blob: fa6e25b94a54bfe1d39450cd9bc568c4043ddfa3 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001
2 Linux kernel coding style
3
4This is a short document describing the preferred coding style for the
5linux kernel. Coding style is very personal, and I won't _force_ my
6views on anybody, but this is what goes for anything that I have to be
7able to maintain, and I'd prefer it for most other things too. Please
8at least consider the points made here.
9
10First off, I'd suggest printing out a copy of the GNU coding standards,
11and NOT read it. Burn them, it's a great symbolic gesture.
12
13Anyway, here goes:
14
15
16 Chapter 1: Indentation
17
18Tabs are 8 characters, and thus indentations are also 8 characters.
19There are heretic movements that try to make indentations 4 (or even 2!)
20characters deep, and that is akin to trying to define the value of PI to
21be 3.
22
23Rationale: The whole idea behind indentation is to clearly define where
24a block of control starts and ends. Especially when you've been looking
25at your screen for 20 straight hours, you'll find it a lot easier to see
26how the indentation works if you have large indentations.
27
28Now, some people will claim that having 8-character indentations makes
29the code move too far to the right, and makes it hard to read on a
3080-character terminal screen. The answer to that is that if you need
31more than 3 levels of indentation, you're screwed anyway, and should fix
32your program.
33
34In short, 8-char indents make things easier to read, and have the added
35benefit of warning you when you're nesting your functions too deep.
36Heed that warning.
37
Randy Dunlapb3fc9942006-12-10 02:18:56 -080038The preferred way to ease multiple indentation levels in a switch statement is
39to align the "switch" and its subordinate "case" labels in the same column
40instead of "double-indenting" the "case" labels. E.g.:
41
42 switch (suffix) {
43 case 'G':
44 case 'g':
45 mem <<= 30;
46 break;
47 case 'M':
48 case 'm':
49 mem <<= 20;
50 break;
51 case 'K':
52 case 'k':
53 mem <<= 10;
54 /* fall through */
55 default:
56 break;
57 }
58
59
Linus Torvalds1da177e2005-04-16 15:20:36 -070060Don't put multiple statements on a single line unless you have
61something to hide:
62
63 if (condition) do_this;
64 do_something_everytime;
65
Randy Dunlapb3fc9942006-12-10 02:18:56 -080066Don't put multiple assignments on a single line either. Kernel coding style
67is super simple. Avoid tricky expressions.
68
Linus Torvalds1da177e2005-04-16 15:20:36 -070069Outside of comments, documentation and except in Kconfig, spaces are never
70used for indentation, and the above example is deliberately broken.
71
72Get a decent editor and don't leave whitespace at the end of lines.
73
74
75 Chapter 2: Breaking long lines and strings
76
77Coding style is all about readability and maintainability using commonly
78available tools.
79
Alan Coxdff49822007-10-16 23:27:33 -070080The limit on the length of lines is 80 columns and this is a strongly
81preferred limit.
Linus Torvalds1da177e2005-04-16 15:20:36 -070082
83Statements longer than 80 columns will be broken into sensible chunks.
84Descendants are always substantially shorter than the parent and are placed
85substantially to the right. The same applies to function headers with a long
Alan Coxdff49822007-10-16 23:27:33 -070086argument list. Long strings are as well broken into shorter strings. The
87only exception to this is where exceeding 80 columns significantly increases
88readability and does not hide information.
Linus Torvalds1da177e2005-04-16 15:20:36 -070089
90void fun(int a, int b, int c)
91{
92 if (condition)
93 printk(KERN_WARNING "Warning this is a long printk with "
94 "3 parameters a: %u b: %u "
95 "c: %u \n", a, b, c);
96 else
97 next_statement;
98}
99
Randy Dunlapb3fc9942006-12-10 02:18:56 -0800100 Chapter 3: Placing Braces and Spaces
Linus Torvalds1da177e2005-04-16 15:20:36 -0700101
102The other issue that always comes up in C styling is the placement of
103braces. Unlike the indent size, there are few technical reasons to
104choose one placement strategy over the other, but the preferred way, as
105shown to us by the prophets Kernighan and Ritchie, is to put the opening
106brace last on the line, and put the closing brace first, thusly:
107
108 if (x is true) {
109 we do y
110 }
111
Randy Dunlapb3fc9942006-12-10 02:18:56 -0800112This applies to all non-function statement blocks (if, switch, for,
113while, do). E.g.:
114
115 switch (action) {
116 case KOBJ_ADD:
117 return "add";
118 case KOBJ_REMOVE:
119 return "remove";
120 case KOBJ_CHANGE:
121 return "change";
122 default:
123 return NULL;
124 }
125
Linus Torvalds1da177e2005-04-16 15:20:36 -0700126However, there is one special case, namely functions: they have the
127opening brace at the beginning of the next line, thus:
128
129 int function(int x)
130 {
131 body of function
132 }
133
134Heretic people all over the world have claimed that this inconsistency
135is ... well ... inconsistent, but all right-thinking people know that
136(a) K&R are _right_ and (b) K&R are right. Besides, functions are
137special anyway (you can't nest them in C).
138
139Note that the closing brace is empty on a line of its own, _except_ in
140the cases where it is followed by a continuation of the same statement,
141ie a "while" in a do-statement or an "else" in an if-statement, like
142this:
143
144 do {
145 body of do-loop
146 } while (condition);
147
148and
149
150 if (x == y) {
151 ..
152 } else if (x > y) {
153 ...
154 } else {
155 ....
156 }
157
158Rationale: K&R.
159
160Also, note that this brace-placement also minimizes the number of empty
161(or almost empty) lines, without any loss of readability. Thus, as the
162supply of new-lines on your screen is not a renewable resource (think
16325-line terminal screens here), you have more empty lines to put
164comments on.
165
Oliver Neukume659ba42007-05-08 00:30:34 -0700166Do not unnecessarily use braces where a single statement will do.
167
168if (condition)
169 action();
170
Harry Wei38829dc2011-03-22 16:35:01 -0700171and
172
173if (condition)
174 do_this();
175else
176 do_that();
177
Oliver Neukume659ba42007-05-08 00:30:34 -0700178This does not apply if one branch of a conditional statement is a single
179statement. Use braces in both branches.
180
181if (condition) {
182 do_this();
183 do_that();
184} else {
185 otherwise();
186}
187
Randy Dunlapb3fc9942006-12-10 02:18:56 -0800188 3.1: Spaces
189
190Linux kernel style for use of spaces depends (mostly) on
191function-versus-keyword usage. Use a space after (most) keywords. The
192notable exceptions are sizeof, typeof, alignof, and __attribute__, which look
193somewhat like functions (and are usually used with parentheses in Linux,
194although they are not required in the language, as in: "sizeof info" after
195"struct fileinfo info;" is declared).
196
197So use a space after these keywords:
198 if, switch, case, for, do, while
199but not with sizeof, typeof, alignof, or __attribute__. E.g.,
200 s = sizeof(struct file);
201
202Do not add spaces around (inside) parenthesized expressions. This example is
203*bad*:
204
205 s = sizeof( struct file );
206
207When declaring pointer data or a function that returns a pointer type, the
208preferred use of '*' is adjacent to the data name or function name and not
209adjacent to the type name. Examples:
210
211 char *linux_banner;
212 unsigned long long memparse(char *ptr, char **retptr);
213 char *match_strdup(substring_t *s);
214
215Use one space around (on each side of) most binary and ternary operators,
216such as any of these:
217
218 = + - < > * / % | & ^ <= >= == != ? :
219
220but no space after unary operators:
221 & * + - ~ ! sizeof typeof alignof __attribute__ defined
222
223no space before the postfix increment & decrement unary operators:
224 ++ --
225
226no space after the prefix increment & decrement unary operators:
227 ++ --
228
229and no space around the '.' and "->" structure member operators.
230
Josh Tripletta923fd62007-07-15 23:41:37 -0700231Do not leave trailing whitespace at the ends of lines. Some editors with
232"smart" indentation will insert whitespace at the beginning of new lines as
233appropriate, so you can start typing the next line of code right away.
234However, some such editors do not remove the whitespace if you end up not
235putting a line of code there, such as if you leave a blank line. As a result,
236you end up with lines containing trailing whitespace.
237
238Git will warn you about patches that introduce trailing whitespace, and can
239optionally strip the trailing whitespace for you; however, if applying a series
240of patches, this may make later patches in the series fail by changing their
241context lines.
242
Linus Torvalds1da177e2005-04-16 15:20:36 -0700243
244 Chapter 4: Naming
245
246C is a Spartan language, and so should your naming be. Unlike Modula-2
247and Pascal programmers, C programmers do not use cute names like
248ThisVariableIsATemporaryCounter. A C programmer would call that
249variable "tmp", which is much easier to write, and not the least more
250difficult to understand.
251
252HOWEVER, while mixed-case names are frowned upon, descriptive names for
253global variables are a must. To call a global function "foo" is a
254shooting offense.
255
256GLOBAL variables (to be used only if you _really_ need them) need to
257have descriptive names, as do global functions. If you have a function
258that counts the number of active users, you should call that
259"count_active_users()" or similar, you should _not_ call it "cntusr()".
260
261Encoding the type of a function into the name (so-called Hungarian
262notation) is brain damaged - the compiler knows the types anyway and can
263check those, and it only confuses the programmer. No wonder MicroSoft
264makes buggy programs.
265
266LOCAL variable names should be short, and to the point. If you have
267some random integer loop counter, it should probably be called "i".
268Calling it "loop_counter" is non-productive, if there is no chance of it
269being mis-understood. Similarly, "tmp" can be just about any type of
270variable that is used to hold a temporary value.
271
272If you are afraid to mix up your local variable names, you have another
273problem, which is called the function-growth-hormone-imbalance syndrome.
Randy Dunlapb3fc9942006-12-10 02:18:56 -0800274See chapter 6 (Functions).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700275
276
Randy Dunlap226a6b82006-06-23 02:05:58 -0700277 Chapter 5: Typedefs
278
279Please don't use things like "vps_t".
280
281It's a _mistake_ to use typedef for structures and pointers. When you see a
282
283 vps_t a;
284
285in the source, what does it mean?
286
287In contrast, if it says
288
289 struct virtual_container *a;
290
291you can actually tell what "a" is.
292
293Lots of people think that typedefs "help readability". Not so. They are
294useful only for:
295
296 (a) totally opaque objects (where the typedef is actively used to _hide_
297 what the object is).
298
299 Example: "pte_t" etc. opaque objects that you can only access using
300 the proper accessor functions.
301
302 NOTE! Opaqueness and "accessor functions" are not good in themselves.
303 The reason we have them for things like pte_t etc. is that there
304 really is absolutely _zero_ portably accessible information there.
305
306 (b) Clear integer types, where the abstraction _helps_ avoid confusion
307 whether it is "int" or "long".
308
309 u8/u16/u32 are perfectly fine typedefs, although they fit into
310 category (d) better than here.
311
312 NOTE! Again - there needs to be a _reason_ for this. If something is
313 "unsigned long", then there's no reason to do
314
315 typedef unsigned long myflags_t;
316
317 but if there is a clear reason for why it under certain circumstances
318 might be an "unsigned int" and under other configurations might be
319 "unsigned long", then by all means go ahead and use a typedef.
320
321 (c) when you use sparse to literally create a _new_ type for
322 type-checking.
323
324 (d) New types which are identical to standard C99 types, in certain
325 exceptional circumstances.
326
327 Although it would only take a short amount of time for the eyes and
328 brain to become accustomed to the standard types like 'uint32_t',
329 some people object to their use anyway.
330
331 Therefore, the Linux-specific 'u8/u16/u32/u64' types and their
332 signed equivalents which are identical to standard types are
333 permitted -- although they are not mandatory in new code of your
334 own.
335
336 When editing existing code which already uses one or the other set
337 of types, you should conform to the existing choices in that code.
338
339 (e) Types safe for use in userspace.
340
341 In certain structures which are visible to userspace, we cannot
342 require C99 types and cannot use the 'u32' form above. Thus, we
343 use __u32 and similar types in all structures which are shared
344 with userspace.
345
346Maybe there are other cases too, but the rule should basically be to NEVER
347EVER use a typedef unless you can clearly match one of those rules.
348
349In general, a pointer, or a struct that has elements that can reasonably
350be directly accessed should _never_ be a typedef.
351
352
353 Chapter 6: Functions
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354
355Functions should be short and sweet, and do just one thing. They should
356fit on one or two screenfuls of text (the ISO/ANSI screen size is 80x24,
357as we all know), and do one thing and do that well.
358
359The maximum length of a function is inversely proportional to the
360complexity and indentation level of that function. So, if you have a
361conceptually simple function that is just one long (but simple)
362case-statement, where you have to do lots of small things for a lot of
363different cases, it's OK to have a longer function.
364
365However, if you have a complex function, and you suspect that a
366less-than-gifted first-year high-school student might not even
367understand what the function is all about, you should adhere to the
368maximum limits all the more closely. Use helper functions with
369descriptive names (you can ask the compiler to in-line them if you think
370it's performance-critical, and it will probably do a better job of it
371than you would have done).
372
373Another measure of the function is the number of local variables. They
374shouldn't exceed 5-10, or you're doing something wrong. Re-think the
375function, and split it into smaller pieces. A human brain can
376generally easily keep track of about 7 different things, anything more
377and it gets confused. You know you're brilliant, but maybe you'd like
378to understand what you did 2 weeks from now.
379
Randy Dunlapb3fc9942006-12-10 02:18:56 -0800380In source files, separate functions with one blank line. If the function is
381exported, the EXPORT* macro for it should follow immediately after the closing
382function brace line. E.g.:
383
384int system_is_up(void)
385{
386 return system_state == SYSTEM_RUNNING;
387}
388EXPORT_SYMBOL(system_is_up);
389
390In function prototypes, include parameter names with their data types.
391Although this is not required by the C language, it is preferred in Linux
392because it is a simple way to add valuable information for the reader.
393
Linus Torvalds1da177e2005-04-16 15:20:36 -0700394
Randy Dunlap226a6b82006-06-23 02:05:58 -0700395 Chapter 7: Centralized exiting of functions
Linus Torvalds1da177e2005-04-16 15:20:36 -0700396
397Albeit deprecated by some people, the equivalent of the goto statement is
398used frequently by compilers in form of the unconditional jump instruction.
399
400The goto statement comes in handy when a function exits from multiple
401locations and some common work such as cleanup has to be done.
402
403The rationale is:
404
405- unconditional statements are easier to understand and follow
406- nesting is reduced
407- errors by not updating individual exit points when making
408 modifications are prevented
409- saves the compiler work to optimize redundant code away ;)
410
Jesper Juhldc3d28d2006-01-09 20:53:51 -0800411int fun(int a)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700412{
413 int result = 0;
414 char *buffer = kmalloc(SIZE);
415
416 if (buffer == NULL)
417 return -ENOMEM;
418
419 if (condition1) {
420 while (loop1) {
421 ...
422 }
423 result = 1;
424 goto out;
425 }
426 ...
427out:
428 kfree(buffer);
429 return result;
430}
431
Randy Dunlap226a6b82006-06-23 02:05:58 -0700432 Chapter 8: Commenting
Linus Torvalds1da177e2005-04-16 15:20:36 -0700433
434Comments are good, but there is also a danger of over-commenting. NEVER
435try to explain HOW your code works in a comment: it's much better to
436write the code so that the _working_ is obvious, and it's a waste of
437time to explain badly written code.
438
439Generally, you want your comments to tell WHAT your code does, not HOW.
440Also, try to avoid putting comments inside a function body: if the
441function is so complex that you need to separately comment parts of it,
Randy Dunlapb3fc9942006-12-10 02:18:56 -0800442you should probably go back to chapter 6 for a while. You can make
Linus Torvalds1da177e2005-04-16 15:20:36 -0700443small comments to note or warn about something particularly clever (or
444ugly), but try to avoid excess. Instead, put the comments at the head
445of the function, telling people what it does, and possibly WHY it does
446it.
447
Randy Dunlapb3fc9942006-12-10 02:18:56 -0800448When commenting the kernel API functions, please use the kernel-doc format.
Pekka J Enberge776eba2005-09-10 00:26:44 -0700449See the files Documentation/kernel-doc-nano-HOWTO.txt and scripts/kernel-doc
450for details.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700451
Randy Dunlapb3fc9942006-12-10 02:18:56 -0800452Linux style for comments is the C89 "/* ... */" style.
453Don't use C99-style "// ..." comments.
454
455The preferred style for long (multi-line) comments is:
456
457 /*
458 * This is the preferred style for multi-line
459 * comments in the Linux kernel source code.
460 * Please use it consistently.
461 *
462 * Description: A column of asterisks on the left side,
463 * with beginning and ending almost-blank lines.
464 */
465
466It's also important to comment data, whether they are basic types or derived
467types. To this end, use just one data declaration per line (no commas for
468multiple data declarations). This leaves you room for a small comment on each
469item, explaining its use.
470
471
Randy Dunlap226a6b82006-06-23 02:05:58 -0700472 Chapter 9: You've made a mess of it
Linus Torvalds1da177e2005-04-16 15:20:36 -0700473
474That's OK, we all do. You've probably been told by your long-time Unix
475user helper that "GNU emacs" automatically formats the C sources for
476you, and you've noticed that yes, it does do that, but the defaults it
477uses are less than desirable (in fact, they are worse than random
478typing - an infinite number of monkeys typing into GNU emacs would never
479make a good program).
480
481So, you can either get rid of GNU emacs, or change it to use saner
482values. To do the latter, you can stick the following in your .emacs file:
483
Johannes Weinera7f371e2008-07-25 01:45:51 -0700484(defun c-lineup-arglist-tabs-only (ignored)
485 "Line up argument lists by tabs, not spaces"
486 (let* ((anchor (c-langelem-pos c-syntactic-element))
487 (column (c-langelem-2nd-pos c-syntactic-element))
488 (offset (- (1+ column) anchor))
489 (steps (floor offset c-basic-offset)))
490 (* (max steps 1)
491 c-basic-offset)))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700492
Teemu Likonen0acbc6c2009-01-29 16:28:16 -0800493(add-hook 'c-mode-common-hook
494 (lambda ()
495 ;; Add kernel style
496 (c-add-style
497 "linux-tabs-only"
498 '("linux" (c-offsets-alist
499 (arglist-cont-nonempty
500 c-lineup-gcc-asm-reg
501 c-lineup-arglist-tabs-only))))))
502
Johannes Weinera7f371e2008-07-25 01:45:51 -0700503(add-hook 'c-mode-hook
504 (lambda ()
505 (let ((filename (buffer-file-name)))
506 ;; Enable kernel mode for the appropriate files
507 (when (and filename
Dan Carpenter70221392009-01-29 16:28:28 -0800508 (string-match (expand-file-name "~/src/linux-trees")
509 filename))
Johannes Weinera7f371e2008-07-25 01:45:51 -0700510 (setq indent-tabs-mode t)
Teemu Likonen0acbc6c2009-01-29 16:28:16 -0800511 (c-set-style "linux-tabs-only")))))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700512
Johannes Weinera7f371e2008-07-25 01:45:51 -0700513This will make emacs go better with the kernel coding style for C
514files below ~/src/linux-trees.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700515
516But even if you fail in getting emacs to do sane formatting, not
517everything is lost: use "indent".
518
519Now, again, GNU indent has the same brain-dead settings that GNU emacs
520has, which is why you need to give it a few command line options.
521However, that's not too bad, because even the makers of GNU indent
522recognize the authority of K&R (the GNU people aren't evil, they are
523just severely misguided in this matter), so you just give indent the
524options "-kr -i8" (stands for "K&R, 8 character indents"), or use
525"scripts/Lindent", which indents in the latest style.
526
527"indent" has a lot of options, and especially when it comes to comment
528re-formatting you may want to take a look at the man page. But
529remember: "indent" is not a fix for bad programming.
530
531
Robert P. J. Day6754bb42007-05-23 13:57:42 -0700532 Chapter 10: Kconfig configuration files
Linus Torvalds1da177e2005-04-16 15:20:36 -0700533
Robert P. J. Day6754bb42007-05-23 13:57:42 -0700534For all of the Kconfig* configuration files throughout the source tree,
535the indentation is somewhat different. Lines under a "config" definition
536are indented with one tab, while help text is indented an additional two
537spaces. Example:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700538
Robert P. J. Day6754bb42007-05-23 13:57:42 -0700539config AUDIT
540 bool "Auditing support"
541 depends on NET
Linus Torvalds1da177e2005-04-16 15:20:36 -0700542 help
Robert P. J. Day6754bb42007-05-23 13:57:42 -0700543 Enable auditing infrastructure that can be used with another
544 kernel subsystem, such as SELinux (which requires this for
545 logging of avc messages output). Does not do system-call
546 auditing without CONFIG_AUDITSYSCALL.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700547
Robert P. J. Day6754bb42007-05-23 13:57:42 -0700548Features that might still be considered unstable should be defined as
549dependent on "EXPERIMENTAL":
550
551config SLUB
552 depends on EXPERIMENTAL && !ARCH_USES_SLAB_PAGE_STRUCT
553 bool "SLUB (Unqueued Allocator)"
554 ...
555
556while seriously dangerous features (such as write support for certain
557filesystems) should advertise this prominently in their prompt string:
558
559config ADFS_FS_RW
560 bool "ADFS write support (DANGEROUS)"
561 depends on ADFS_FS
562 ...
563
564For full documentation on the configuration files, see the file
565Documentation/kbuild/kconfig-language.txt.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700566
567
Randy Dunlap226a6b82006-06-23 02:05:58 -0700568 Chapter 11: Data structures
Linus Torvalds1da177e2005-04-16 15:20:36 -0700569
570Data structures that have visibility outside the single-threaded
571environment they are created and destroyed in should always have
572reference counts. In the kernel, garbage collection doesn't exist (and
573outside the kernel garbage collection is slow and inefficient), which
574means that you absolutely _have_ to reference count all your uses.
575
576Reference counting means that you can avoid locking, and allows multiple
577users to have access to the data structure in parallel - and not having
578to worry about the structure suddenly going away from under them just
579because they slept or did something else for a while.
580
581Note that locking is _not_ a replacement for reference counting.
582Locking is used to keep data structures coherent, while reference
583counting is a memory management technique. Usually both are needed, and
584they are not to be confused with each other.
585
586Many data structures can indeed have two levels of reference counting,
587when there are users of different "classes". The subclass count counts
588the number of subclass users, and decrements the global count just once
589when the subclass count goes to zero.
590
591Examples of this kind of "multi-level-reference-counting" can be found in
592memory management ("struct mm_struct": mm_users and mm_count), and in
593filesystem code ("struct super_block": s_count and s_active).
594
595Remember: if another thread can find your data structure, and you don't
596have a reference count on it, you almost certainly have a bug.
597
598
Randy Dunlap226a6b82006-06-23 02:05:58 -0700599 Chapter 12: Macros, Enums and RTL
Linus Torvalds1da177e2005-04-16 15:20:36 -0700600
601Names of macros defining constants and labels in enums are capitalized.
602
603#define CONSTANT 0x12345
604
605Enums are preferred when defining several related constants.
606
607CAPITALIZED macro names are appreciated but macros resembling functions
608may be named in lower case.
609
610Generally, inline functions are preferable to macros resembling functions.
611
612Macros with multiple statements should be enclosed in a do - while block:
613
614#define macrofun(a, b, c) \
615 do { \
616 if (a == 5) \
617 do_this(b, c); \
618 } while (0)
619
620Things to avoid when using macros:
621
6221) macros that affect control flow:
623
624#define FOO(x) \
625 do { \
626 if (blah(x) < 0) \
627 return -EBUGGERED; \
628 } while(0)
629
630is a _very_ bad idea. It looks like a function call but exits the "calling"
631function; don't break the internal parsers of those who will read the code.
632
6332) macros that depend on having a local variable with a magic name:
634
635#define FOO(val) bar(index, val)
636
637might look like a good thing, but it's confusing as hell when one reads the
638code and it's prone to breakage from seemingly innocent changes.
639
6403) macros with arguments that are used as l-values: FOO(x) = y; will
641bite you if somebody e.g. turns FOO into an inline function.
642
6434) forgetting about precedence: macros defining constants using expressions
644must enclose the expression in parentheses. Beware of similar issues with
645macros using parameters.
646
647#define CONSTANT 0x4000
648#define CONSTEXP (CONSTANT | 3)
649
650The cpp manual deals with macros exhaustively. The gcc internals manual also
651covers RTL which is used frequently with assembly language in the kernel.
652
653
Randy Dunlap226a6b82006-06-23 02:05:58 -0700654 Chapter 13: Printing kernel messages
Linus Torvalds1da177e2005-04-16 15:20:36 -0700655
656Kernel developers like to be seen as literate. Do mind the spelling
657of kernel messages to make a good impression. Do not use crippled
David Brownell6b094482007-07-13 16:32:09 -0700658words like "dont"; use "do not" or "don't" instead. Make the messages
659concise, clear, and unambiguous.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660
661Kernel messages do not have to be terminated with a period.
662
663Printing numbers in parentheses (%d) adds no value and should be avoided.
664
David Brownell6b094482007-07-13 16:32:09 -0700665There are a number of driver model diagnostic macros in <linux/device.h>
666which you should use to make sure messages are matched to the right device
667and driver, and are tagged with the right level: dev_err(), dev_warn(),
668dev_info(), and so forth. For messages that aren't associated with a
Roland Kammereraa4862c2011-03-15 16:11:52 -0700669particular device, <linux/printk.h> defines pr_debug() and pr_info().
David Brownell6b094482007-07-13 16:32:09 -0700670
671Coming up with good debugging messages can be quite a challenge; and once
672you have them, they can be a huge help for remote troubleshooting. Such
673messages should be compiled out when the DEBUG symbol is not defined (that
674is, by default they are not included). When you use dev_dbg() or pr_debug(),
675that's automatic. Many subsystems have Kconfig options to turn on -DDEBUG.
676A related convention uses VERBOSE_DEBUG to add dev_vdbg() messages to the
677ones already enabled by DEBUG.
678
Linus Torvalds1da177e2005-04-16 15:20:36 -0700679
Randy Dunlap226a6b82006-06-23 02:05:58 -0700680 Chapter 14: Allocating memory
Pekka J Enbergaf4e5a22005-09-16 19:28:11 -0700681
682The kernel provides the following general purpose memory allocators:
Jesper Juhl316b3792011-07-06 11:27:17 -0700683kmalloc(), kzalloc(), kcalloc(), vmalloc(), and vzalloc(). Please refer to
684the API documentation for further information about them.
Pekka J Enbergaf4e5a22005-09-16 19:28:11 -0700685
686The preferred form for passing a size of a struct is the following:
687
688 p = kmalloc(sizeof(*p), ...);
689
690The alternative form where struct name is spelled out hurts readability and
691introduces an opportunity for a bug when the pointer variable type is changed
692but the corresponding sizeof that is passed to a memory allocator is not.
693
694Casting the return value which is a void pointer is redundant. The conversion
695from void pointer to any other pointer type is guaranteed by the C programming
696language.
697
698
Randy Dunlap226a6b82006-06-23 02:05:58 -0700699 Chapter 15: The inline disease
Arjan van de Vena771f2b2006-01-08 01:05:04 -0800700
701There appears to be a common misperception that gcc has a magic "make me
702faster" speedup option called "inline". While the use of inlines can be
Jesper Juhl53ab97a2007-05-08 00:31:06 -0700703appropriate (for example as a means of replacing macros, see Chapter 12), it
Arjan van de Vena771f2b2006-01-08 01:05:04 -0800704very often is not. Abundant use of the inline keyword leads to a much bigger
705kernel, which in turn slows the system as a whole down, due to a bigger
706icache footprint for the CPU and simply because there is less memory
707available for the pagecache. Just think about it; a pagecache miss causes a
Martin Olsson19af5cd2009-04-23 11:37:37 +0200708disk seek, which easily takes 5 milliseconds. There are a LOT of cpu cycles
709that can go into these 5 milliseconds.
Arjan van de Vena771f2b2006-01-08 01:05:04 -0800710
711A reasonable rule of thumb is to not put inline at functions that have more
712than 3 lines of code in them. An exception to this rule are the cases where
713a parameter is known to be a compiletime constant, and as a result of this
714constantness you *know* the compiler will be able to optimize most of your
715function away at compile time. For a good example of this later case, see
716the kmalloc() inline function.
717
718Often people argue that adding inline to functions that are static and used
719only once is always a win since there is no space tradeoff. While this is
720technically correct, gcc is capable of inlining these automatically without
721help, and the maintenance issue of removing the inline when a second user
722appears outweighs the potential value of the hint that tells gcc to do
723something it would have done anyway.
724
725
Alan Sternc16a02d2006-09-29 02:01:21 -0700726 Chapter 16: Function return values and names
727
728Functions can return values of many different kinds, and one of the
729most common is a value indicating whether the function succeeded or
730failed. Such a value can be represented as an error-code integer
731(-Exxx = failure, 0 = success) or a "succeeded" boolean (0 = failure,
732non-zero = success).
733
734Mixing up these two sorts of representations is a fertile source of
735difficult-to-find bugs. If the C language included a strong distinction
736between integers and booleans then the compiler would find these mistakes
737for us... but it doesn't. To help prevent such bugs, always follow this
738convention:
739
740 If the name of a function is an action or an imperative command,
741 the function should return an error-code integer. If the name
742 is a predicate, the function should return a "succeeded" boolean.
743
744For example, "add work" is a command, and the add_work() function returns 0
745for success or -EBUSY for failure. In the same way, "PCI device present" is
746a predicate, and the pci_dev_present() function returns 1 if it succeeds in
747finding a matching device or 0 if it doesn't.
748
749All EXPORTed functions must respect this convention, and so should all
750public functions. Private (static) functions need not, but it is
751recommended that they do.
752
753Functions whose return value is the actual result of a computation, rather
754than an indication of whether the computation succeeded, are not subject to
755this rule. Generally they indicate failure by returning some out-of-range
756result. Typical examples would be functions that return pointers; they use
757NULL or the ERR_PTR mechanism to report failure.
758
759
Robert P. J. Day58637ec2006-12-22 01:09:11 -0800760 Chapter 17: Don't re-invent the kernel macros
761
762The header file include/linux/kernel.h contains a number of macros that
763you should use, rather than explicitly coding some variant of them yourself.
764For example, if you need to calculate the length of an array, take advantage
765of the macro
766
767 #define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
768
769Similarly, if you need to calculate the size of some structure member, use
770
771 #define FIELD_SIZEOF(t, f) (sizeof(((t*)0)->f))
772
773There are also min() and max() macros that do strict type checking if you
774need them. Feel free to peruse that header file to see what else is already
775defined that you shouldn't reproduce in your code.
776
777
Josh Triplett4e7bd662007-07-15 23:41:37 -0700778 Chapter 18: Editor modelines and other cruft
779
780Some editors can interpret configuration information embedded in source files,
781indicated with special markers. For example, emacs interprets lines marked
782like this:
783
784-*- mode: c -*-
785
786Or like this:
787
788/*
789Local Variables:
790compile-command: "gcc -DMAGIC_DEBUG_FLAG foo.c"
791End:
792*/
793
794Vim interprets markers that look like this:
795
796/* vim:set sw=8 noet */
797
798Do not include any of these in source files. People have their own personal
799editor configurations, and your source files should not override them. This
800includes markers for indentation and mode configuration. People may use their
801own custom mode, or may have some other magic method for making indentation
802work correctly.
803
804
Arjan van de Vena771f2b2006-01-08 01:05:04 -0800805
Randy Dunlap226a6b82006-06-23 02:05:58 -0700806 Appendix I: References
Linus Torvalds1da177e2005-04-16 15:20:36 -0700807
808The C Programming Language, Second Edition
809by Brian W. Kernighan and Dennis M. Ritchie.
810Prentice Hall, Inc., 1988.
811ISBN 0-13-110362-8 (paperback), 0-13-110370-9 (hardback).
812URL: http://cm.bell-labs.com/cm/cs/cbook/
813
814The Practice of Programming
815by Brian W. Kernighan and Rob Pike.
816Addison-Wesley, Inc., 1999.
817ISBN 0-201-61586-X.
818URL: http://cm.bell-labs.com/cm/cs/tpop/
819
820GNU manuals - where in compliance with K&R and this text - for cpp, gcc,
Xose Vazquez Perez5b0ed2c2006-01-08 01:02:49 -0800821gcc internals and indent, all available from http://www.gnu.org/manual/
Linus Torvalds1da177e2005-04-16 15:20:36 -0700822
823WG14 is the international standardization working group for the programming
Xose Vazquez Perez5b0ed2c2006-01-08 01:02:49 -0800824language C, URL: http://www.open-std.org/JTC1/SC22/WG14/
825
826Kernel CodingStyle, by greg@kroah.com at OLS 2002:
827http://www.kroah.com/linux/talks/ols_2002_kernel_codingstyle_talk/html/
Linus Torvalds1da177e2005-04-16 15:20:36 -0700828