Loading...
1.. _codingstyle:
2
3Linux kernel coding style
4=========================
5
6This is a short document describing the preferred coding style for the
7linux kernel. Coding style is very personal, and I won't **force** my
8views on anybody, but this is what goes for anything that I have to be
9able to maintain, and I'd prefer it for most other things too. Please
10at least consider the points made here.
11
12First off, I'd suggest printing out a copy of the GNU coding standards,
13and NOT read it. Burn them, it's a great symbolic gesture.
14
15Anyway, here goes:
16
17
181) Indentation
19--------------
20
21Tabs are 8 characters, and thus indentations are also 8 characters.
22There are heretic movements that try to make indentations 4 (or even 2!)
23characters deep, and that is akin to trying to define the value of PI to
24be 3.
25
26Rationale: The whole idea behind indentation is to clearly define where
27a block of control starts and ends. Especially when you've been looking
28at your screen for 20 straight hours, you'll find it a lot easier to see
29how the indentation works if you have large indentations.
30
31Now, some people will claim that having 8-character indentations makes
32the code move too far to the right, and makes it hard to read on a
3380-character terminal screen. The answer to that is that if you need
34more than 3 levels of indentation, you're screwed anyway, and should fix
35your program.
36
37In short, 8-char indents make things easier to read, and have the added
38benefit of warning you when you're nesting your functions too deep.
39Heed that warning.
40
41The preferred way to ease multiple indentation levels in a switch statement is
42to align the ``switch`` and its subordinate ``case`` labels in the same column
43instead 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
64Don't put multiple statements on a single line unless you have
65something to hide:
66
67.. code-block:: c
68
69 if (condition) do_this;
70 do_something_everytime;
71
72Don't put multiple assignments on a single line either. Kernel coding style
73is super simple. Avoid tricky expressions.
74
75Outside of comments, documentation and except in Kconfig, spaces are never
76used for indentation, and the above example is deliberately broken.
77
78Get a decent editor and don't leave whitespace at the end of lines.
79
80
812) Breaking long lines and strings
82----------------------------------
83
84Coding style is all about readability and maintainability using commonly
85available tools.
86
87The limit on the length of lines is 80 columns and this is a strongly
88preferred limit.
89
90Statements longer than 80 columns will be broken into sensible chunks, unless
91exceeding 80 columns significantly increases readability and does not hide
92information. Descendants are always substantially shorter than the parent and
93are placed substantially to the right. The same applies to function headers
94with a long argument list. However, never break user-visible strings such as
95printk messages, because that breaks the ability to grep for them.
96
97
983) Placing Braces and Spaces
99----------------------------
100
101The other issue that always comes up in C styling is the placement of
102braces. Unlike the indent size, there are few technical reasons to
103choose one placement strategy over the other, but the preferred way, as
104shown to us by the prophets Kernighan and Ritchie, is to put the opening
105brace last on the line, and put the closing brace first, thusly:
106
107.. code-block:: c
108
109 if (x is true) {
110 we do y
111 }
112
113This applies to all non-function statement blocks (if, switch, for,
114while, do). E.g.:
115
116.. code-block:: c
117
118 switch (action) {
119 case KOBJ_ADD:
120 return "add";
121 case KOBJ_REMOVE:
122 return "remove";
123 case KOBJ_CHANGE:
124 return "change";
125 default:
126 return NULL;
127 }
128
129However, there is one special case, namely functions: they have the
130opening brace at the beginning of the next line, thus:
131
132.. code-block:: c
133
134 int function(int x)
135 {
136 body of function
137 }
138
139Heretic people all over the world have claimed that this inconsistency
140is ... well ... inconsistent, but all right-thinking people know that
141(a) K&R are **right** and (b) K&R are right. Besides, functions are
142special anyway (you can't nest them in C).
143
144Note that the closing brace is empty on a line of its own, **except** in
145the cases where it is followed by a continuation of the same statement,
146ie a ``while`` in a do-statement or an ``else`` in an if-statement, like
147this:
148
149.. code-block:: c
150
151 do {
152 body of do-loop
153 } while (condition);
154
155and
156
157.. code-block:: c
158
159 if (x == y) {
160 ..
161 } else if (x > y) {
162 ...
163 } else {
164 ....
165 }
166
167Rationale: K&R.
168
169Also, note that this brace-placement also minimizes the number of empty
170(or almost empty) lines, without any loss of readability. Thus, as the
171supply of new-lines on your screen is not a renewable resource (think
17225-line terminal screens here), you have more empty lines to put
173comments on.
174
175Do not unnecessarily use braces where a single statement will do.
176
177.. code-block:: c
178
179 if (condition)
180 action();
181
182and
183
184.. code-block:: none
185
186 if (condition)
187 do_this();
188 else
189 do_that();
190
191This does not apply if only one branch of a conditional statement is a single
192statement; in the latter case use braces in both branches:
193
194.. code-block:: c
195
196 if (condition) {
197 do_this();
198 do_that();
199 } else {
200 otherwise();
201 }
202
203Also, use braces when a loop contains more than a single simple statement:
204
205.. code-block:: c
206
207 while (condition) {
208 if (test)
209 do_something();
210 }
211
2123.1) Spaces
213***********
214
215Linux kernel style for use of spaces depends (mostly) on
216function-versus-keyword usage. Use a space after (most) keywords. The
217notable exceptions are sizeof, typeof, alignof, and __attribute__, which look
218somewhat like functions (and are usually used with parentheses in Linux,
219although they are not required in the language, as in: ``sizeof info`` after
220``struct fileinfo info;`` is declared).
221
222So use a space after these keywords::
223
224 if, switch, case, for, do, while
225
226but not with sizeof, typeof, alignof, or __attribute__. E.g.,
227
228.. code-block:: c
229
230
231 s = sizeof(struct file);
232
233Do not add spaces around (inside) parenthesized expressions. This example is
234**bad**:
235
236.. code-block:: c
237
238
239 s = sizeof( struct file );
240
241When declaring pointer data or a function that returns a pointer type, the
242preferred use of ``*`` is adjacent to the data name or function name and not
243adjacent to the type name. Examples:
244
245.. code-block:: c
246
247
248 char *linux_banner;
249 unsigned long long memparse(char *ptr, char **retptr);
250 char *match_strdup(substring_t *s);
251
252Use one space around (on each side of) most binary and ternary operators,
253such as any of these::
254
255 = + - < > * / % | & ^ <= >= == != ? :
256
257but no space after unary operators::
258
259 & * + - ~ ! sizeof typeof alignof __attribute__ defined
260
261no space before the postfix increment & decrement unary operators::
262
263 ++ --
264
265no space after the prefix increment & decrement unary operators::
266
267 ++ --
268
269and no space around the ``.`` and ``->`` structure member operators.
270
271Do not leave trailing whitespace at the ends of lines. Some editors with
272``smart`` indentation will insert whitespace at the beginning of new lines as
273appropriate, so you can start typing the next line of code right away.
274However, some such editors do not remove the whitespace if you end up not
275putting a line of code there, such as if you leave a blank line. As a result,
276you end up with lines containing trailing whitespace.
277
278Git will warn you about patches that introduce trailing whitespace, and can
279optionally strip the trailing whitespace for you; however, if applying a series
280of patches, this may make later patches in the series fail by changing their
281context lines.
282
283
2844) Naming
285---------
286
287C is a Spartan language, and so should your naming be. Unlike Modula-2
288and Pascal programmers, C programmers do not use cute names like
289ThisVariableIsATemporaryCounter. A C programmer would call that
290variable ``tmp``, which is much easier to write, and not the least more
291difficult to understand.
292
293HOWEVER, while mixed-case names are frowned upon, descriptive names for
294global variables are a must. To call a global function ``foo`` is a
295shooting offense.
296
297GLOBAL variables (to be used only if you **really** need them) need to
298have descriptive names, as do global functions. If you have a function
299that counts the number of active users, you should call that
300``count_active_users()`` or similar, you should **not** call it ``cntusr()``.
301
302Encoding the type of a function into the name (so-called Hungarian
303notation) is brain damaged - the compiler knows the types anyway and can
304check those, and it only confuses the programmer. No wonder MicroSoft
305makes buggy programs.
306
307LOCAL variable names should be short, and to the point. If you have
308some random integer loop counter, it should probably be called ``i``.
309Calling it ``loop_counter`` is non-productive, if there is no chance of it
310being mis-understood. Similarly, ``tmp`` can be just about any type of
311variable that is used to hold a temporary value.
312
313If you are afraid to mix up your local variable names, you have another
314problem, which is called the function-growth-hormone-imbalance syndrome.
315See chapter 6 (Functions).
316
317
3185) Typedefs
319-----------
320
321Please don't use things like ``vps_t``.
322It's a **mistake** to use typedef for structures and pointers. When you see a
323
324.. code-block:: c
325
326
327 vps_t a;
328
329in the source, what does it mean?
330In contrast, if it says
331
332.. code-block:: c
333
334 struct virtual_container *a;
335
336you can actually tell what ``a`` is.
337
338Lots of people think that typedefs ``help readability``. Not so. They are
339useful only for:
340
341 (a) totally opaque objects (where the typedef is actively used to **hide**
342 what the object is).
343
344 Example: ``pte_t`` etc. opaque objects that you can only access using
345 the proper accessor functions.
346
347 .. note::
348
349 Opaqueness and ``accessor functions`` are not good in themselves.
350 The reason we have them for things like pte_t etc. is that there
351 really is absolutely **zero** portably accessible information there.
352
353 (b) Clear integer types, where the abstraction **helps** avoid confusion
354 whether it is ``int`` or ``long``.
355
356 u8/u16/u32 are perfectly fine typedefs, although they fit into
357 category (d) better than here.
358
359 .. note::
360
361 Again - there needs to be a **reason** for this. If something is
362 ``unsigned long``, then there's no reason to do
363
364 typedef unsigned long myflags_t;
365
366 but if there is a clear reason for why it under certain circumstances
367 might be an ``unsigned int`` and under other configurations might be
368 ``unsigned long``, then by all means go ahead and use a typedef.
369
370 (c) when you use sparse to literally create a **new** type for
371 type-checking.
372
373 (d) New types which are identical to standard C99 types, in certain
374 exceptional circumstances.
375
376 Although it would only take a short amount of time for the eyes and
377 brain to become accustomed to the standard types like ``uint32_t``,
378 some people object to their use anyway.
379
380 Therefore, the Linux-specific ``u8/u16/u32/u64`` types and their
381 signed equivalents which are identical to standard types are
382 permitted -- although they are not mandatory in new code of your
383 own.
384
385 When editing existing code which already uses one or the other set
386 of types, you should conform to the existing choices in that code.
387
388 (e) Types safe for use in userspace.
389
390 In certain structures which are visible to userspace, we cannot
391 require C99 types and cannot use the ``u32`` form above. Thus, we
392 use __u32 and similar types in all structures which are shared
393 with userspace.
394
395Maybe there are other cases too, but the rule should basically be to NEVER
396EVER use a typedef unless you can clearly match one of those rules.
397
398In general, a pointer, or a struct that has elements that can reasonably
399be directly accessed should **never** be a typedef.
400
401
4026) Functions
403------------
404
405Functions should be short and sweet, and do just one thing. They should
406fit on one or two screenfuls of text (the ISO/ANSI screen size is 80x24,
407as we all know), and do one thing and do that well.
408
409The maximum length of a function is inversely proportional to the
410complexity and indentation level of that function. So, if you have a
411conceptually simple function that is just one long (but simple)
412case-statement, where you have to do lots of small things for a lot of
413different cases, it's OK to have a longer function.
414
415However, if you have a complex function, and you suspect that a
416less-than-gifted first-year high-school student might not even
417understand what the function is all about, you should adhere to the
418maximum limits all the more closely. Use helper functions with
419descriptive names (you can ask the compiler to in-line them if you think
420it's performance-critical, and it will probably do a better job of it
421than you would have done).
422
423Another measure of the function is the number of local variables. They
424shouldn't exceed 5-10, or you're doing something wrong. Re-think the
425function, and split it into smaller pieces. A human brain can
426generally easily keep track of about 7 different things, anything more
427and it gets confused. You know you're brilliant, but maybe you'd like
428to understand what you did 2 weeks from now.
429
430In source files, separate functions with one blank line. If the function is
431exported, the **EXPORT** macro for it should follow immediately after the
432closing function brace line. E.g.:
433
434.. code-block:: c
435
436 int system_is_up(void)
437 {
438 return system_state == SYSTEM_RUNNING;
439 }
440 EXPORT_SYMBOL(system_is_up);
441
442In function prototypes, include parameter names with their data types.
443Although this is not required by the C language, it is preferred in Linux
444because it is a simple way to add valuable information for the reader.
445
446Do not use the ``extern`` keyword with function prototypes as this makes
447lines longer and isn't strictly necessary.
448
449
4507) Centralized exiting of functions
451-----------------------------------
452
453Albeit deprecated by some people, the equivalent of the goto statement is
454used frequently by compilers in form of the unconditional jump instruction.
455
456The goto statement comes in handy when a function exits from multiple
457locations and some common work such as cleanup has to be done. If there is no
458cleanup needed then just return directly.
459
460Choose label names which say what the goto does or why the goto exists. An
461example of a good name could be ``out_free_buffer:`` if the goto frees ``buffer``.
462Avoid using GW-BASIC names like ``err1:`` and ``err2:``, as you would have to
463renumber them if you ever add or remove exit paths, and they make correctness
464difficult to verify anyway.
465
466The rationale for using gotos is:
467
468- unconditional statements are easier to understand and follow
469- nesting is reduced
470- errors by not updating individual exit points when making
471 modifications are prevented
472- saves the compiler work to optimize redundant code away ;)
473
474.. code-block:: c
475
476 int fun(int a)
477 {
478 int result = 0;
479 char *buffer;
480
481 buffer = kmalloc(SIZE, GFP_KERNEL);
482 if (!buffer)
483 return -ENOMEM;
484
485 if (condition1) {
486 while (loop1) {
487 ...
488 }
489 result = 1;
490 goto out_free_buffer;
491 }
492 ...
493 out_free_buffer:
494 kfree(buffer);
495 return result;
496 }
497
498A common type of bug to be aware of is ``one err bugs`` which look like this:
499
500.. code-block:: c
501
502 err:
503 kfree(foo->bar);
504 kfree(foo);
505 return ret;
506
507The bug in this code is that on some exit paths ``foo`` is NULL. Normally the
508fix for this is to split it up into two error labels ``err_free_bar:`` and
509``err_free_foo:``:
510
511.. code-block:: c
512
513 err_free_bar:
514 kfree(foo->bar);
515 err_free_foo:
516 kfree(foo);
517 return ret;
518
519Ideally you should simulate errors to test all exit paths.
520
521
5228) Commenting
523-------------
524
525Comments are good, but there is also a danger of over-commenting. NEVER
526try to explain HOW your code works in a comment: it's much better to
527write the code so that the **working** is obvious, and it's a waste of
528time to explain badly written code.
529
530Generally, you want your comments to tell WHAT your code does, not HOW.
531Also, try to avoid putting comments inside a function body: if the
532function is so complex that you need to separately comment parts of it,
533you should probably go back to chapter 6 for a while. You can make
534small comments to note or warn about something particularly clever (or
535ugly), but try to avoid excess. Instead, put the comments at the head
536of the function, telling people what it does, and possibly WHY it does
537it.
538
539When commenting the kernel API functions, please use the kernel-doc format.
540See the files at :ref:`Documentation/doc-guide/ <doc_guide>` and
541``scripts/kernel-doc`` for details.
542
543The preferred style for long (multi-line) comments is:
544
545.. code-block:: c
546
547 /*
548 * This is the preferred style for multi-line
549 * comments in the Linux kernel source code.
550 * Please use it consistently.
551 *
552 * Description: A column of asterisks on the left side,
553 * with beginning and ending almost-blank lines.
554 */
555
556For files in net/ and drivers/net/ the preferred style for long (multi-line)
557comments is a little different.
558
559.. code-block:: c
560
561 /* The preferred comment style for files in net/ and drivers/net
562 * looks like this.
563 *
564 * It is nearly the same as the generally preferred comment style,
565 * but there is no initial almost-blank line.
566 */
567
568It's also important to comment data, whether they are basic types or derived
569types. To this end, use just one data declaration per line (no commas for
570multiple data declarations). This leaves you room for a small comment on each
571item, explaining its use.
572
573
5749) You've made a mess of it
575---------------------------
576
577That's OK, we all do. You've probably been told by your long-time Unix
578user helper that ``GNU emacs`` automatically formats the C sources for
579you, and you've noticed that yes, it does do that, but the defaults it
580uses are less than desirable (in fact, they are worse than random
581typing - an infinite number of monkeys typing into GNU emacs would never
582make a good program).
583
584So, you can either get rid of GNU emacs, or change it to use saner
585values. To do the latter, you can stick the following in your .emacs file:
586
587.. code-block:: none
588
589 (defun c-lineup-arglist-tabs-only (ignored)
590 "Line up argument lists by tabs, not spaces"
591 (let* ((anchor (c-langelem-pos c-syntactic-element))
592 (column (c-langelem-2nd-pos c-syntactic-element))
593 (offset (- (1+ column) anchor))
594 (steps (floor offset c-basic-offset)))
595 (* (max steps 1)
596 c-basic-offset)))
597
598 (dir-locals-set-class-variables
599 'linux-kernel
600 '((c-mode . (
601 (c-basic-offset . 8)
602 (c-label-minimum-indentation . 0)
603 (c-offsets-alist . (
604 (arglist-close . c-lineup-arglist-tabs-only)
605 (arglist-cont-nonempty .
606 (c-lineup-gcc-asm-reg c-lineup-arglist-tabs-only))
607 (arglist-intro . +)
608 (brace-list-intro . +)
609 (c . c-lineup-C-comments)
610 (case-label . 0)
611 (comment-intro . c-lineup-comment)
612 (cpp-define-intro . +)
613 (cpp-macro . -1000)
614 (cpp-macro-cont . +)
615 (defun-block-intro . +)
616 (else-clause . 0)
617 (func-decl-cont . +)
618 (inclass . +)
619 (inher-cont . c-lineup-multi-inher)
620 (knr-argdecl-intro . 0)
621 (label . -1000)
622 (statement . 0)
623 (statement-block-intro . +)
624 (statement-case-intro . +)
625 (statement-cont . +)
626 (substatement . +)
627 ))
628 (indent-tabs-mode . t)
629 (show-trailing-whitespace . t)
630 ))))
631
632 (dir-locals-set-directory-class
633 (expand-file-name "~/src/linux-trees")
634 'linux-kernel)
635
636This will make emacs go better with the kernel coding style for C
637files below ``~/src/linux-trees``.
638
639But even if you fail in getting emacs to do sane formatting, not
640everything is lost: use ``indent``.
641
642Now, again, GNU indent has the same brain-dead settings that GNU emacs
643has, which is why you need to give it a few command line options.
644However, that's not too bad, because even the makers of GNU indent
645recognize the authority of K&R (the GNU people aren't evil, they are
646just severely misguided in this matter), so you just give indent the
647options ``-kr -i8`` (stands for ``K&R, 8 character indents``), or use
648``scripts/Lindent``, which indents in the latest style.
649
650``indent`` has a lot of options, and especially when it comes to comment
651re-formatting you may want to take a look at the man page. But
652remember: ``indent`` is not a fix for bad programming.
653
654Note that you can also use the ``clang-format`` tool to help you with
655these rules, to quickly re-format parts of your code automatically,
656and to review full files in order to spot coding style mistakes,
657typos and possible improvements. It is also handy for sorting ``#includes``,
658for aligning variables/macros, for reflowing text and other similar tasks.
659See the file :ref:`Documentation/process/clang-format.rst <clangformat>`
660for more details.
661
662
66310) Kconfig configuration files
664-------------------------------
665
666For all of the Kconfig* configuration files throughout the source tree,
667the indentation is somewhat different. Lines under a ``config`` definition
668are indented with one tab, while help text is indented an additional two
669spaces. Example::
670
671 config AUDIT
672 bool "Auditing support"
673 depends on NET
674 help
675 Enable auditing infrastructure that can be used with another
676 kernel subsystem, such as SELinux (which requires this for
677 logging of avc messages output). Does not do system-call
678 auditing without CONFIG_AUDITSYSCALL.
679
680Seriously dangerous features (such as write support for certain
681filesystems) should advertise this prominently in their prompt string::
682
683 config ADFS_FS_RW
684 bool "ADFS write support (DANGEROUS)"
685 depends on ADFS_FS
686 ...
687
688For full documentation on the configuration files, see the file
689Documentation/kbuild/kconfig-language.rst.
690
691
69211) Data structures
693-------------------
694
695Data structures that have visibility outside the single-threaded
696environment they are created and destroyed in should always have
697reference counts. In the kernel, garbage collection doesn't exist (and
698outside the kernel garbage collection is slow and inefficient), which
699means that you absolutely **have** to reference count all your uses.
700
701Reference counting means that you can avoid locking, and allows multiple
702users to have access to the data structure in parallel - and not having
703to worry about the structure suddenly going away from under them just
704because they slept or did something else for a while.
705
706Note that locking is **not** a replacement for reference counting.
707Locking is used to keep data structures coherent, while reference
708counting is a memory management technique. Usually both are needed, and
709they are not to be confused with each other.
710
711Many data structures can indeed have two levels of reference counting,
712when there are users of different ``classes``. The subclass count counts
713the number of subclass users, and decrements the global count just once
714when the subclass count goes to zero.
715
716Examples of this kind of ``multi-level-reference-counting`` can be found in
717memory management (``struct mm_struct``: mm_users and mm_count), and in
718filesystem code (``struct super_block``: s_count and s_active).
719
720Remember: if another thread can find your data structure, and you don't
721have a reference count on it, you almost certainly have a bug.
722
723
72412) Macros, Enums and RTL
725-------------------------
726
727Names of macros defining constants and labels in enums are capitalized.
728
729.. code-block:: c
730
731 #define CONSTANT 0x12345
732
733Enums are preferred when defining several related constants.
734
735CAPITALIZED macro names are appreciated but macros resembling functions
736may be named in lower case.
737
738Generally, inline functions are preferable to macros resembling functions.
739
740Macros with multiple statements should be enclosed in a do - while block:
741
742.. code-block:: c
743
744 #define macrofun(a, b, c) \
745 do { \
746 if (a == 5) \
747 do_this(b, c); \
748 } while (0)
749
750Things to avoid when using macros:
751
7521) macros that affect control flow:
753
754.. code-block:: c
755
756 #define FOO(x) \
757 do { \
758 if (blah(x) < 0) \
759 return -EBUGGERED; \
760 } while (0)
761
762is a **very** bad idea. It looks like a function call but exits the ``calling``
763function; don't break the internal parsers of those who will read the code.
764
7652) macros that depend on having a local variable with a magic name:
766
767.. code-block:: c
768
769 #define FOO(val) bar(index, val)
770
771might look like a good thing, but it's confusing as hell when one reads the
772code and it's prone to breakage from seemingly innocent changes.
773
7743) macros with arguments that are used as l-values: FOO(x) = y; will
775bite you if somebody e.g. turns FOO into an inline function.
776
7774) forgetting about precedence: macros defining constants using expressions
778must enclose the expression in parentheses. Beware of similar issues with
779macros using parameters.
780
781.. code-block:: c
782
783 #define CONSTANT 0x4000
784 #define CONSTEXP (CONSTANT | 3)
785
7865) namespace collisions when defining local variables in macros resembling
787functions:
788
789.. code-block:: c
790
791 #define FOO(x) \
792 ({ \
793 typeof(x) ret; \
794 ret = calc_ret(x); \
795 (ret); \
796 })
797
798ret is a common name for a local variable - __foo_ret is less likely
799to collide with an existing variable.
800
801The cpp manual deals with macros exhaustively. The gcc internals manual also
802covers RTL which is used frequently with assembly language in the kernel.
803
804
80513) Printing kernel messages
806----------------------------
807
808Kernel developers like to be seen as literate. Do mind the spelling
809of kernel messages to make a good impression. Do not use crippled
810words like ``dont``; use ``do not`` or ``don't`` instead. Make the messages
811concise, clear, and unambiguous.
812
813Kernel messages do not have to be terminated with a period.
814
815Printing numbers in parentheses (%d) adds no value and should be avoided.
816
817There are a number of driver model diagnostic macros in <linux/device.h>
818which you should use to make sure messages are matched to the right device
819and driver, and are tagged with the right level: dev_err(), dev_warn(),
820dev_info(), and so forth. For messages that aren't associated with a
821particular device, <linux/printk.h> defines pr_notice(), pr_info(),
822pr_warn(), pr_err(), etc.
823
824Coming up with good debugging messages can be quite a challenge; and once
825you have them, they can be a huge help for remote troubleshooting. However
826debug message printing is handled differently than printing other non-debug
827messages. While the other pr_XXX() functions print unconditionally,
828pr_debug() does not; it is compiled out by default, unless either DEBUG is
829defined or CONFIG_DYNAMIC_DEBUG is set. That is true for dev_dbg() also,
830and a related convention uses VERBOSE_DEBUG to add dev_vdbg() messages to
831the ones already enabled by DEBUG.
832
833Many subsystems have Kconfig debug options to turn on -DDEBUG in the
834corresponding Makefile; in other cases specific files #define DEBUG. And
835when a debug message should be unconditionally printed, such as if it is
836already inside a debug-related #ifdef section, printk(KERN_DEBUG ...) can be
837used.
838
839
84014) Allocating memory
841---------------------
842
843The kernel provides the following general purpose memory allocators:
844kmalloc(), kzalloc(), kmalloc_array(), kcalloc(), vmalloc(), and
845vzalloc(). Please refer to the API documentation for further information
846about them. :ref:`Documentation/core-api/memory-allocation.rst
847<memory_allocation>`
848
849The preferred form for passing a size of a struct is the following:
850
851.. code-block:: c
852
853 p = kmalloc(sizeof(*p), ...);
854
855The alternative form where struct name is spelled out hurts readability and
856introduces an opportunity for a bug when the pointer variable type is changed
857but the corresponding sizeof that is passed to a memory allocator is not.
858
859Casting the return value which is a void pointer is redundant. The conversion
860from void pointer to any other pointer type is guaranteed by the C programming
861language.
862
863The preferred form for allocating an array is the following:
864
865.. code-block:: c
866
867 p = kmalloc_array(n, sizeof(...), ...);
868
869The preferred form for allocating a zeroed array is the following:
870
871.. code-block:: c
872
873 p = kcalloc(n, sizeof(...), ...);
874
875Both forms check for overflow on the allocation size n * sizeof(...),
876and return NULL if that occurred.
877
878These generic allocation functions all emit a stack dump on failure when used
879without __GFP_NOWARN so there is no use in emitting an additional failure
880message when NULL is returned.
881
88215) The inline disease
883----------------------
884
885There appears to be a common misperception that gcc has a magic "make me
886faster" speedup option called ``inline``. While the use of inlines can be
887appropriate (for example as a means of replacing macros, see Chapter 12), it
888very often is not. Abundant use of the inline keyword leads to a much bigger
889kernel, which in turn slows the system as a whole down, due to a bigger
890icache footprint for the CPU and simply because there is less memory
891available for the pagecache. Just think about it; a pagecache miss causes a
892disk seek, which easily takes 5 milliseconds. There are a LOT of cpu cycles
893that can go into these 5 milliseconds.
894
895A reasonable rule of thumb is to not put inline at functions that have more
896than 3 lines of code in them. An exception to this rule are the cases where
897a parameter is known to be a compiletime constant, and as a result of this
898constantness you *know* the compiler will be able to optimize most of your
899function away at compile time. For a good example of this later case, see
900the kmalloc() inline function.
901
902Often people argue that adding inline to functions that are static and used
903only once is always a win since there is no space tradeoff. While this is
904technically correct, gcc is capable of inlining these automatically without
905help, and the maintenance issue of removing the inline when a second user
906appears outweighs the potential value of the hint that tells gcc to do
907something it would have done anyway.
908
909
91016) Function return values and names
911------------------------------------
912
913Functions can return values of many different kinds, and one of the
914most common is a value indicating whether the function succeeded or
915failed. Such a value can be represented as an error-code integer
916(-Exxx = failure, 0 = success) or a ``succeeded`` boolean (0 = failure,
917non-zero = success).
918
919Mixing up these two sorts of representations is a fertile source of
920difficult-to-find bugs. If the C language included a strong distinction
921between integers and booleans then the compiler would find these mistakes
922for us... but it doesn't. To help prevent such bugs, always follow this
923convention::
924
925 If the name of a function is an action or an imperative command,
926 the function should return an error-code integer. If the name
927 is a predicate, the function should return a "succeeded" boolean.
928
929For example, ``add work`` is a command, and the add_work() function returns 0
930for success or -EBUSY for failure. In the same way, ``PCI device present`` is
931a predicate, and the pci_dev_present() function returns 1 if it succeeds in
932finding a matching device or 0 if it doesn't.
933
934All EXPORTed functions must respect this convention, and so should all
935public functions. Private (static) functions need not, but it is
936recommended that they do.
937
938Functions whose return value is the actual result of a computation, rather
939than an indication of whether the computation succeeded, are not subject to
940this rule. Generally they indicate failure by returning some out-of-range
941result. Typical examples would be functions that return pointers; they use
942NULL or the ERR_PTR mechanism to report failure.
943
944
94517) Using bool
946--------------
947
948The Linux kernel bool type is an alias for the C99 _Bool type. bool values can
949only evaluate to 0 or 1, and implicit or explicit conversion to bool
950automatically converts the value to true or false. When using bool types the
951!! construction is not needed, which eliminates a class of bugs.
952
953When working with bool values the true and false definitions should be used
954instead of 1 and 0.
955
956bool function return types and stack variables are always fine to use whenever
957appropriate. Use of bool is encouraged to improve readability and is often a
958better option than 'int' for storing boolean values.
959
960Do not use bool if cache line layout or size of the value matters, as its size
961and alignment varies based on the compiled architecture. Structures that are
962optimized for alignment and size should not use bool.
963
964If a structure has many true/false values, consider consolidating them into a
965bitfield with 1 bit members, or using an appropriate fixed width type, such as
966u8.
967
968Similarly for function arguments, many true/false values can be consolidated
969into a single bitwise 'flags' argument and 'flags' can often be a more
970readable alternative if the call-sites have naked true/false constants.
971
972Otherwise limited use of bool in structures and arguments can improve
973readability.
974
97518) Don't re-invent the kernel macros
976-------------------------------------
977
978The header file include/linux/kernel.h contains a number of macros that
979you should use, rather than explicitly coding some variant of them yourself.
980For example, if you need to calculate the length of an array, take advantage
981of the macro
982
983.. code-block:: c
984
985 #define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
986
987Similarly, if you need to calculate the size of some structure member, use
988
989.. code-block:: c
990
991 #define FIELD_SIZEOF(t, f) (sizeof(((t*)0)->f))
992
993There are also min() and max() macros that do strict type checking if you
994need them. Feel free to peruse that header file to see what else is already
995defined that you shouldn't reproduce in your code.
996
997
99819) Editor modelines and other cruft
999------------------------------------
1000
1001Some editors can interpret configuration information embedded in source files,
1002indicated with special markers. For example, emacs interprets lines marked
1003like this:
1004
1005.. code-block:: c
1006
1007 -*- mode: c -*-
1008
1009Or like this:
1010
1011.. code-block:: c
1012
1013 /*
1014 Local Variables:
1015 compile-command: "gcc -DMAGIC_DEBUG_FLAG foo.c"
1016 End:
1017 */
1018
1019Vim interprets markers that look like this:
1020
1021.. code-block:: c
1022
1023 /* vim:set sw=8 noet */
1024
1025Do not include any of these in source files. People have their own personal
1026editor configurations, and your source files should not override them. This
1027includes markers for indentation and mode configuration. People may use their
1028own custom mode, or may have some other magic method for making indentation
1029work correctly.
1030
1031
103220) Inline assembly
1033-------------------
1034
1035In architecture-specific code, you may need to use inline assembly to interface
1036with CPU or platform functionality. Don't hesitate to do so when necessary.
1037However, don't use inline assembly gratuitously when C can do the job. You can
1038and should poke hardware from C when possible.
1039
1040Consider writing simple helper functions that wrap common bits of inline
1041assembly, rather than repeatedly writing them with slight variations. Remember
1042that inline assembly can use C parameters.
1043
1044Large, non-trivial assembly functions should go in .S files, with corresponding
1045C prototypes defined in C header files. The C prototypes for assembly
1046functions should use ``asmlinkage``.
1047
1048You may need to mark your asm statement as volatile, to prevent GCC from
1049removing it if GCC doesn't notice any side effects. You don't always need to
1050do so, though, and doing so unnecessarily can limit optimization.
1051
1052When writing a single inline assembly statement containing multiple
1053instructions, put each instruction on a separate line in a separate quoted
1054string, and end each string except the last with ``\n\t`` to properly indent
1055the next instruction in the assembly output:
1056
1057.. code-block:: c
1058
1059 asm ("magic %reg1, #42\n\t"
1060 "more_magic %reg2, %reg3"
1061 : /* outputs */ : /* inputs */ : /* clobbers */);
1062
1063
106421) Conditional Compilation
1065---------------------------
1066
1067Wherever possible, don't use preprocessor conditionals (#if, #ifdef) in .c
1068files; doing so makes code harder to read and logic harder to follow. Instead,
1069use such conditionals in a header file defining functions for use in those .c
1070files, providing no-op stub versions in the #else case, and then call those
1071functions unconditionally from .c files. The compiler will avoid generating
1072any code for the stub calls, producing identical results, but the logic will
1073remain easy to follow.
1074
1075Prefer to compile out entire functions, rather than portions of functions or
1076portions of expressions. Rather than putting an ifdef in an expression, factor
1077out part or all of the expression into a separate helper function and apply the
1078conditional to that function.
1079
1080If you have a function or variable which may potentially go unused in a
1081particular configuration, and the compiler would warn about its definition
1082going unused, mark the definition as __maybe_unused rather than wrapping it in
1083a preprocessor conditional. (However, if a function or variable *always* goes
1084unused, delete it.)
1085
1086Within code, where possible, use the IS_ENABLED macro to convert a Kconfig
1087symbol into a C boolean expression, and use it in a normal C conditional:
1088
1089.. code-block:: c
1090
1091 if (IS_ENABLED(CONFIG_SOMETHING)) {
1092 ...
1093 }
1094
1095The compiler will constant-fold the conditional away, and include or exclude
1096the block of code just as with an #ifdef, so this will not add any runtime
1097overhead. However, this approach still allows the C compiler to see the code
1098inside the block, and check it for correctness (syntax, types, symbol
1099references, etc). Thus, you still have to use an #ifdef if the code inside the
1100block references symbols that will not exist if the condition is not met.
1101
1102At the end of any non-trivial #if or #ifdef block (more than a few lines),
1103place a comment after the #endif on the same line, noting the conditional
1104expression used. For instance:
1105
1106.. code-block:: c
1107
1108 #ifdef CONFIG_SOMETHING
1109 ...
1110 #endif /* CONFIG_SOMETHING */
1111
1112
1113Appendix I) References
1114----------------------
1115
1116The C Programming Language, Second Edition
1117by Brian W. Kernighan and Dennis M. Ritchie.
1118Prentice Hall, Inc., 1988.
1119ISBN 0-13-110362-8 (paperback), 0-13-110370-9 (hardback).
1120
1121The Practice of Programming
1122by Brian W. Kernighan and Rob Pike.
1123Addison-Wesley, Inc., 1999.
1124ISBN 0-201-61586-X.
1125
1126GNU manuals - where in compliance with K&R and this text - for cpp, gcc,
1127gcc internals and indent, all available from http://www.gnu.org/manual/
1128
1129WG14 is the international standardization working group for the programming
1130language C, URL: http://www.open-std.org/JTC1/SC22/WG14/
1131
1132Kernel :ref:`process/coding-style.rst <codingstyle>`, by greg@kroah.com at OLS 2002:
1133http://www.kroah.com/linux/talks/ols_2002_kernel_codingstyle_talk/html/
1.. _codingstyle:
2
3Linux kernel coding style
4=========================
5
6This is a short document describing the preferred coding style for the
7linux kernel. Coding style is very personal, and I won't **force** my
8views on anybody, but this is what goes for anything that I have to be
9able to maintain, and I'd prefer it for most other things too. Please
10at least consider the points made here.
11
12First off, I'd suggest printing out a copy of the GNU coding standards,
13and NOT read it. Burn them, it's a great symbolic gesture.
14
15Anyway, here goes:
16
17
181) Indentation
19--------------
20
21Tabs are 8 characters, and thus indentations are also 8 characters.
22There are heretic movements that try to make indentations 4 (or even 2!)
23characters deep, and that is akin to trying to define the value of PI to
24be 3.
25
26Rationale: The whole idea behind indentation is to clearly define where
27a block of control starts and ends. Especially when you've been looking
28at your screen for 20 straight hours, you'll find it a lot easier to see
29how the indentation works if you have large indentations.
30
31Now, some people will claim that having 8-character indentations makes
32the code move too far to the right, and makes it hard to read on a
3380-character terminal screen. The answer to that is that if you need
34more than 3 levels of indentation, you're screwed anyway, and should fix
35your program.
36
37In short, 8-char indents make things easier to read, and have the added
38benefit of warning you when you're nesting your functions too deep.
39Heed that warning.
40
41The preferred way to ease multiple indentation levels in a switch statement is
42to align the ``switch`` and its subordinate ``case`` labels in the same column
43instead 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
64Don't put multiple statements on a single line unless you have
65something to hide:
66
67.. code-block:: c
68
69 if (condition) do_this;
70 do_something_everytime;
71
72Don't use commas to avoid using braces:
73
74.. code-block:: c
75
76 if (condition)
77 do_this(), do_that();
78
79Always uses braces for multiple statements:
80
81.. code-block:: c
82
83 if (condition) {
84 do_this();
85 do_that();
86 }
87
88Don't put multiple assignments on a single line either. Kernel coding style
89is super simple. Avoid tricky expressions.
90
91
92Outside of comments, documentation and except in Kconfig, spaces are never
93used for indentation, and the above example is deliberately broken.
94
95Get a decent editor and don't leave whitespace at the end of lines.
96
97
982) Breaking long lines and strings
99----------------------------------
100
101Coding style is all about readability and maintainability using commonly
102available tools.
103
104The preferred limit on the length of a single line is 80 columns.
105
106Statements longer than 80 columns should be broken into sensible chunks,
107unless exceeding 80 columns significantly increases readability and does
108not hide information.
109
110Descendants are always substantially shorter than the parent and
111are placed substantially to the right. A very commonly used style
112is to align descendants to a function open parenthesis.
113
114These same rules are applied to function headers with a long argument list.
115
116However, never break user-visible strings such as printk messages because
117that breaks the ability to grep for them.
118
119
1203) Placing Braces and Spaces
121----------------------------
122
123The other issue that always comes up in C styling is the placement of
124braces. Unlike the indent size, there are few technical reasons to
125choose one placement strategy over the other, but the preferred way, as
126shown to us by the prophets Kernighan and Ritchie, is to put the opening
127brace 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
135This applies to all non-function statement blocks (if, switch, for,
136while, 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
151However, there is one special case, namely functions: they have the
152opening 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
161Heretic people all over the world have claimed that this inconsistency
162is ... well ... inconsistent, but all right-thinking people know that
163(a) K&R are **right** and (b) K&R are right. Besides, functions are
164special anyway (you can't nest them in C).
165
166Note that the closing brace is empty on a line of its own, **except** in
167the cases where it is followed by a continuation of the same statement,
168ie a ``while`` in a do-statement or an ``else`` in an if-statement, like
169this:
170
171.. code-block:: c
172
173 do {
174 body of do-loop
175 } while (condition);
176
177and
178
179.. code-block:: c
180
181 if (x == y) {
182 ..
183 } else if (x > y) {
184 ...
185 } else {
186 ....
187 }
188
189Rationale: K&R.
190
191Also, note that this brace-placement also minimizes the number of empty
192(or almost empty) lines, without any loss of readability. Thus, as the
193supply of new-lines on your screen is not a renewable resource (think
19425-line terminal screens here), you have more empty lines to put
195comments on.
196
197Do not unnecessarily use braces where a single statement will do.
198
199.. code-block:: c
200
201 if (condition)
202 action();
203
204and
205
206.. code-block:: c
207
208 if (condition)
209 do_this();
210 else
211 do_that();
212
213This does not apply if only one branch of a conditional statement is a single
214statement; 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
225Also, 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
2343.1) Spaces
235***********
236
237Linux kernel style for use of spaces depends (mostly) on
238function-versus-keyword usage. Use a space after (most) keywords. The
239notable exceptions are sizeof, typeof, alignof, and __attribute__, which look
240somewhat like functions (and are usually used with parentheses in Linux,
241although they are not required in the language, as in: ``sizeof info`` after
242``struct fileinfo info;`` is declared).
243
244So use a space after these keywords::
245
246 if, switch, case, for, do, while
247
248but not with sizeof, typeof, alignof, or __attribute__. E.g.,
249
250.. code-block:: c
251
252
253 s = sizeof(struct file);
254
255Do 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
263When declaring pointer data or a function that returns a pointer type, the
264preferred use of ``*`` is adjacent to the data name or function name and not
265adjacent 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
274Use one space around (on each side of) most binary and ternary operators,
275such as any of these::
276
277 = + - < > * / % | & ^ <= >= == != ? :
278
279but no space after unary operators::
280
281 & * + - ~ ! sizeof typeof alignof __attribute__ defined
282
283no space before the postfix increment & decrement unary operators::
284
285 ++ --
286
287no space after the prefix increment & decrement unary operators::
288
289 ++ --
290
291and no space around the ``.`` and ``->`` structure member operators.
292
293Do 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
295appropriate, so you can start typing the next line of code right away.
296However, some such editors do not remove the whitespace if you end up not
297putting a line of code there, such as if you leave a blank line. As a result,
298you end up with lines containing trailing whitespace.
299
300Git will warn you about patches that introduce trailing whitespace, and can
301optionally strip the trailing whitespace for you; however, if applying a series
302of patches, this may make later patches in the series fail by changing their
303context lines.
304
305
3064) Naming
307---------
308
309C is a Spartan language, and your naming conventions should follow suit.
310Unlike Modula-2 and Pascal programmers, C programmers do not use cute
311names like ThisVariableIsATemporaryCounter. A C programmer would call that
312variable ``tmp``, which is much easier to write, and not the least more
313difficult to understand.
314
315HOWEVER, while mixed-case names are frowned upon, descriptive names for
316global variables are a must. To call a global function ``foo`` is a
317shooting offense.
318
319GLOBAL variables (to be used only if you **really** need them) need to
320have descriptive names, as do global functions. If you have a function
321that counts the number of active users, you should call that
322``count_active_users()`` or similar, you should **not** call it ``cntusr()``.
323
324Encoding the type of a function into the name (so-called Hungarian
325notation) is asinine - the compiler knows the types anyway and can check
326those, and it only confuses the programmer.
327
328LOCAL variable names should be short, and to the point. If you have
329some random integer loop counter, it should probably be called ``i``.
330Calling it ``loop_counter`` is non-productive, if there is no chance of it
331being mis-understood. Similarly, ``tmp`` can be just about any type of
332variable that is used to hold a temporary value.
333
334If you are afraid to mix up your local variable names, you have another
335problem, which is called the function-growth-hormone-imbalance syndrome.
336See chapter 6 (Functions).
337
338For symbol names and documentation, avoid introducing new usage of
339'master / slave' (or 'slave' independent of 'master') and 'blacklist /
340whitelist'.
341
342Recommended 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
349Recommended replacements for 'blacklist/whitelist' are:
350 'denylist / allowlist'
351 'blocklist / passlist'
352
353Exceptions for introducing new usage is to maintain a userspace ABI/API,
354or when updating code for an existing (as of 2020) hardware or protocol
355specification that mandates those terms. For new specifications
356translate specification usage of the terminology to the kernel coding
357standard where possible.
358
3595) Typedefs
360-----------
361
362Please don't use things like ``vps_t``.
363It'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
370in the source, what does it mean?
371In contrast, if it says
372
373.. code-block:: c
374
375 struct virtual_container *a;
376
377you can actually tell what ``a`` is.
378
379Lots of people think that typedefs ``help readability``. Not so. They are
380useful 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
436Maybe there are other cases too, but the rule should basically be to NEVER
437EVER use a typedef unless you can clearly match one of those rules.
438
439In general, a pointer, or a struct that has elements that can reasonably
440be directly accessed should **never** be a typedef.
441
442
4436) Functions
444------------
445
446Functions should be short and sweet, and do just one thing. They should
447fit on one or two screenfuls of text (the ISO/ANSI screen size is 80x24,
448as we all know), and do one thing and do that well.
449
450The maximum length of a function is inversely proportional to the
451complexity and indentation level of that function. So, if you have a
452conceptually simple function that is just one long (but simple)
453case-statement, where you have to do lots of small things for a lot of
454different cases, it's OK to have a longer function.
455
456However, if you have a complex function, and you suspect that a
457less-than-gifted first-year high-school student might not even
458understand what the function is all about, you should adhere to the
459maximum limits all the more closely. Use helper functions with
460descriptive names (you can ask the compiler to in-line them if you think
461it's performance-critical, and it will probably do a better job of it
462than you would have done).
463
464Another measure of the function is the number of local variables. They
465shouldn't exceed 5-10, or you're doing something wrong. Re-think the
466function, and split it into smaller pieces. A human brain can
467generally easily keep track of about 7 different things, anything more
468and it gets confused. You know you're brilliant, but maybe you'd like
469to understand what you did 2 weeks from now.
470
471In source files, separate functions with one blank line. If the function is
472exported, the **EXPORT** macro for it should follow immediately after the
473closing 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
4836.1) Function prototypes
484************************
485
486In function prototypes, include parameter names with their data types.
487Although this is not required by the C language, it is preferred in Linux
488because it is a simple way to add valuable information for the reader.
489
490Do not use the ``extern`` keyword with function declarations as this makes
491lines longer and isn't strictly necessary.
492
493When writing function prototypes, please keep the `order of elements regular
494<https://lore.kernel.org/mm-commits/CAHk-=wiOCLRny5aifWNhr621kYrJwhfURsa0vFPeUEm8mF0ufg@mail.gmail.com/>`_.
495For 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
500The 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
514Note that for a function **definition** (i.e. the actual function body),
515the compiler does not allow function parameter attributes after the
516function parameters. In these cases, they should go after the storage
517class attributes (e.g. note the changed position of ``__printf(4, 5)``
518below, 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
5267) Centralized exiting of functions
527-----------------------------------
528
529Albeit deprecated by some people, the equivalent of the goto statement is
530used frequently by compilers in form of the unconditional jump instruction.
531
532The goto statement comes in handy when a function exits from multiple
533locations and some common work such as cleanup has to be done. If there is no
534cleanup needed then just return directly.
535
536Choose label names which say what the goto does or why the goto exists. An
537example of a good name could be ``out_free_buffer:`` if the goto frees ``buffer``.
538Avoid using GW-BASIC names like ``err1:`` and ``err2:``, as you would have to
539renumber them if you ever add or remove exit paths, and they make correctness
540difficult to verify anyway.
541
542The 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
574A 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
583The bug in this code is that on some exit paths ``foo`` is NULL. Normally the
584fix 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
595Ideally you should simulate errors to test all exit paths.
596
597
5988) Commenting
599-------------
600
601Comments are good, but there is also a danger of over-commenting. NEVER
602try to explain HOW your code works in a comment: it's much better to
603write the code so that the **working** is obvious, and it's a waste of
604time to explain badly written code.
605
606Generally, you want your comments to tell WHAT your code does, not HOW.
607Also, try to avoid putting comments inside a function body: if the
608function is so complex that you need to separately comment parts of it,
609you should probably go back to chapter 6 for a while. You can make
610small comments to note or warn about something particularly clever (or
611ugly), but try to avoid excess. Instead, put the comments at the head
612of the function, telling people what it does, and possibly WHY it does
613it.
614
615When commenting the kernel API functions, please use the kernel-doc format.
616See the files at :ref:`Documentation/doc-guide/ <doc_guide>` and
617``scripts/kernel-doc`` for details.
618
619The 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
632For files in net/ and drivers/net/ the preferred style for long (multi-line)
633comments is a little different.
634
635.. code-block:: c
636
637 /* The preferred comment style for files in net/ and drivers/net
638 * looks like this.
639 *
640 * It is nearly the same as the generally preferred comment style,
641 * but there is no initial almost-blank line.
642 */
643
644It's also important to comment data, whether they are basic types or derived
645types. To this end, use just one data declaration per line (no commas for
646multiple data declarations). This leaves you room for a small comment on each
647item, explaining its use.
648
649
6509) You've made a mess of it
651---------------------------
652
653That's OK, we all do. You've probably been told by your long-time Unix
654user helper that ``GNU emacs`` automatically formats the C sources for
655you, and you've noticed that yes, it does do that, but the defaults it
656uses are less than desirable (in fact, they are worse than random
657typing - an infinite number of monkeys typing into GNU emacs would never
658make a good program).
659
660So, you can either get rid of GNU emacs, or change it to use saner
661values. To do the latter, you can stick the following in your .emacs file:
662
663.. code-block:: elisp
664
665 (defun c-lineup-arglist-tabs-only (ignored)
666 "Line up argument lists by tabs, not spaces"
667 (let* ((anchor (c-langelem-pos c-syntactic-element))
668 (column (c-langelem-2nd-pos c-syntactic-element))
669 (offset (- (1+ column) anchor))
670 (steps (floor offset c-basic-offset)))
671 (* (max steps 1)
672 c-basic-offset)))
673
674 (dir-locals-set-class-variables
675 'linux-kernel
676 '((c-mode . (
677 (c-basic-offset . 8)
678 (c-label-minimum-indentation . 0)
679 (c-offsets-alist . (
680 (arglist-close . c-lineup-arglist-tabs-only)
681 (arglist-cont-nonempty .
682 (c-lineup-gcc-asm-reg c-lineup-arglist-tabs-only))
683 (arglist-intro . +)
684 (brace-list-intro . +)
685 (c . c-lineup-C-comments)
686 (case-label . 0)
687 (comment-intro . c-lineup-comment)
688 (cpp-define-intro . +)
689 (cpp-macro . -1000)
690 (cpp-macro-cont . +)
691 (defun-block-intro . +)
692 (else-clause . 0)
693 (func-decl-cont . +)
694 (inclass . +)
695 (inher-cont . c-lineup-multi-inher)
696 (knr-argdecl-intro . 0)
697 (label . -1000)
698 (statement . 0)
699 (statement-block-intro . +)
700 (statement-case-intro . +)
701 (statement-cont . +)
702 (substatement . +)
703 ))
704 (indent-tabs-mode . t)
705 (show-trailing-whitespace . t)
706 ))))
707
708 (dir-locals-set-directory-class
709 (expand-file-name "~/src/linux-trees")
710 'linux-kernel)
711
712This will make emacs go better with the kernel coding style for C
713files below ``~/src/linux-trees``.
714
715But even if you fail in getting emacs to do sane formatting, not
716everything is lost: use ``indent``.
717
718Now, again, GNU indent has the same brain-dead settings that GNU emacs
719has, which is why you need to give it a few command line options.
720However, that's not too bad, because even the makers of GNU indent
721recognize the authority of K&R (the GNU people aren't evil, they are
722just severely misguided in this matter), so you just give indent the
723options ``-kr -i8`` (stands for ``K&R, 8 character indents``), or use
724``scripts/Lindent``, which indents in the latest style.
725
726``indent`` has a lot of options, and especially when it comes to comment
727re-formatting you may want to take a look at the man page. But
728remember: ``indent`` is not a fix for bad programming.
729
730Note that you can also use the ``clang-format`` tool to help you with
731these rules, to quickly re-format parts of your code automatically,
732and to review full files in order to spot coding style mistakes,
733typos and possible improvements. It is also handy for sorting ``#includes``,
734for aligning variables/macros, for reflowing text and other similar tasks.
735See the file :ref:`Documentation/process/clang-format.rst <clangformat>`
736for more details.
737
738Some basic editor settings, such as indentation and line endings, will be
739set automatically if you are using an editor that is compatible with
740EditorConfig. See the official EditorConfig website for more information:
741https://editorconfig.org/
742
74310) Kconfig configuration files
744-------------------------------
745
746For all of the Kconfig* configuration files throughout the source tree,
747the indentation is somewhat different. Lines under a ``config`` definition
748are indented with one tab, while help text is indented an additional two
749spaces. Example::
750
751 config AUDIT
752 bool "Auditing support"
753 depends on NET
754 help
755 Enable auditing infrastructure that can be used with another
756 kernel subsystem, such as SELinux (which requires this for
757 logging of avc messages output). Does not do system-call
758 auditing without CONFIG_AUDITSYSCALL.
759
760Seriously dangerous features (such as write support for certain
761filesystems) should advertise this prominently in their prompt string::
762
763 config ADFS_FS_RW
764 bool "ADFS write support (DANGEROUS)"
765 depends on ADFS_FS
766 ...
767
768For full documentation on the configuration files, see the file
769Documentation/kbuild/kconfig-language.rst.
770
771
77211) Data structures
773-------------------
774
775Data structures that have visibility outside the single-threaded
776environment they are created and destroyed in should always have
777reference counts. In the kernel, garbage collection doesn't exist (and
778outside the kernel garbage collection is slow and inefficient), which
779means that you absolutely **have** to reference count all your uses.
780
781Reference counting means that you can avoid locking, and allows multiple
782users to have access to the data structure in parallel - and not having
783to worry about the structure suddenly going away from under them just
784because they slept or did something else for a while.
785
786Note that locking is **not** a replacement for reference counting.
787Locking is used to keep data structures coherent, while reference
788counting is a memory management technique. Usually both are needed, and
789they are not to be confused with each other.
790
791Many data structures can indeed have two levels of reference counting,
792when there are users of different ``classes``. The subclass count counts
793the number of subclass users, and decrements the global count just once
794when the subclass count goes to zero.
795
796Examples of this kind of ``multi-level-reference-counting`` can be found in
797memory management (``struct mm_struct``: mm_users and mm_count), and in
798filesystem code (``struct super_block``: s_count and s_active).
799
800Remember: if another thread can find your data structure, and you don't
801have a reference count on it, you almost certainly have a bug.
802
803
80412) Macros, Enums and RTL
805-------------------------
806
807Names of macros defining constants and labels in enums are capitalized.
808
809.. code-block:: c
810
811 #define CONSTANT 0x12345
812
813Enums are preferred when defining several related constants.
814
815CAPITALIZED macro names are appreciated but macros resembling functions
816may be named in lower case.
817
818Generally, inline functions are preferable to macros resembling functions.
819
820Macros with multiple statements should be enclosed in a do - while block:
821
822.. code-block:: c
823
824 #define macrofun(a, b, c) \
825 do { \
826 if (a == 5) \
827 do_this(b, c); \
828 } while (0)
829
830Things to avoid when using macros:
831
8321) macros that affect control flow:
833
834.. code-block:: c
835
836 #define FOO(x) \
837 do { \
838 if (blah(x) < 0) \
839 return -EBUGGERED; \
840 } while (0)
841
842is a **very** bad idea. It looks like a function call but exits the ``calling``
843function; don't break the internal parsers of those who will read the code.
844
8452) macros that depend on having a local variable with a magic name:
846
847.. code-block:: c
848
849 #define FOO(val) bar(index, val)
850
851might look like a good thing, but it's confusing as hell when one reads the
852code and it's prone to breakage from seemingly innocent changes.
853
8543) macros with arguments that are used as l-values: FOO(x) = y; will
855bite you if somebody e.g. turns FOO into an inline function.
856
8574) forgetting about precedence: macros defining constants using expressions
858must enclose the expression in parentheses. Beware of similar issues with
859macros using parameters.
860
861.. code-block:: c
862
863 #define CONSTANT 0x4000
864 #define CONSTEXP (CONSTANT | 3)
865
8665) namespace collisions when defining local variables in macros resembling
867functions:
868
869.. code-block:: c
870
871 #define FOO(x) \
872 ({ \
873 typeof(x) ret; \
874 ret = calc_ret(x); \
875 (ret); \
876 })
877
878ret is a common name for a local variable - __foo_ret is less likely
879to collide with an existing variable.
880
881The cpp manual deals with macros exhaustively. The gcc internals manual also
882covers RTL which is used frequently with assembly language in the kernel.
883
884
88513) Printing kernel messages
886----------------------------
887
888Kernel developers like to be seen as literate. Do mind the spelling
889of kernel messages to make a good impression. Do not use incorrect
890contractions like ``dont``; use ``do not`` or ``don't`` instead. Make the
891messages concise, clear, and unambiguous.
892
893Kernel messages do not have to be terminated with a period.
894
895Printing numbers in parentheses (%d) adds no value and should be avoided.
896
897There are a number of driver model diagnostic macros in <linux/dev_printk.h>
898which you should use to make sure messages are matched to the right device
899and driver, and are tagged with the right level: dev_err(), dev_warn(),
900dev_info(), and so forth. For messages that aren't associated with a
901particular device, <linux/printk.h> defines pr_notice(), pr_info(),
902pr_warn(), pr_err(), etc. When drivers are working properly they are quiet,
903so prefer to use dev_dbg/pr_debug unless something is wrong.
904
905Coming up with good debugging messages can be quite a challenge; and once
906you have them, they can be a huge help for remote troubleshooting. However
907debug message printing is handled differently than printing other non-debug
908messages. While the other pr_XXX() functions print unconditionally,
909pr_debug() does not; it is compiled out by default, unless either DEBUG is
910defined or CONFIG_DYNAMIC_DEBUG is set. That is true for dev_dbg() also,
911and a related convention uses VERBOSE_DEBUG to add dev_vdbg() messages to
912the ones already enabled by DEBUG.
913
914Many subsystems have Kconfig debug options to turn on -DDEBUG in the
915corresponding Makefile; in other cases specific files #define DEBUG. And
916when a debug message should be unconditionally printed, such as if it is
917already inside a debug-related #ifdef section, printk(KERN_DEBUG ...) can be
918used.
919
920
92114) Allocating memory
922---------------------
923
924The kernel provides the following general purpose memory allocators:
925kmalloc(), kzalloc(), kmalloc_array(), kcalloc(), vmalloc(), and
926vzalloc(). Please refer to the API documentation for further information
927about them. :ref:`Documentation/core-api/memory-allocation.rst
928<memory_allocation>`
929
930The preferred form for passing a size of a struct is the following:
931
932.. code-block:: c
933
934 p = kmalloc(sizeof(*p), ...);
935
936The alternative form where struct name is spelled out hurts readability and
937introduces an opportunity for a bug when the pointer variable type is changed
938but the corresponding sizeof that is passed to a memory allocator is not.
939
940Casting the return value which is a void pointer is redundant. The conversion
941from void pointer to any other pointer type is guaranteed by the C programming
942language.
943
944The preferred form for allocating an array is the following:
945
946.. code-block:: c
947
948 p = kmalloc_array(n, sizeof(...), ...);
949
950The preferred form for allocating a zeroed array is the following:
951
952.. code-block:: c
953
954 p = kcalloc(n, sizeof(...), ...);
955
956Both forms check for overflow on the allocation size n * sizeof(...),
957and return NULL if that occurred.
958
959These generic allocation functions all emit a stack dump on failure when used
960without __GFP_NOWARN so there is no use in emitting an additional failure
961message when NULL is returned.
962
96315) The inline disease
964----------------------
965
966There appears to be a common misperception that gcc has a magic "make me
967faster" speedup option called ``inline``. While the use of inlines can be
968appropriate (for example as a means of replacing macros, see Chapter 12), it
969very often is not. Abundant use of the inline keyword leads to a much bigger
970kernel, which in turn slows the system as a whole down, due to a bigger
971icache footprint for the CPU and simply because there is less memory
972available for the pagecache. Just think about it; a pagecache miss causes a
973disk seek, which easily takes 5 milliseconds. There are a LOT of cpu cycles
974that can go into these 5 milliseconds.
975
976A reasonable rule of thumb is to not put inline at functions that have more
977than 3 lines of code in them. An exception to this rule are the cases where
978a parameter is known to be a compiletime constant, and as a result of this
979constantness you *know* the compiler will be able to optimize most of your
980function away at compile time. For a good example of this later case, see
981the kmalloc() inline function.
982
983Often people argue that adding inline to functions that are static and used
984only once is always a win since there is no space tradeoff. While this is
985technically correct, gcc is capable of inlining these automatically without
986help, and the maintenance issue of removing the inline when a second user
987appears outweighs the potential value of the hint that tells gcc to do
988something it would have done anyway.
989
990
99116) Function return values and names
992------------------------------------
993
994Functions can return values of many different kinds, and one of the
995most common is a value indicating whether the function succeeded or
996failed. Such a value can be represented as an error-code integer
997(-Exxx = failure, 0 = success) or a ``succeeded`` boolean (0 = failure,
998non-zero = success).
999
1000Mixing up these two sorts of representations is a fertile source of
1001difficult-to-find bugs. If the C language included a strong distinction
1002between integers and booleans then the compiler would find these mistakes
1003for us... but it doesn't. To help prevent such bugs, always follow this
1004convention::
1005
1006 If the name of a function is an action or an imperative command,
1007 the function should return an error-code integer. If the name
1008 is a predicate, the function should return a "succeeded" boolean.
1009
1010For example, ``add work`` is a command, and the add_work() function returns 0
1011for success or -EBUSY for failure. In the same way, ``PCI device present`` is
1012a predicate, and the pci_dev_present() function returns 1 if it succeeds in
1013finding a matching device or 0 if it doesn't.
1014
1015All EXPORTed functions must respect this convention, and so should all
1016public functions. Private (static) functions need not, but it is
1017recommended that they do.
1018
1019Functions whose return value is the actual result of a computation, rather
1020than an indication of whether the computation succeeded, are not subject to
1021this rule. Generally they indicate failure by returning some out-of-range
1022result. Typical examples would be functions that return pointers; they use
1023NULL or the ERR_PTR mechanism to report failure.
1024
1025
102617) Using bool
1027--------------
1028
1029The Linux kernel bool type is an alias for the C99 _Bool type. bool values can
1030only evaluate to 0 or 1, and implicit or explicit conversion to bool
1031automatically converts the value to true or false. When using bool types the
1032!! construction is not needed, which eliminates a class of bugs.
1033
1034When working with bool values the true and false definitions should be used
1035instead of 1 and 0.
1036
1037bool function return types and stack variables are always fine to use whenever
1038appropriate. Use of bool is encouraged to improve readability and is often a
1039better option than 'int' for storing boolean values.
1040
1041Do not use bool if cache line layout or size of the value matters, as its size
1042and alignment varies based on the compiled architecture. Structures that are
1043optimized for alignment and size should not use bool.
1044
1045If a structure has many true/false values, consider consolidating them into a
1046bitfield with 1 bit members, or using an appropriate fixed width type, such as
1047u8.
1048
1049Similarly for function arguments, many true/false values can be consolidated
1050into a single bitwise 'flags' argument and 'flags' can often be a more
1051readable alternative if the call-sites have naked true/false constants.
1052
1053Otherwise limited use of bool in structures and arguments can improve
1054readability.
1055
105618) Don't re-invent the kernel macros
1057-------------------------------------
1058
1059The header file include/linux/kernel.h contains a number of macros that
1060you should use, rather than explicitly coding some variant of them yourself.
1061For example, if you need to calculate the length of an array, take advantage
1062of the macro
1063
1064.. code-block:: c
1065
1066 #define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
1067
1068Similarly, if you need to calculate the size of some structure member, use
1069
1070.. code-block:: c
1071
1072 #define sizeof_field(t, f) (sizeof(((t*)0)->f))
1073
1074There are also min() and max() macros that do strict type checking if you
1075need them. Feel free to peruse that header file to see what else is already
1076defined that you shouldn't reproduce in your code.
1077
1078
107919) Editor modelines and other cruft
1080------------------------------------
1081
1082Some editors can interpret configuration information embedded in source files,
1083indicated with special markers. For example, emacs interprets lines marked
1084like this:
1085
1086.. code-block:: c
1087
1088 -*- mode: c -*-
1089
1090Or like this:
1091
1092.. code-block:: c
1093
1094 /*
1095 Local Variables:
1096 compile-command: "gcc -DMAGIC_DEBUG_FLAG foo.c"
1097 End:
1098 */
1099
1100Vim interprets markers that look like this:
1101
1102.. code-block:: c
1103
1104 /* vim:set sw=8 noet */
1105
1106Do not include any of these in source files. People have their own personal
1107editor configurations, and your source files should not override them. This
1108includes markers for indentation and mode configuration. People may use their
1109own custom mode, or may have some other magic method for making indentation
1110work correctly.
1111
1112
111320) Inline assembly
1114-------------------
1115
1116In architecture-specific code, you may need to use inline assembly to interface
1117with CPU or platform functionality. Don't hesitate to do so when necessary.
1118However, don't use inline assembly gratuitously when C can do the job. You can
1119and should poke hardware from C when possible.
1120
1121Consider writing simple helper functions that wrap common bits of inline
1122assembly, rather than repeatedly writing them with slight variations. Remember
1123that inline assembly can use C parameters.
1124
1125Large, non-trivial assembly functions should go in .S files, with corresponding
1126C prototypes defined in C header files. The C prototypes for assembly
1127functions should use ``asmlinkage``.
1128
1129You may need to mark your asm statement as volatile, to prevent GCC from
1130removing it if GCC doesn't notice any side effects. You don't always need to
1131do so, though, and doing so unnecessarily can limit optimization.
1132
1133When writing a single inline assembly statement containing multiple
1134instructions, put each instruction on a separate line in a separate quoted
1135string, and end each string except the last with ``\n\t`` to properly indent
1136the next instruction in the assembly output:
1137
1138.. code-block:: c
1139
1140 asm ("magic %reg1, #42\n\t"
1141 "more_magic %reg2, %reg3"
1142 : /* outputs */ : /* inputs */ : /* clobbers */);
1143
1144
114521) Conditional Compilation
1146---------------------------
1147
1148Wherever possible, don't use preprocessor conditionals (#if, #ifdef) in .c
1149files; doing so makes code harder to read and logic harder to follow. Instead,
1150use such conditionals in a header file defining functions for use in those .c
1151files, providing no-op stub versions in the #else case, and then call those
1152functions unconditionally from .c files. The compiler will avoid generating
1153any code for the stub calls, producing identical results, but the logic will
1154remain easy to follow.
1155
1156Prefer to compile out entire functions, rather than portions of functions or
1157portions of expressions. Rather than putting an ifdef in an expression, factor
1158out part or all of the expression into a separate helper function and apply the
1159conditional to that function.
1160
1161If you have a function or variable which may potentially go unused in a
1162particular configuration, and the compiler would warn about its definition
1163going unused, mark the definition as __maybe_unused rather than wrapping it in
1164a preprocessor conditional. (However, if a function or variable *always* goes
1165unused, delete it.)
1166
1167Within code, where possible, use the IS_ENABLED macro to convert a Kconfig
1168symbol into a C boolean expression, and use it in a normal C conditional:
1169
1170.. code-block:: c
1171
1172 if (IS_ENABLED(CONFIG_SOMETHING)) {
1173 ...
1174 }
1175
1176The compiler will constant-fold the conditional away, and include or exclude
1177the block of code just as with an #ifdef, so this will not add any runtime
1178overhead. However, this approach still allows the C compiler to see the code
1179inside the block, and check it for correctness (syntax, types, symbol
1180references, etc). Thus, you still have to use an #ifdef if the code inside the
1181block references symbols that will not exist if the condition is not met.
1182
1183At the end of any non-trivial #if or #ifdef block (more than a few lines),
1184place a comment after the #endif on the same line, noting the conditional
1185expression used. For instance:
1186
1187.. code-block:: c
1188
1189 #ifdef CONFIG_SOMETHING
1190 ...
1191 #endif /* CONFIG_SOMETHING */
1192
1193
119422) Do not crash the kernel
1195---------------------------
1196
1197In general, the decision to crash the kernel belongs to the user, rather
1198than to the kernel developer.
1199
1200Avoid panic()
1201*************
1202
1203panic() should be used with care and primarily only during system boot.
1204panic() is, for example, acceptable when running out of memory during boot and
1205not being able to continue.
1206
1207Use WARN() rather than BUG()
1208****************************
1209
1210Do not add new code that uses any of the BUG() variants, such as BUG(),
1211BUG_ON(), or VM_BUG_ON(). Instead, use a WARN*() variant, preferably
1212WARN_ON_ONCE(), and possibly with recovery code. Recovery code is not
1213required if there is no reasonable way to at least partially recover.
1214
1215"I'm too lazy to do error handling" is not an excuse for using BUG(). Major
1216internal corruptions with no way of continuing may still use BUG(), but need
1217good justification.
1218
1219Use WARN_ON_ONCE() rather than WARN() or WARN_ON()
1220**************************************************
1221
1222WARN_ON_ONCE() is generally preferred over WARN() or WARN_ON(), because it
1223is common for a given warning condition, if it occurs at all, to occur
1224multiple times. This can fill up and wrap the kernel log, and can even slow
1225the system enough that the excessive logging turns into its own, additional
1226problem.
1227
1228Do not WARN lightly
1229*******************
1230
1231WARN*() is intended for unexpected, this-should-never-happen situations.
1232WARN*() macros are not to be used for anything that is expected to happen
1233during normal operation. These are not pre- or post-condition asserts, for
1234example. Again: WARN*() must not be used for a condition that is expected
1235to trigger easily, for example, by user space actions. pr_warn_once() is a
1236possible alternative, if you need to notify the user of a problem.
1237
1238Do not worry about panic_on_warn users
1239**************************************
1240
1241A few more words about panic_on_warn: Remember that ``panic_on_warn`` is an
1242available kernel option, and that many users set this option. This is why
1243there is a "Do not WARN lightly" writeup, above. However, the existence of
1244panic_on_warn users is not a valid reason to avoid the judicious use
1245WARN*(). That is because, whoever enables panic_on_warn has explicitly
1246asked the kernel to crash if a WARN*() fires, and such users must be
1247prepared to deal with the consequences of a system that is somewhat more
1248likely to crash.
1249
1250Use BUILD_BUG_ON() for compile-time assertions
1251**********************************************
1252
1253The use of BUILD_BUG_ON() is acceptable and encouraged, because it is a
1254compile-time assertion that has no effect at runtime.
1255
1256Appendix I) References
1257----------------------
1258
1259The C Programming Language, Second Edition
1260by Brian W. Kernighan and Dennis M. Ritchie.
1261Prentice Hall, Inc., 1988.
1262ISBN 0-13-110362-8 (paperback), 0-13-110370-9 (hardback).
1263
1264The Practice of Programming
1265by Brian W. Kernighan and Rob Pike.
1266Addison-Wesley, Inc., 1999.
1267ISBN 0-201-61586-X.
1268
1269GNU manuals - where in compliance with K&R and this text - for cpp, gcc,
1270gcc internals and indent, all available from https://www.gnu.org/manual/
1271
1272WG14 is the international standardization working group for the programming
1273language C, URL: http://www.open-std.org/JTC1/SC22/WG14/
1274
1275Kernel CodingStyle, by greg@kroah.com at OLS 2002:
1276http://www.kroah.com/linux/talks/ols_2002_kernel_codingstyle_talk/html/