CodingStyle 35.6 KB
Newer Older
1 2
.. _codingstyle:

3 4
Linux kernel coding style
=========================
L
Linus Torvalds 已提交
5 6

This is a short document describing the preferred coding style for the
7
linux kernel.  Coding style is very personal, and I won't **force** my
L
Linus Torvalds 已提交
8 9 10 11 12 13 14 15 16 17
views on anybody, but this is what goes for anything that I have to be
able to maintain, and I'd prefer it for most other things too.  Please
at least consider the points made here.

First off, I'd suggest printing out a copy of the GNU coding standards,
and NOT read it.  Burn them, it's a great symbolic gesture.

Anyway, here goes:


18 19
1) Indentation
--------------
L
Linus Torvalds 已提交
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

Tabs are 8 characters, and thus indentations are also 8 characters.
There are heretic movements that try to make indentations 4 (or even 2!)
characters deep, and that is akin to trying to define the value of PI to
be 3.

Rationale: The whole idea behind indentation is to clearly define where
a block of control starts and ends.  Especially when you've been looking
at your screen for 20 straight hours, you'll find it a lot easier to see
how the indentation works if you have large indentations.

Now, some people will claim that having 8-character indentations makes
the code move too far to the right, and makes it hard to read on a
80-character terminal screen.  The answer to that is that if you need
more than 3 levels of indentation, you're screwed anyway, and should fix
your program.

In short, 8-char indents make things easier to read, and have the added
benefit of warning you when you're nesting your functions too deep.
Heed that warning.

R
Randy Dunlap 已提交
41
The preferred way to ease multiple indentation levels in a switch statement is
42 43
to align the ``switch`` and its subordinate ``case`` labels in the same column
instead of ``double-indenting`` the ``case`` labels.  E.g.:
R
Randy Dunlap 已提交
44

45 46
.. code-block:: c

R
Randy Dunlap 已提交
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63
	switch (suffix) {
	case 'G':
	case 'g':
		mem <<= 30;
		break;
	case 'M':
	case 'm':
		mem <<= 20;
		break;
	case 'K':
	case 'k':
		mem <<= 10;
		/* fall through */
	default:
		break;
	}

L
Linus Torvalds 已提交
64 65 66
Don't put multiple statements on a single line unless you have
something to hide:

67 68
.. code-block:: c

L
Linus Torvalds 已提交
69 70 71
	if (condition) do_this;
	  do_something_everytime;

R
Randy Dunlap 已提交
72 73 74
Don't put multiple assignments on a single line either.  Kernel coding style
is super simple.  Avoid tricky expressions.

L
Linus Torvalds 已提交
75 76 77 78 79 80
Outside of comments, documentation and except in Kconfig, spaces are never
used for indentation, and the above example is deliberately broken.

Get a decent editor and don't leave whitespace at the end of lines.


81 82
2) Breaking long lines and strings
----------------------------------
L
Linus Torvalds 已提交
83 84 85 86

Coding style is all about readability and maintainability using commonly
available tools.

A
Alan Cox 已提交
87 88
The limit on the length of lines is 80 columns and this is a strongly
preferred limit.
L
Linus Torvalds 已提交
89

90 91 92 93 94 95 96
Statements longer than 80 columns will be broken into sensible chunks, unless
exceeding 80 columns significantly increases readability and does not hide
information. Descendants are always substantially shorter than the parent and
are placed substantially to the right. The same applies to function headers
with a long argument list. However, never break user-visible strings such as
printk messages, because that breaks the ability to grep for them.

L
Linus Torvalds 已提交
97

98 99
3) Placing Braces and Spaces
----------------------------
L
Linus Torvalds 已提交
100 101 102 103 104 105 106

The other issue that always comes up in C styling is the placement of
braces.  Unlike the indent size, there are few technical reasons to
choose one placement strategy over the other, but the preferred way, as
shown to us by the prophets Kernighan and Ritchie, is to put the opening
brace last on the line, and put the closing brace first, thusly:

107 108
.. code-block:: c

L
Linus Torvalds 已提交
109 110 111 112
	if (x is true) {
		we do y
	}

R
Randy Dunlap 已提交
113 114 115
This applies to all non-function statement blocks (if, switch, for,
while, do).  E.g.:

116 117
.. code-block:: c

R
Randy Dunlap 已提交
118 119 120 121 122 123 124 125 126 127 128
	switch (action) {
	case KOBJ_ADD:
		return "add";
	case KOBJ_REMOVE:
		return "remove";
	case KOBJ_CHANGE:
		return "change";
	default:
		return NULL;
	}

L
Linus Torvalds 已提交
129 130 131
However, there is one special case, namely functions: they have the
opening brace at the beginning of the next line, thus:

132 133
.. code-block:: c

L
Linus Torvalds 已提交
134 135 136 137 138 139 140
	int function(int x)
	{
		body of function
	}

Heretic people all over the world have claimed that this inconsistency
is ...  well ...  inconsistent, but all right-thinking people know that
141
(a) K&R are **right** and (b) K&R are right.  Besides, functions are
L
Linus Torvalds 已提交
142 143
special anyway (you can't nest them in C).

144
Note that the closing brace is empty on a line of its own, **except** in
L
Linus Torvalds 已提交
145
the cases where it is followed by a continuation of the same statement,
146
ie a ``while`` in a do-statement or an ``else`` in an if-statement, like
L
Linus Torvalds 已提交
147 148
this:

149 150
.. code-block:: c

L
Linus Torvalds 已提交
151 152 153 154 155 156
	do {
		body of do-loop
	} while (condition);

and

157 158
.. code-block:: c

L
Linus Torvalds 已提交
159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
	if (x == y) {
		..
	} else if (x > y) {
		...
	} else {
		....
	}

Rationale: K&R.

Also, note that this brace-placement also minimizes the number of empty
(or almost empty) lines, without any loss of readability.  Thus, as the
supply of new-lines on your screen is not a renewable resource (think
25-line terminal screens here), you have more empty lines to put
comments on.

175 176
Do not unnecessarily use braces where a single statement will do.

177 178
.. code-block:: c

179 180
	if (condition)
		action();
181

182 183
and

184 185
.. code-block:: none

186 187 188 189
	if (condition)
		do_this();
	else
		do_that();
190

191 192
This does not apply if only one branch of a conditional statement is a single
statement; in the latter case use braces in both branches:
193

194 195
.. code-block:: c

196 197 198 199 200 201
	if (condition) {
		do_this();
		do_that();
	} else {
		otherwise();
	}
202

203 204
3.1) Spaces
***********
R
Randy Dunlap 已提交
205 206 207 208 209

Linux kernel style for use of spaces depends (mostly) on
function-versus-keyword usage.  Use a space after (most) keywords.  The
notable exceptions are sizeof, typeof, alignof, and __attribute__, which look
somewhat like functions (and are usually used with parentheses in Linux,
210 211
although they are not required in the language, as in: ``sizeof info`` after
``struct fileinfo info;`` is declared).
R
Randy Dunlap 已提交
212

213
So use a space after these keywords::
214

R
Randy Dunlap 已提交
215
	if, switch, case, for, do, while
216

R
Randy Dunlap 已提交
217
but not with sizeof, typeof, alignof, or __attribute__.  E.g.,
218

219 220 221
.. code-block:: c


R
Randy Dunlap 已提交
222 223 224
	s = sizeof(struct file);

Do not add spaces around (inside) parenthesized expressions.  This example is
225 226 227 228
**bad**:

.. code-block:: c

R
Randy Dunlap 已提交
229 230 231 232

	s = sizeof( struct file );

When declaring pointer data or a function that returns a pointer type, the
233
preferred use of ``*`` is adjacent to the data name or function name and not
R
Randy Dunlap 已提交
234 235
adjacent to the type name.  Examples:

236 237 238
.. code-block:: c


R
Randy Dunlap 已提交
239 240 241 242 243
	char *linux_banner;
	unsigned long long memparse(char *ptr, char **retptr);
	char *match_strdup(substring_t *s);

Use one space around (on each side of) most binary and ternary operators,
244
such as any of these::
R
Randy Dunlap 已提交
245 246 247

	=  +  -  <  >  *  /  %  |  &  ^  <=  >=  ==  !=  ?  :

248
but no space after unary operators::
249

R
Randy Dunlap 已提交
250 251
	&  *  +  -  ~  !  sizeof  typeof  alignof  __attribute__  defined

252
no space before the postfix increment & decrement unary operators::
253

R
Randy Dunlap 已提交
254 255
	++  --

256
no space after the prefix increment & decrement unary operators::
257

R
Randy Dunlap 已提交
258 259
	++  --

260
and no space around the ``.`` and ``->`` structure member operators.
R
Randy Dunlap 已提交
261

262
Do not leave trailing whitespace at the ends of lines.  Some editors with
263
``smart`` indentation will insert whitespace at the beginning of new lines as
264 265 266 267 268 269 270 271 272 273
appropriate, so you can start typing the next line of code right away.
However, some such editors do not remove the whitespace if you end up not
putting a line of code there, such as if you leave a blank line.  As a result,
you end up with lines containing trailing whitespace.

Git will warn you about patches that introduce trailing whitespace, and can
optionally strip the trailing whitespace for you; however, if applying a series
of patches, this may make later patches in the series fail by changing their
context lines.

L
Linus Torvalds 已提交
274

275 276
4) Naming
---------
L
Linus Torvalds 已提交
277 278 279 280

C is a Spartan language, and so should your naming be.  Unlike Modula-2
and Pascal programmers, C programmers do not use cute names like
ThisVariableIsATemporaryCounter.  A C programmer would call that
281
variable ``tmp``, which is much easier to write, and not the least more
L
Linus Torvalds 已提交
282 283 284
difficult to understand.

HOWEVER, while mixed-case names are frowned upon, descriptive names for
285
global variables are a must.  To call a global function ``foo`` is a
L
Linus Torvalds 已提交
286 287
shooting offense.

288
GLOBAL variables (to be used only if you **really** need them) need to
L
Linus Torvalds 已提交
289 290
have descriptive names, as do global functions.  If you have a function
that counts the number of active users, you should call that
291
``count_active_users()`` or similar, you should **not** call it ``cntusr()``.
L
Linus Torvalds 已提交
292 293 294 295 296 297 298

Encoding the type of a function into the name (so-called Hungarian
notation) is brain damaged - the compiler knows the types anyway and can
check those, and it only confuses the programmer.  No wonder MicroSoft
makes buggy programs.

LOCAL variable names should be short, and to the point.  If you have
299 300 301
some random integer loop counter, it should probably be called ``i``.
Calling it ``loop_counter`` is non-productive, if there is no chance of it
being mis-understood.  Similarly, ``tmp`` can be just about any type of
L
Linus Torvalds 已提交
302 303 304 305
variable that is used to hold a temporary value.

If you are afraid to mix up your local variable names, you have another
problem, which is called the function-growth-hormone-imbalance syndrome.
R
Randy Dunlap 已提交
306
See chapter 6 (Functions).
L
Linus Torvalds 已提交
307 308


309 310
5) Typedefs
-----------
311

312
Please don't use things like ``vps_t``.
313
It's a **mistake** to use typedef for structures and pointers. When you see a
314

315 316 317
.. code-block:: c


318 319 320 321 322
	vps_t a;

in the source, what does it mean?
In contrast, if it says

323 324
.. code-block:: c

325 326
	struct virtual_container *a;

327
you can actually tell what ``a`` is.
328

329
Lots of people think that typedefs ``help readability``. Not so. They are
330 331
useful only for:

332
 (a) totally opaque objects (where the typedef is actively used to **hide**
333 334
     what the object is).

335
     Example: ``pte_t`` etc. opaque objects that you can only access using
336 337
     the proper accessor functions.

338 339 340 341 342
     .. note::

       Opaqueness and ``accessor functions`` are not good in themselves.
       The reason we have them for things like pte_t etc. is that there
       really is absolutely **zero** portably accessible information there.
343

344
 (b) Clear integer types, where the abstraction **helps** avoid confusion
345
     whether it is ``int`` or ``long``.
346 347 348 349

     u8/u16/u32 are perfectly fine typedefs, although they fit into
     category (d) better than here.

350 351 352 353
     .. note::

       Again - there needs to be a **reason** for this. If something is
       ``unsigned long``, then there's no reason to do
354 355 356 357

	typedef unsigned long myflags_t;

     but if there is a clear reason for why it under certain circumstances
358 359
     might be an ``unsigned int`` and under other configurations might be
     ``unsigned long``, then by all means go ahead and use a typedef.
360

361
 (c) when you use sparse to literally create a **new** type for
362 363 364 365 366 367
     type-checking.

 (d) New types which are identical to standard C99 types, in certain
     exceptional circumstances.

     Although it would only take a short amount of time for the eyes and
368
     brain to become accustomed to the standard types like ``uint32_t``,
369 370
     some people object to their use anyway.

371
     Therefore, the Linux-specific ``u8/u16/u32/u64`` types and their
372 373 374 375 376 377 378 379 380 381
     signed equivalents which are identical to standard types are
     permitted -- although they are not mandatory in new code of your
     own.

     When editing existing code which already uses one or the other set
     of types, you should conform to the existing choices in that code.

 (e) Types safe for use in userspace.

     In certain structures which are visible to userspace, we cannot
382
     require C99 types and cannot use the ``u32`` form above. Thus, we
383 384 385 386 387 388 389
     use __u32 and similar types in all structures which are shared
     with userspace.

Maybe there are other cases too, but the rule should basically be to NEVER
EVER use a typedef unless you can clearly match one of those rules.

In general, a pointer, or a struct that has elements that can reasonably
390
be directly accessed should **never** be a typedef.
391 392


393 394
6) Functions
------------
L
Linus Torvalds 已提交
395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420

Functions should be short and sweet, and do just one thing.  They should
fit on one or two screenfuls of text (the ISO/ANSI screen size is 80x24,
as we all know), and do one thing and do that well.

The maximum length of a function is inversely proportional to the
complexity and indentation level of that function.  So, if you have a
conceptually simple function that is just one long (but simple)
case-statement, where you have to do lots of small things for a lot of
different cases, it's OK to have a longer function.

However, if you have a complex function, and you suspect that a
less-than-gifted first-year high-school student might not even
understand what the function is all about, you should adhere to the
maximum limits all the more closely.  Use helper functions with
descriptive names (you can ask the compiler to in-line them if you think
it's performance-critical, and it will probably do a better job of it
than you would have done).

Another measure of the function is the number of local variables.  They
shouldn't exceed 5-10, or you're doing something wrong.  Re-think the
function, and split it into smaller pieces.  A human brain can
generally easily keep track of about 7 different things, anything more
and it gets confused.  You know you're brilliant, but maybe you'd like
to understand what you did 2 weeks from now.

R
Randy Dunlap 已提交
421
In source files, separate functions with one blank line.  If the function is
422 423 424 425
exported, the **EXPORT** macro for it should follow immediately after the
closing function brace line.  E.g.:

.. code-block:: c
R
Randy Dunlap 已提交
426

427 428 429 430 431
	int system_is_up(void)
	{
		return system_state == SYSTEM_RUNNING;
	}
	EXPORT_SYMBOL(system_is_up);
R
Randy Dunlap 已提交
432 433 434 435 436

In function prototypes, include parameter names with their data types.
Although this is not required by the C language, it is preferred in Linux
because it is a simple way to add valuable information for the reader.

L
Linus Torvalds 已提交
437

438 439
7) Centralized exiting of functions
-----------------------------------
L
Linus Torvalds 已提交
440 441 442 443 444

Albeit deprecated by some people, the equivalent of the goto statement is
used frequently by compilers in form of the unconditional jump instruction.

The goto statement comes in handy when a function exits from multiple
445 446
locations and some common work such as cleanup has to be done.  If there is no
cleanup needed then just return directly.
L
Linus Torvalds 已提交
447

448
Choose label names which say what the goto does or why the goto exists.  An
449 450
example of a good name could be ``out_free_buffer:`` if the goto frees ``buffer``.
Avoid using GW-BASIC names like ``err1:`` and ``err2:``, as you would have to
451 452 453 454
renumber them if you ever add or remove exit paths, and they make correctness
difficult to verify anyway.

It is advised to indent labels with a single space (not tab), so that
455
``diff -p`` does not confuse labels with functions.
456 457

The rationale for using gotos is:
L
Linus Torvalds 已提交
458 459 460 461

- unconditional statements are easier to understand and follow
- nesting is reduced
- errors by not updating individual exit points when making
462
  modifications are prevented
L
Linus Torvalds 已提交
463 464
- saves the compiler work to optimize redundant code away ;)

465 466
.. code-block:: c

467 468 469 470 471 472 473 474 475 476 477 478 479 480 481
	int fun(int a)
	{
		int result = 0;
		char *buffer;

		buffer = kmalloc(SIZE, GFP_KERNEL);
		if (!buffer)
			return -ENOMEM;

		if (condition1) {
			while (loop1) {
				...
			}
			result = 1;
			goto out_buffer;
L
Linus Torvalds 已提交
482
		}
483
		...
484
	 out_free_buffer:
485 486
		kfree(buffer);
		return result;
L
Linus Torvalds 已提交
487 488
	}

489
A common type of bug to be aware of is ``one err bugs`` which look like this:
490

491 492
.. code-block:: c

493
	 err:
494 495 496
		kfree(foo->bar);
		kfree(foo);
		return ret;
497

498 499 500
The bug in this code is that on some exit paths ``foo`` is NULL.  Normally the
fix for this is to split it up into two error labels ``err_free_bar:`` and
``err_free_foo:``:
501

502 503
.. code-block:: c

504 505 506 507 508 509 510
	 err_free_bar:
		kfree(foo->bar);
	 err_free_foo:
		kfree(foo);
		return ret;

Ideally you should simulate errors to test all exit paths.
511 512


513 514
8) Commenting
-------------
L
Linus Torvalds 已提交
515 516 517

Comments are good, but there is also a danger of over-commenting.  NEVER
try to explain HOW your code works in a comment: it's much better to
518
write the code so that the **working** is obvious, and it's a waste of
L
Linus Torvalds 已提交
519 520 521 522 523
time to explain badly written code.

Generally, you want your comments to tell WHAT your code does, not HOW.
Also, try to avoid putting comments inside a function body: if the
function is so complex that you need to separately comment parts of it,
R
Randy Dunlap 已提交
524
you should probably go back to chapter 6 for a while.  You can make
L
Linus Torvalds 已提交
525 526 527 528 529
small comments to note or warn about something particularly clever (or
ugly), but try to avoid excess.  Instead, put the comments at the head
of the function, telling people what it does, and possibly WHY it does
it.

R
Randy Dunlap 已提交
530
When commenting the kernel API functions, please use the kernel-doc format.
531
See the files Documentation/kernel-documentation.rst and scripts/kernel-doc
532
for details.
L
Linus Torvalds 已提交
533

R
Randy Dunlap 已提交
534 535
The preferred style for long (multi-line) comments is:

536 537
.. code-block:: c

R
Randy Dunlap 已提交
538 539 540 541 542 543 544 545 546
	/*
	 * This is the preferred style for multi-line
	 * comments in the Linux kernel source code.
	 * Please use it consistently.
	 *
	 * Description:  A column of asterisks on the left side,
	 * with beginning and ending almost-blank lines.
	 */

547 548 549
For files in net/ and drivers/net/ the preferred style for long (multi-line)
comments is a little different.

550 551
.. code-block:: c

552 553 554 555 556 557 558
	/* The preferred comment style for files in net/ and drivers/net
	 * looks like this.
	 *
	 * It is nearly the same as the generally preferred comment style,
	 * but there is no initial almost-blank line.
	 */

R
Randy Dunlap 已提交
559 560 561 562 563 564
It's also important to comment data, whether they are basic types or derived
types.  To this end, use just one data declaration per line (no commas for
multiple data declarations).  This leaves you room for a small comment on each
item, explaining its use.


565 566
9) You've made a mess of it
---------------------------
L
Linus Torvalds 已提交
567 568

That's OK, we all do.  You've probably been told by your long-time Unix
569
user helper that ``GNU emacs`` automatically formats the C sources for
L
Linus Torvalds 已提交
570 571 572 573 574 575 576 577
you, and you've noticed that yes, it does do that, but the defaults it
uses are less than desirable (in fact, they are worse than random
typing - an infinite number of monkeys typing into GNU emacs would never
make a good program).

So, you can either get rid of GNU emacs, or change it to use saner
values.  To do the latter, you can stick the following in your .emacs file:

578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608
.. code-block:: none

  (defun c-lineup-arglist-tabs-only (ignored)
    "Line up argument lists by tabs, not spaces"
    (let* ((anchor (c-langelem-pos c-syntactic-element))
           (column (c-langelem-2nd-pos c-syntactic-element))
           (offset (- (1+ column) anchor))
           (steps (floor offset c-basic-offset)))
      (* (max steps 1)
         c-basic-offset)))

  (add-hook 'c-mode-common-hook
            (lambda ()
              ;; Add kernel style
              (c-add-style
               "linux-tabs-only"
               '("linux" (c-offsets-alist
                          (arglist-cont-nonempty
                           c-lineup-gcc-asm-reg
                           c-lineup-arglist-tabs-only))))))

  (add-hook 'c-mode-hook
            (lambda ()
              (let ((filename (buffer-file-name)))
                ;; Enable kernel mode for the appropriate files
                (when (and filename
                           (string-match (expand-file-name "~/src/linux-trees")
                                         filename))
                  (setq indent-tabs-mode t)
                  (setq show-trailing-whitespace t)
                  (c-set-style "linux-tabs-only")))))
609 610

This will make emacs go better with the kernel coding style for C
611
files below ``~/src/linux-trees``.
L
Linus Torvalds 已提交
612 613

But even if you fail in getting emacs to do sane formatting, not
614
everything is lost: use ``indent``.
L
Linus Torvalds 已提交
615 616 617 618 619 620

Now, again, GNU indent has the same brain-dead settings that GNU emacs
has, which is why you need to give it a few command line options.
However, that's not too bad, because even the makers of GNU indent
recognize the authority of K&R (the GNU people aren't evil, they are
just severely misguided in this matter), so you just give indent the
621 622
options ``-kr -i8`` (stands for ``K&R, 8 character indents``), or use
``scripts/Lindent``, which indents in the latest style.
L
Linus Torvalds 已提交
623

624
``indent`` has a lot of options, and especially when it comes to comment
L
Linus Torvalds 已提交
625
re-formatting you may want to take a look at the man page.  But
626
remember: ``indent`` is not a fix for bad programming.
L
Linus Torvalds 已提交
627 628


629 630
10) Kconfig configuration files
-------------------------------
L
Linus Torvalds 已提交
631

632
For all of the Kconfig* configuration files throughout the source tree,
633
the indentation is somewhat different.  Lines under a ``config`` definition
634
are indented with one tab, while help text is indented an additional two
635
spaces.  Example::
L
Linus Torvalds 已提交
636

637
  config AUDIT
638 639
	bool "Auditing support"
	depends on NET
L
Linus Torvalds 已提交
640
	help
641 642 643 644 645
	  Enable auditing infrastructure that can be used with another
	  kernel subsystem, such as SELinux (which requires this for
	  logging of avc messages output).  Does not do system-call
	  auditing without CONFIG_AUDITSYSCALL.

646
Seriously dangerous features (such as write support for certain
647
filesystems) should advertise this prominently in their prompt string::
648

649
  config ADFS_FS_RW
650 651 652
	bool "ADFS write support (DANGEROUS)"
	depends on ADFS_FS
	...
L
Linus Torvalds 已提交
653

654 655
For full documentation on the configuration files, see the file
Documentation/kbuild/kconfig-language.txt.
L
Linus Torvalds 已提交
656 657


658 659
11) Data structures
-------------------
L
Linus Torvalds 已提交
660 661 662 663 664

Data structures that have visibility outside the single-threaded
environment they are created and destroyed in should always have
reference counts.  In the kernel, garbage collection doesn't exist (and
outside the kernel garbage collection is slow and inefficient), which
665
means that you absolutely **have** to reference count all your uses.
L
Linus Torvalds 已提交
666 667 668 669 670 671

Reference counting means that you can avoid locking, and allows multiple
users to have access to the data structure in parallel - and not having
to worry about the structure suddenly going away from under them just
because they slept or did something else for a while.

672
Note that locking is **not** a replacement for reference counting.
L
Linus Torvalds 已提交
673 674 675 676 677
Locking is used to keep data structures coherent, while reference
counting is a memory management technique.  Usually both are needed, and
they are not to be confused with each other.

Many data structures can indeed have two levels of reference counting,
678
when there are users of different ``classes``.  The subclass count counts
L
Linus Torvalds 已提交
679 680 681
the number of subclass users, and decrements the global count just once
when the subclass count goes to zero.

682 683 684
Examples of this kind of ``multi-level-reference-counting`` can be found in
memory management (``struct mm_struct``: mm_users and mm_count), and in
filesystem code (``struct super_block``: s_count and s_active).
L
Linus Torvalds 已提交
685 686 687 688 689

Remember: if another thread can find your data structure, and you don't
have a reference count on it, you almost certainly have a bug.


690 691
12) Macros, Enums and RTL
-------------------------
L
Linus Torvalds 已提交
692 693 694

Names of macros defining constants and labels in enums are capitalized.

695 696
.. code-block:: c

697
	#define CONSTANT 0x12345
L
Linus Torvalds 已提交
698 699 700 701 702 703 704 705 706 707

Enums are preferred when defining several related constants.

CAPITALIZED macro names are appreciated but macros resembling functions
may be named in lower case.

Generally, inline functions are preferable to macros resembling functions.

Macros with multiple statements should be enclosed in a do - while block:

708 709 710
.. code-block:: c

	#define macrofun(a, b, c)			\
711 712 713 714
		do {					\
			if (a == 5)			\
				do_this(b, c);		\
		} while (0)
L
Linus Torvalds 已提交
715 716 717 718 719

Things to avoid when using macros:

1) macros that affect control flow:

720 721
.. code-block:: c

722 723 724 725
	#define FOO(x)					\
		do {					\
			if (blah(x) < 0)		\
				return -EBUGGERED;	\
726
		} while (0)
L
Linus Torvalds 已提交
727

728
is a **very** bad idea.  It looks like a function call but exits the ``calling``
L
Linus Torvalds 已提交
729 730 731 732
function; don't break the internal parsers of those who will read the code.

2) macros that depend on having a local variable with a magic name:

733 734
.. code-block:: c

735
	#define FOO(val) bar(index, val)
L
Linus Torvalds 已提交
736 737 738 739 740 741 742 743 744 745 746

might look like a good thing, but it's confusing as hell when one reads the
code and it's prone to breakage from seemingly innocent changes.

3) macros with arguments that are used as l-values: FOO(x) = y; will
bite you if somebody e.g. turns FOO into an inline function.

4) forgetting about precedence: macros defining constants using expressions
must enclose the expression in parentheses. Beware of similar issues with
macros using parameters.

747 748
.. code-block:: c

749 750
	#define CONSTANT 0x4000
	#define CONSTEXP (CONSTANT | 3)
L
Linus Torvalds 已提交
751

752 753 754
5) namespace collisions when defining local variables in macros resembling
functions:

755 756 757 758 759 760 761 762
.. code-block:: c

	#define FOO(x)				\
	({					\
		typeof(x) ret;			\
		ret = calc_ret(x);		\
		(ret);				\
	})
763 764 765 766

ret is a common name for a local variable - __foo_ret is less likely
to collide with an existing variable.

L
Linus Torvalds 已提交
767 768 769 770
The cpp manual deals with macros exhaustively. The gcc internals manual also
covers RTL which is used frequently with assembly language in the kernel.


771 772
13) Printing kernel messages
----------------------------
L
Linus Torvalds 已提交
773 774 775

Kernel developers like to be seen as literate. Do mind the spelling
of kernel messages to make a good impression. Do not use crippled
776
words like ``dont``; use ``do not`` or ``don't`` instead.  Make the messages
D
David Brownell 已提交
777
concise, clear, and unambiguous.
L
Linus Torvalds 已提交
778 779 780 781 782

Kernel messages do not have to be terminated with a period.

Printing numbers in parentheses (%d) adds no value and should be avoided.

D
David Brownell 已提交
783 784 785 786
There are a number of driver model diagnostic macros in <linux/device.h>
which you should use to make sure messages are matched to the right device
and driver, and are tagged with the right level:  dev_err(), dev_warn(),
dev_info(), and so forth.  For messages that aren't associated with a
787 788
particular device, <linux/printk.h> defines pr_notice(), pr_info(),
pr_warn(), pr_err(), etc.
D
David Brownell 已提交
789 790

Coming up with good debugging messages can be quite a challenge; and once
791 792 793 794 795 796 797 798 799 800 801
you have them, they can be a huge help for remote troubleshooting.  However
debug message printing is handled differently than printing other non-debug
messages.  While the other pr_XXX() functions print unconditionally,
pr_debug() does not; it is compiled out by default, unless either DEBUG is
defined or CONFIG_DYNAMIC_DEBUG is set.  That is true for dev_dbg() also,
and a related convention uses VERBOSE_DEBUG to add dev_vdbg() messages to
the ones already enabled by DEBUG.

Many subsystems have Kconfig debug options to turn on -DDEBUG in the
corresponding Makefile; in other cases specific files #define DEBUG.  And
when a debug message should be unconditionally printed, such as if it is
802
already inside a debug-related #ifdef section, printk(KERN_DEBUG ...) can be
803
used.
D
David Brownell 已提交
804

L
Linus Torvalds 已提交
805

806 807
14) Allocating memory
---------------------
808 809

The kernel provides the following general purpose memory allocators:
810 811 812
kmalloc(), kzalloc(), kmalloc_array(), kcalloc(), vmalloc(), and
vzalloc().  Please refer to the API documentation for further information
about them.
813 814 815

The preferred form for passing a size of a struct is the following:

816 817
.. code-block:: c

818 819 820 821 822 823 824 825 826 827
	p = kmalloc(sizeof(*p), ...);

The alternative form where struct name is spelled out hurts readability and
introduces an opportunity for a bug when the pointer variable type is changed
but the corresponding sizeof that is passed to a memory allocator is not.

Casting the return value which is a void pointer is redundant. The conversion
from void pointer to any other pointer type is guaranteed by the C programming
language.

828 829
The preferred form for allocating an array is the following:

830 831
.. code-block:: c

832 833 834 835
	p = kmalloc_array(n, sizeof(...), ...);

The preferred form for allocating a zeroed array is the following:

836 837
.. code-block:: c

838 839 840 841 842
	p = kcalloc(n, sizeof(...), ...);

Both forms check for overflow on the allocation size n * sizeof(...),
and return NULL if that occurred.

843

844 845
15) The inline disease
----------------------
846 847

There appears to be a common misperception that gcc has a magic "make me
848
faster" speedup option called ``inline``. While the use of inlines can be
849
appropriate (for example as a means of replacing macros, see Chapter 12), it
850 851 852 853
very often is not. Abundant use of the inline keyword leads to a much bigger
kernel, which in turn slows the system as a whole down, due to a bigger
icache footprint for the CPU and simply because there is less memory
available for the pagecache. Just think about it; a pagecache miss causes a
854 855
disk seek, which easily takes 5 milliseconds. There are a LOT of cpu cycles
that can go into these 5 milliseconds.
856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871

A reasonable rule of thumb is to not put inline at functions that have more
than 3 lines of code in them. An exception to this rule are the cases where
a parameter is known to be a compiletime constant, and as a result of this
constantness you *know* the compiler will be able to optimize most of your
function away at compile time. For a good example of this later case, see
the kmalloc() inline function.

Often people argue that adding inline to functions that are static and used
only once is always a win since there is no space tradeoff. While this is
technically correct, gcc is capable of inlining these automatically without
help, and the maintenance issue of removing the inline when a second user
appears outweighs the potential value of the hint that tells gcc to do
something it would have done anyway.


872 873
16) Function return values and names
------------------------------------
874 875 876 877

Functions can return values of many different kinds, and one of the
most common is a value indicating whether the function succeeded or
failed.  Such a value can be represented as an error-code integer
878
(-Exxx = failure, 0 = success) or a ``succeeded`` boolean (0 = failure,
879 880 881 882 883 884
non-zero = success).

Mixing up these two sorts of representations is a fertile source of
difficult-to-find bugs.  If the C language included a strong distinction
between integers and booleans then the compiler would find these mistakes
for us... but it doesn't.  To help prevent such bugs, always follow this
885
convention::
886 887 888 889 890

	If the name of a function is an action or an imperative command,
	the function should return an error-code integer.  If the name
	is a predicate, the function should return a "succeeded" boolean.

891 892
For example, ``add work`` is a command, and the add_work() function returns 0
for success or -EBUSY for failure.  In the same way, ``PCI device present`` is
893 894 895 896 897 898 899 900 901 902 903 904 905 906
a predicate, and the pci_dev_present() function returns 1 if it succeeds in
finding a matching device or 0 if it doesn't.

All EXPORTed functions must respect this convention, and so should all
public functions.  Private (static) functions need not, but it is
recommended that they do.

Functions whose return value is the actual result of a computation, rather
than an indication of whether the computation succeeded, are not subject to
this rule.  Generally they indicate failure by returning some out-of-range
result.  Typical examples would be functions that return pointers; they use
NULL or the ERR_PTR mechanism to report failure.


907 908
17) Don't re-invent the kernel macros
-------------------------------------
909 910 911 912 913 914

The header file include/linux/kernel.h contains a number of macros that
you should use, rather than explicitly coding some variant of them yourself.
For example, if you need to calculate the length of an array, take advantage
of the macro

915 916
.. code-block:: c

917
	#define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
918 919 920

Similarly, if you need to calculate the size of some structure member, use

921 922
.. code-block:: c

923
	#define FIELD_SIZEOF(t, f) (sizeof(((t*)0)->f))
924 925 926 927 928 929

There are also min() and max() macros that do strict type checking if you
need them.  Feel free to peruse that header file to see what else is already
defined that you shouldn't reproduce in your code.


930 931
18) Editor modelines and other cruft
------------------------------------
932 933 934 935 936

Some editors can interpret configuration information embedded in source files,
indicated with special markers.  For example, emacs interprets lines marked
like this:

937 938
.. code-block:: c

939
	-*- mode: c -*-
940 941 942

Or like this:

943 944
.. code-block:: c

945 946 947 948 949
	/*
	Local Variables:
	compile-command: "gcc -DMAGIC_DEBUG_FLAG foo.c"
	End:
	*/
950 951 952

Vim interprets markers that look like this:

953 954
.. code-block:: c

955
	/* vim:set sw=8 noet */
956 957 958 959 960 961 962 963

Do not include any of these in source files.  People have their own personal
editor configurations, and your source files should not override them.  This
includes markers for indentation and mode configuration.  People may use their
own custom mode, or may have some other magic method for making indentation
work correctly.


964 965
19) Inline assembly
-------------------
966 967 968 969 970 971 972 973 974 975 976 977

In architecture-specific code, you may need to use inline assembly to interface
with CPU or platform functionality.  Don't hesitate to do so when necessary.
However, don't use inline assembly gratuitously when C can do the job.  You can
and should poke hardware from C when possible.

Consider writing simple helper functions that wrap common bits of inline
assembly, rather than repeatedly writing them with slight variations.  Remember
that inline assembly can use C parameters.

Large, non-trivial assembly functions should go in .S files, with corresponding
C prototypes defined in C header files.  The C prototypes for assembly
978
functions should use ``asmlinkage``.
979 980 981 982 983 984 985 986 987 988

You may need to mark your asm statement as volatile, to prevent GCC from
removing it if GCC doesn't notice any side effects.  You don't always need to
do so, though, and doing so unnecessarily can limit optimization.

When writing a single inline assembly statement containing multiple
instructions, put each instruction on a separate line in a separate quoted
string, and end each string except the last with \n\t to properly indent the
next instruction in the assembly output:

989 990
.. code-block:: c

991 992 993 994 995
	asm ("magic %reg1, #42\n\t"
	     "more_magic %reg2, %reg3"
	     : /* outputs */ : /* inputs */ : /* clobbers */);


996 997
20) Conditional Compilation
---------------------------
998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020

Wherever possible, don't use preprocessor conditionals (#if, #ifdef) in .c
files; doing so makes code harder to read and logic harder to follow.  Instead,
use such conditionals in a header file defining functions for use in those .c
files, providing no-op stub versions in the #else case, and then call those
functions unconditionally from .c files.  The compiler will avoid generating
any code for the stub calls, producing identical results, but the logic will
remain easy to follow.

Prefer to compile out entire functions, rather than portions of functions or
portions of expressions.  Rather than putting an ifdef in an expression, factor
out part or all of the expression into a separate helper function and apply the
conditional to that function.

If you have a function or variable which may potentially go unused in a
particular configuration, and the compiler would warn about its definition
going unused, mark the definition as __maybe_unused rather than wrapping it in
a preprocessor conditional.  (However, if a function or variable *always* goes
unused, delete it.)

Within code, where possible, use the IS_ENABLED macro to convert a Kconfig
symbol into a C boolean expression, and use it in a normal C conditional:

1021 1022
.. code-block:: c

1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037
	if (IS_ENABLED(CONFIG_SOMETHING)) {
		...
	}

The compiler will constant-fold the conditional away, and include or exclude
the block of code just as with an #ifdef, so this will not add any runtime
overhead.  However, this approach still allows the C compiler to see the code
inside the block, and check it for correctness (syntax, types, symbol
references, etc).  Thus, you still have to use an #ifdef if the code inside the
block references symbols that will not exist if the condition is not met.

At the end of any non-trivial #if or #ifdef block (more than a few lines),
place a comment after the #endif on the same line, noting the conditional
expression used.  For instance:

1038 1039
.. code-block:: c

1040 1041 1042
	#ifdef CONFIG_SOMETHING
	...
	#endif /* CONFIG_SOMETHING */
1043

1044

1045 1046
Appendix I) References
----------------------
L
Linus Torvalds 已提交
1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058

The C Programming Language, Second Edition
by Brian W. Kernighan and Dennis M. Ritchie.
Prentice Hall, Inc., 1988.
ISBN 0-13-110362-8 (paperback), 0-13-110370-9 (hardback).

The Practice of Programming
by Brian W. Kernighan and Rob Pike.
Addison-Wesley, Inc., 1999.
ISBN 0-201-61586-X.

GNU manuals - where in compliance with K&R and this text - for cpp, gcc,
1059
gcc internals and indent, all available from http://www.gnu.org/manual/
L
Linus Torvalds 已提交
1060 1061

WG14 is the international standardization working group for the programming
1062 1063 1064 1065
language C, URL: http://www.open-std.org/JTC1/SC22/WG14/

Kernel CodingStyle, by greg@kroah.com at OLS 2002:
http://www.kroah.com/linux/talks/ols_2002_kernel_codingstyle_talk/html/