures that have visibility outside the single-threaded
764 environment they are created and destroyed in should always have
765 reference counts. In the kernel, garbage collection doesn't exist (and
766 outside the kernel garbage collection is slow and inefficient), which
767 means that you absolutely **have** to reference count all your uses.
768
769 Reference counting means that you can avoid locking, and allows multiple
770 users to have access to the data structure in parallel - and not having
771 to worry about the structure suddenly going away from under them just
772 because they slept or did something else for a while.
773
774 Note that locking is **not** a replacement for reference counting.
775 Locking is used to keep data structures coherent, while reference
776 counting is a memory management technique. Usually both are needed, and
777 they are not to be confused with each other.
778
779 Many data structures can indeed have two levels of reference counting,
780 when there are users of different ``classes``. The subclass count counts
781 the number of subclass users, and decrements the global count just once
782 when the subclass count goes to zero.
783
784 Examples of this kind of ``multi-level-reference-counting`` can be found in
785 memory management (``struct mm_struct``: mm_users and mm_count), and in
786 filesystem code (``struct super_block``: s_count and s_active).
787
788 Remember: if another thread can find your data structure, and you don't
789 have a reference count on it, you almost certainly have a bug.
790
791
792 12) Macros, Enums and RTL
793 -------------------------
794
795 Names of macros defining constants and labels in enums are capitalized.
796
797 .. code-block:: c
798
799 #define CONSTANT 0x12345
800
801 Enums are preferred when defining several related constants.
802
803 CAPITALIZED macro names are appreciated but macros resembling functions
804 may be named in lower case.
805
806 Generally, inline functions are preferable to macros resembling functions.
807
808 Macros with multiple statements should be enclosed in a do - while block:
809
810 .. code-block:: c
811
812 #define macrofun(a, b, c) \
813 do { \
814 if (a == 5) \
815 do_this(b, c); \
816 } while (0)
817
818 Function-like macros with unused parameters should be replaced by static
819 inline functions to avoid the issue of unused variables:
820
821 .. code-block:: c
822
823 static inline void fun(struct foo *foo)
824 {
825 }
826
827 Due to historical practices, many files still employ the "cast to (void)"
828 approach to evaluate parameters. However, this method is not advisable.
829 Inline functions address the issue of "expression with side effects
830 evaluated more than once", circumvent unused-variable problems, and
831 are generally better documented than macros for some reason.
832
833 .. code-block:: c
834
835 /*
836 * Avoid doing this whenever possible and instead opt for static
837 * inline functions
838 */
839 #define macrofun(foo) do { (void) (foo); } while (0)
840
841 Things to avoid when using macros:
842
843 1) macros that affect control flow:
844
845 .. code-block:: c
846
847 #define FOO(x) \
848 do { \
849 if (blah(x) < 0) \
850 return -EBUGGERED; \
851 } while (0)
852
853 is a **very** bad idea. It looks like a function call but exits the ``calling``
854 function; don't break the internal parsers of those who will read the code.
855
856 2) macros that depend on having a local variable with a magic name:
857
858 .. code-block:: c
859
860 #define FOO(val) bar(index, val)
861
862 might look like a good thing, but it's confusing as hell when one reads the
863 code and it's prone to breakage from seemingly innocent changes.
864
865 3) macros with arguments that are used as l-values: FOO(x) = y; will
866 bite you if somebody e.g. turns FOO into an inline function.
867
868 4) forgetting about precedence: macros defining constants using expressions
869 must enclose the expression in parentheses. Beware of similar issues with
870 macros using parameters.
871
872 .. code-block:: c
873
874 #define CONSTANT 0x4000
875 #define CONSTEXP (CONSTANT | 3)
876
877 5) namespace collisions when defining local variables in macros resembling
878 functions:
879
880 .. code-block:: c
881
882 #define FOO(x) \
883 ({ \
884 typeof(x) ret; \
885 ret = calc_ret(x); \
886 (ret); \
887 })
888
889 ret is a common name for a local variable - __foo_ret is less likely
890 to collide with an existing variable.
891
892 The cpp manual deals with macros exhaustively. The gcc internals manual also
893 covers RTL which is used frequently with assembly language in the kernel.
894
895
896 13) Printing kernel messages
897 ----------------------------
898
899 Kernel developers like to be seen as literate. Do mind the spelling
900 of kernel messages to make a good impression. Do not use incorrect
901 contractions like ``dont``; use ``do not`` or ``don't`` instead. Make the
902 messages concise, clear, and unambiguous.
903
904 Kernel messages do not have to be terminated with a period.
905
906 Printing numbers in parentheses (%d) adds no value and should be avoided.
907
908 There are a number of driver model diagnostic macros in <linux/dev_printk.h>
909 which you should use to make sure messages are matched to the right device
910 and driver, and are tagged with the right level: dev_err(), dev_warn(),
911 dev_info(), and so forth. For messages that aren't associated with a
912 particular device, <linux/printk.h> defines pr_notice(), pr_info(),
913 pr_warn(), pr_err(), etc. When drivers are working properly they are quiet,
914 so prefer to use dev_dbg/pr_debug unless something is wrong.
915
916 Coming up with good debugging messages can be quite a challenge; and once
917 you have them, they can be a huge help for remote troubleshooting. However
918 debug message printing is handled differently than printing other non-debug
919 messages. While the other pr_XXX() functions print unconditionally,
920 pr_debug() does not; it is compiled out by default, unless either DEBUG is
921 defined or CONFIG_DYNAMIC_DEBUG is set. That is true for dev_dbg() also,
922 and a related convention uses VERBOSE_DEBUG to add dev_vdbg() messages to
923 the ones already enabled by DEBUG.
924
925 Many subsystems have Kconfig debug options to turn on -DDEBUG in the
926 corresponding Makefile; in other cases specific files #define DEBUG. And
927 when a debug message should be unconditionally printed, such as if it is
928 already inside a debug-related #ifdef section, printk(KERN_DEBUG ...) can be
929 used.
930
931
932 14) Allocating memory
933 ---------------------
934
935 The kernel provides the following general purpose memory allocators:
936 kmalloc(), kzalloc(), kmalloc_array(), kcalloc(), vmalloc(), and
937 vzalloc(). Please refer to the API documentation for further information
938 about them. :ref:`Documentation/core-api/memory-allocation.rst
939 <memory_allocation>`
940
941 The preferred form for passing a size of a struct is the following:
942
943 .. code-block:: c
944
945 p = kmalloc(sizeof(*p), ...);
946
947 The alternative form where struct name is spelled out hurts readability and
948 introduces an opportunity for a bug when the pointer variable type is changed
949 but the corresponding sizeof that is passed to a memory allocator is not.
950
951 Casting the return value which is a void pointer is redundant. The conversion
952 from void pointer to any other pointer type is guaranteed by the C programming
953 language.
954
955 The preferred form for allocating an array is the following:
956
957 .. code-block:: c
958
959 p = kmalloc_array(n, sizeof(...), ...);
960
961 The preferred form for allocating a zeroed array is the following:
962
963 .. code-block:: c
964
965 p = kcalloc(n, sizeof(...), ...);
966
967 Both forms check for overflow on the allocation size n * sizeof(...),
968 and return NULL if that occurred.
969
970 These generic allocation functions all emit a stack dump on failure when used
971 without __GFP_NOWARN so there is no use in emitting an additional failure
972 message when NULL is returned.
973
974 15) The inline disease
975 ----------------------
976
977 There appears to be a common misperception that gcc has a magic "make me
978 faster" speedup option called ``inline``. While the use of inlines can be
979 appropriate (for example as a means of replacing macros, see Chapter 12), it
980 very often is not. Abundant use of the inline keyword leads to a much bigger
981 kernel, which in turn slows the system as a whole down, due to a bigger
982 icache footprint for the CPU and simply because there is less memory
983 available for the pagecache. Just think about it; a pagecache miss causes a
984 disk seek, which easily takes 5 milliseconds. There are a LOT of cpu cycles
985 that can go into these 5 milliseconds.
986
987 A reasonable rule of thumb is to not put inline at functions that have more
988 than 3 lines of code in them. An exception to this rule are the cases where
989 a parameter is known to be a compiletime constant, and as a result of this
990 constantness you *know* the compiler will be able to optimize most of your
991 function away at compile time. For a good example of this later case, see
992 the kmalloc() inline function.
993
994 Often people argue that adding inline to functions that are static and used
995 only once is always a win since there is no space tradeoff. While this is
996 technically correct, gcc is capable of inlining these automatically without
997 help, and the maintenance issue of removing the inline when a second user
998 appears outweighs the potential value of the hint that tells gcc to do
999 something it would have done anyway.
1000
1001
1002 16) Function return values and names
1003 ------------------------------------
1004
1005 Functions can return values of many different kinds, and one of the
1006 most common is a value indicating whether the function succeeded or
1007 failed. Such a value can be represented as an error-code integer
1008 (-Exxx = failure, 0 = success) or a ``succeeded`` boolean (0 = failure,
1009 non-zero = success).
1010
1011 Mixing up these two sorts of representations is a fertile source of
1012 difficult-to-find bugs. If the C language included a strong distinction
1013 between integers and booleans then the compiler would find these mistakes
1014 for us... but it doesn't. To help prevent such bugs, always follow this
1015 convention::
1016
1017 If the name of a function is an action or an imperative command,
1018 the function should return an error-code integer. If the name
1019 is a predicate, the function should return a "succeeded" boolean.
1020
1021 For example, ``add work`` is a command, and the add_work() function returns 0
1022 for success or -EBUSY for failure. In the same way, ``PCI device present`` is
1023 a predicate, and the pci_dev_present() function returns 1 if it succeeds in
1024 finding a matching device or 0 if it doesn't.
1025
1026 All EXPORTed functions must respect this convention, and so should all
1027 public functions. Private (static) functions need not, but it is
1028 recommended that they do.
1029
1030 Functions whose return value is the actual result of a computation, rather
1031 than an indication of whether the computation succeeded, are not subject to
1032 this rule. Generally they indicate failure by returning some out-of-range
1033 result. Typical examples would be functions that return pointers; they use
1034 NULL or the ERR_PTR mechanism to report failure.
1035
1036
1037 17) Using bool
1038 --------------
1039
1040 The Linux kernel bool type is an alias for the C99 _Bool type. bool values can
1041 only evaluate to 0 or 1, and implicit or explicit conversion to bool
1042 automatically converts the value to true or false. When using bool types the
1043 !! construction is not needed, which eliminates a class of bugs.
1044
1045 When working with bool values the true and false definitions should be used
1046 instead of 1 and 0.
1047
1048 bool function return types and stack variables are always fine to use whenever
1049 appropriate. Use of bool is encouraged to improve readability and is often a
1050 better option than 'int' for storing boolean values.
1051
1052 Do not use bool if cache line layout or size of the value matters, as its size
1053 and alignment varies based on the compiled architecture. Structures that are
1054 optimized for alignment and size should not use bool.
1055
1056 If a structure has many true/false values, consider consolidating them into a
1057 bitfield with 1 bit members, or using an appropriate fixed width type, such as
1058 u8.
1059
1060 Similarly for function arguments, many true/false values can be consolidated
1061 into a single bitwise 'flags' argument and 'flags' can often be a more
1062 readable alternative if the call-sites have naked true/false constants.
1063
1064 Otherwise limited use of bool in structures and arguments can improve
1065 readability.
1066
1067 18) Don't re-invent the kernel macros
1068 -------------------------------------
1069
1070 The header file include/linux/kernel.h contains a number of macros that
1071 you should use, rather than explicitly coding some variant of them yourself.
1072 For example, if you need to calculate the length of an array, take advantage
1073 of the macro
1074
1075 .. code-block:: c
1076
1077 #define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
1078
1079 Similarly, if you need to calculate the size of some structure member, use
1080
1081 .. code-block:: c
1082
1083 #define sizeof_field(t, f) (sizeof(((t*)0)->f))
1084
1085 There are also min() and max() macros that do strict type checking if you
1086 need them. Feel free to peruse that header file to see what else is already
1087 defined that you shouldn't reproduce in your code.
1088
1089
1090 19) Editor modelines and other cruft
1091 ------------------------------------
1092
1093 Some editors can interpret configuration information embedded in source files,
1094 indicated with special markers. For example, emacs interprets lines marked
1095 like this:
1096
1097 .. code-block:: c
1098
1099 -*- mode: c -*-
1100
1101 Or like this:
1102
1103 .. code-block:: c
1104
1105 /*
1106 Local Variables:
1107 compile-command: "gcc -DMAGIC_DEBUG_FLAG foo.c"
1108 End:
1109 */
1110
1111 Vim interprets markers that look like this:
1112
1113 .. code-block:: c
1114
1115 /* vim:set sw=8 noet */
1116
1117 Do not include any of these in source files. People have their own personal
1118 editor configurations, and your source files should not override them. This
1119 includes markers for indentation and mode configuration. People may use their
1120 own custom mode, or may have some other magic method for making indentation
1121 work correctly.
1122
1123
1124 20) Inline assembly
1125 -------------------
1126
1127 In architecture-specific code, you may need to use inline assembly to interface
1128 with CPU or platform functionality. Don't hesitate to do so when necessary.
1129 However, don't use inline assembly gratuitously when C can do the job. You can
1130 and should poke hardware from C when possible.
1131
1132 Consider writing simple helper functions that wrap common bits of inline
1133 assembly, rather than repeatedly writing them with slight variations. Remember
1134 that inline assembly can use C parameters.
1135
1136 Large, non-trivial assembly functions should go in .S files, with corresponding
1137 C prototypes defined in C header files. The C prototypes for assembly
1138 functions should use ``asmlinkage``.
1139
1140 You may need to mark your asm statement as volatile, to prevent GCC from
1141 removing it if GCC doesn't notice any side effects. You don't always need to
1142 do so, though, and doing so unnecessarily can limit optimization.
1143
1144 When writing a single inline assembly statement containing multiple
1145 instructions, put each instruction on a separate line in a separate quoted
1146 string, and end each string except the last with ``\n\t`` to properly indent
1147 the next instruction in the assembly output:
1148
1149 .. code-block:: c
1150
1151 asm ("magic %reg1, #42\n\t"
1152 "more_magic %reg2, %reg3"
1153 : /* outputs */ : /* inputs */ : /* clobbers */);
1154
1155
1156 21) Conditional Compilation
1157 ---------------------------
1158
1159 Wherever possible, don't use preprocessor conditionals (#if, #ifdef) in .c
1160 files; doing so makes code harder to read and logic harder to follow. Instead,
1161 use such conditionals in a header file defining functions for use in those .c
1162 files, providing no-op stub versions in the #else case, and then call those
1163 functions unconditionally from .c files. The compiler will avoid generating
1164 any code for the stub calls, producing identical results, but the logic will
1165 remain easy to follow.
1166
1167 Prefer to compile out entire functions, rather than portions of functions or
1168 portions of expressions. Rather than putting an ifdef in an expression, factor
1169 out part or all of the expression into a separate helper function and apply the
1170 conditional to that function.
1171
1172 If you have a function or variable which may potentially go unused in a
1173 particular configuration, and the compiler would warn about its definition
1174 going unused, mark the definition as __maybe_unused rather than wrapping it in
1175 a preprocessor conditional. (However, if a function or variable *always* goes
1176 unused, delete it.)
1177
1178 Within code, where possible, use the IS_ENABLED macro to convert a Kconfig
1179 symbol into a C boolean expression, and use it in a normal C conditional:
1180
1181 .. code-block:: c
1182
1183 if (IS_ENABLED(CONFIG_SOMETHING)) {
1184 ...
1185 }
1186
1187 The compiler will constant-fold the conditional away, and include or exclude
1188 the block of code just as with an #ifdef, so this will not add any runtime
1189 overhead. However, this approach still allows the C compiler to see the code
1190 inside the block, and check it for correctness (syntax, types, symbol
1191 references, etc). Thus, you still have to use an #ifdef if the code inside the
1192 block references symbols that will not exist if the condition is not met.
1193
1194 At the end of any non-trivial #if or #ifdef block (more than a few lines),
1195 place a comment after the #endif on the same line, noting the conditional
1196 expression used. For instance:
1197
1198 .. code-block:: c
1199
1200 #ifdef CONFIG_SOMETHING
1201 ...
1202 #endif /* CONFIG_SOMETHING */
1203
1204
1205 22) Do not crash the kernel
1206 ---------------------------
1207
1208 In general, the decision to crash the kernel belongs to the user, rather
1209 than to the kernel developer.
1210
1211 Avoid panic()
1212 *************
1213
1214 panic() should be used with care and primarily only during system boot.
1215 panic() is, for example, acceptable when running out of memory during boot and
1216 not being able to continue.
1217
1218 Use WARN() rather than BUG()
1219 ****************************
1220
1221 Do not add new code that uses any of the BUG() variants, such as BUG(),
1222 BUG_ON(), or VM_BUG_ON(). Instead, use a WARN*() variant, preferably
1223 WARN_ON_ONCE(), and possibly with recovery code. Recovery code is not
1224 required if there is no reasonable way to at least partially recover.
1225
1226 "I'm too lazy to do error handling" is not an excuse for using BUG(). Major
1227 internal corruptions with no way of continuing may still use BUG(), but need
1228 good justification.
1229
1230 Use WARN_ON_ONCE() rather than WARN() or WARN_ON()
1231 **************************************************
1232
1233 WARN_ON_ONCE() is generally preferred over WARN() or WARN_ON(), because it
1234 is common for a given warning condition, if it occurs at all, to occur
1235 multiple times. This can fill up and wrap the kernel log, and can even slow
1236 the system enough that the excessive logging turns into its own, additional
1237 problem.
1238
1239 Do not WARN lightly
1240 *******************
1241
1242 WARN*() is intended for unexpected, this-should-never-happen situations.
1243 WARN*() macros are not to be used for anything that is expected to happen
1244 during normal operation. These are not pre- or post-condition asserts, for
1245 example. Again: WARN*() must not be used for a condition that is expected
1246 to trigger easily, for example, by user space actions. pr_warn_once() is a
1247 possible alternative, if you need to notify the user of a problem.
1248
1249 Do not worry about panic_on_warn users
1250 **************************************
1251
1252 A few more words about panic_on_warn: Remember that ``panic_on_warn`` is an
1253 available kernel option, and that many users set this option. This is why
1254 there is a "Do not WARN lightly" writeup, above. However, the existence of
1255 panic_on_warn users is not a valid reason to avoid the judicious use
1256 WARN*(). That is because, whoever enables panic_on_warn has explicitly
1257 asked the kernel to crash if a WARN*() fires, and such users must be
1258 prepared to deal with the consequences of a system that is somewhat more
1259 likely to crash.
1260
1261 Use BUILD_BUG_ON() for compile-time assertions
1262 **********************************************
1263
1264 The use of BUILD_BUG_ON() is acceptable and encouraged, because it is a
1265 compile-time assertion that has no effect at runtime.
1266
1267 Appendix I) References
1268 ----------------------
1269
1270 The C Programming Language, Second Edition
1271 by Brian W. Kernighan and Dennis M. Ritchie.
1272 Prentice Hall, Inc., 1988.
1273 ISBN 0-13-110362-8 (paperback), 0-13-110370-9 (hardback).
1274
1275 The Practice of Programming
1276 by Brian W. Kernighan and Rob Pike.
1277 Addison-Wesley, Inc., 1999.
1278 ISBN 0-201-61586-X.
1279
1280 GNU manuals - where in compliance with K&R and this text - for cpp, gcc,
1281 gcc internals and indent, all available from https://www.gnu.org/manual/
1282
1283 WG14 is the international standardization working group for the programming
1284 language C, URL: http://www.open-std.org/JTC1/SC22/WG14/
1285
1286 Kernel CodingStyle, by greg@kroah.com at OLS 2002:
1287 http://www.kroah.com/linux/talks/ols_2002_kernel_codingstyle_talk/html/
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.