linux/Documentation/CodingStyle
<<
>>
Prefs
   1
   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
  38The 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
  60Don'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
  66Don't put multiple assignments on a single line either.  Kernel coding style
  67is super simple.  Avoid tricky expressions.
  68
  69Outside 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
  80The limit on the length of lines is 80 columns and this is a strongly
  81preferred limit.
  82
  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
  86argument 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.
  89
  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
 100                Chapter 3: Placing Braces and Spaces
 101
 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
 112This 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
 126However, 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
 166Do not unnecessarily use braces where a single statement will do.
 167
 168if (condition)
 169        action();
 170
 171This does not apply if one branch of a conditional statement is a single
 172statement. Use braces in both branches.
 173
 174if (condition) {
 175        do_this();
 176        do_that();
 177} else {
 178        otherwise();
 179}
 180
 181                3.1:  Spaces
 182
 183Linux kernel style for use of spaces depends (mostly) on
 184function-versus-keyword usage.  Use a space after (most) keywords.  The
 185notable exceptions are sizeof, typeof, alignof, and __attribute__, which look
 186somewhat like functions (and are usually used with parentheses in Linux,
 187although they are not required in the language, as in: "sizeof info" after
 188"struct fileinfo info;" is declared).
 189
 190So use a space after these keywords:
 191        if, switch, case, for, do, while
 192but not with sizeof, typeof, alignof, or __attribute__.  E.g.,
 193        s = sizeof(struct file);
 194
 195Do not add spaces around (inside) parenthesized expressions.  This example is
 196*bad*:
 197
 198        s = sizeof( struct file );
 199
 200When declaring pointer data or a function that returns a pointer type, the
 201preferred use of '*' is adjacent to the data name or function name and not
 202adjacent to the type name.  Examples:
 203
 204        char *linux_banner;
 205        unsigned long long memparse(char *ptr, char **retptr);
 206        char *match_strdup(substring_t *s);
 207
 208Use one space around (on each side of) most binary and ternary operators,
 209such as any of these:
 210
 211        =  +  -  <  >  *  /  %  |  &  ^  <=  >=  ==  !=  ?  :
 212
 213but no space after unary operators:
 214        &  *  +  -  ~  !  sizeof  typeof  alignof  __attribute__  defined
 215
 216no space before the postfix increment & decrement unary operators:
 217        ++  --
 218
 219no space after the prefix increment & decrement unary operators:
 220        ++  --
 221
 222and no space around the '.' and "->" structure member operators.
 223
 224Do not leave trailing whitespace at the ends of lines.  Some editors with
 225"smart" indentation will insert whitespace at the beginning of new lines as
 226appropriate, so you can start typing the next line of code right away.
 227However, some such editors do not remove the whitespace if you end up not
 228putting a line of code there, such as if you leave a blank line.  As a result,
 229you end up with lines containing trailing whitespace.
 230
 231Git will warn you about patches that introduce trailing whitespace, and can
 232optionally strip the trailing whitespace for you; however, if applying a series
 233of patches, this may make later patches in the series fail by changing their
 234context lines.
 235
 236
 237                Chapter 4: Naming
 238
 239C is a Spartan language, and so should your naming be.  Unlike Modula-2
 240and Pascal programmers, C programmers do not use cute names like
 241ThisVariableIsATemporaryCounter.  A C programmer would call that
 242variable "tmp", which is much easier to write, and not the least more
 243difficult to understand.
 244
 245HOWEVER, while mixed-case names are frowned upon, descriptive names for
 246global variables are a must.  To call a global function "foo" is a
 247shooting offense.
 248
 249GLOBAL variables (to be used only if you _really_ need them) need to
 250have descriptive names, as do global functions.  If you have a function
 251that counts the number of active users, you should call that
 252"count_active_users()" or similar, you should _not_ call it "cntusr()".
 253
 254Encoding the type of a function into the name (so-called Hungarian
 255notation) is brain damaged - the compiler knows the types anyway and can
 256check those, and it only confuses the programmer.  No wonder MicroSoft
 257makes buggy programs.
 258
 259LOCAL variable names should be short, and to the point.  If you have
 260some random integer loop counter, it should probably be called "i".
 261Calling it "loop_counter" is non-productive, if there is no chance of it
 262being mis-understood.  Similarly, "tmp" can be just about any type of
 263variable that is used to hold a temporary value.
 264
 265If you are afraid to mix up your local variable names, you have another
 266problem, which is called the function-growth-hormone-imbalance syndrome.
 267See chapter 6 (Functions).
 268
 269
 270                Chapter 5: Typedefs
 271
 272Please don't use things like "vps_t".
 273
 274It's a _mistake_ to use typedef for structures and pointers. When you see a
 275
 276        vps_t a;
 277
 278in the source, what does it mean?
 279
 280In contrast, if it says
 281
 282        struct virtual_container *a;
 283
 284you can actually tell what "a" is.
 285
 286Lots of people think that typedefs "help readability". Not so. They are
 287useful only for:
 288
 289 (a) totally opaque objects (where the typedef is actively used to _hide_
 290     what the object is).
 291
 292     Example: "pte_t" etc. opaque objects that you can only access using
 293     the proper accessor functions.
 294
 295     NOTE! Opaqueness and "accessor functions" are not good in themselves.
 296     The reason we have them for things like pte_t etc. is that there
 297     really is absolutely _zero_ portably accessible information there.
 298
 299 (b) Clear integer types, where the abstraction _helps_ avoid confusion
 300     whether it is "int" or "long".
 301
 302     u8/u16/u32 are perfectly fine typedefs, although they fit into
 303     category (d) better than here.
 304
 305     NOTE! Again - there needs to be a _reason_ for this. If something is
 306     "unsigned long", then there's no reason to do
 307
 308        typedef unsigned long myflags_t;
 309
 310     but if there is a clear reason for why it under certain circumstances
 311     might be an "unsigned int" and under other configurations might be
 312     "unsigned long", then by all means go ahead and use a typedef.
 313
 314 (c) when you use sparse to literally create a _new_ type for
 315     type-checking.
 316
 317 (d) New types which are identical to standard C99 types, in certain
 318     exceptional circumstances.
 319
 320     Although it would only take a short amount of time for the eyes and
 321     brain to become accustomed to the standard types like 'uint32_t',
 322     some people object to their use anyway.
 323
 324     Therefore, the Linux-specific 'u8/u16/u32/u64' types and their
 325     signed equivalents which are identical to standard types are
 326     permitted -- although they are not mandatory in new code of your
 327     own.
 328
 329     When editing existing code which already uses one or the other set
 330     of types, you should conform to the existing choices in that code.
 331
 332 (e) Types safe for use in userspace.
 333
 334     In certain structures which are visible to userspace, we cannot
 335     require C99 types and cannot use the 'u32' form above. Thus, we
 336     use __u32 and similar types in all structures which are shared
 337     with userspace.
 338
 339Maybe there are other cases too, but the rule should basically be to NEVER
 340EVER use a typedef unless you can clearly match one of those rules.
 341
 342In general, a pointer, or a struct that has elements that can reasonably
 343be directly accessed should _never_ be a typedef.
 344
 345
 346                Chapter 6: Functions
 347
 348Functions should be short and sweet, and do just one thing.  They should
 349fit on one or two screenfuls of text (the ISO/ANSI screen size is 80x24,
 350as we all know), and do one thing and do that well.
 351
 352The maximum length of a function is inversely proportional to the
 353complexity and indentation level of that function.  So, if you have a
 354conceptually simple function that is just one long (but simple)
 355case-statement, where you have to do lots of small things for a lot of
 356different cases, it's OK to have a longer function.
 357
 358However, if you have a complex function, and you suspect that a
 359less-than-gifted first-year high-school student might not even
 360understand what the function is all about, you should adhere to the
 361maximum limits all the more closely.  Use helper functions with
 362descriptive names (you can ask the compiler to in-line them if you think
 363it's performance-critical, and it will probably do a better job of it
 364than you would have done).
 365
 366Another measure of the function is the number of local variables.  They
 367shouldn't exceed 5-10, or you're doing something wrong.  Re-think the
 368function, and split it into smaller pieces.  A human brain can
 369generally easily keep track of about 7 different things, anything more
 370and it gets confused.  You know you're brilliant, but maybe you'd like
 371to understand what you did 2 weeks from now.
 372
 373In source files, separate functions with one blank line.  If the function is
 374exported, the EXPORT* macro for it should follow immediately after the closing
 375function brace line.  E.g.:
 376
 377int system_is_up(void)
 378{
 379        return system_state == SYSTEM_RUNNING;
 380}
 381EXPORT_SYMBOL(system_is_up);
 382
 383In function prototypes, include parameter names with their data types.
 384Although this is not required by the C language, it is preferred in Linux
 385because it is a simple way to add valuable information for the reader.
 386
 387
 388                Chapter 7: Centralized exiting of functions
 389
 390Albeit deprecated by some people, the equivalent of the goto statement is
 391used frequently by compilers in form of the unconditional jump instruction.
 392
 393The goto statement comes in handy when a function exits from multiple
 394locations and some common work such as cleanup has to be done.
 395
 396The rationale is:
 397
 398- unconditional statements are easier to understand and follow
 399- nesting is reduced
 400- errors by not updating individual exit points when making
 401    modifications are prevented
 402- saves the compiler work to optimize redundant code away ;)
 403
 404int fun(int a)
 405{
 406        int result = 0;
 407        char *buffer = kmalloc(SIZE);
 408
 409        if (buffer == NULL)
 410                return -ENOMEM;
 411
 412        if (condition1) {
 413                while (loop1) {
 414                        ...
 415                }
 416                result = 1;
 417                goto out;
 418        }
 419        ...
 420out:
 421        kfree(buffer);
 422        return result;
 423}
 424
 425                Chapter 8: Commenting
 426
 427Comments are good, but there is also a danger of over-commenting.  NEVER
 428try to explain HOW your code works in a comment: it's much better to
 429write the code so that the _working_ is obvious, and it's a waste of
 430time to explain badly written code.
 431
 432Generally, you want your comments to tell WHAT your code does, not HOW.
 433Also, try to avoid putting comments inside a function body: if the
 434function is so complex that you need to separately comment parts of it,
 435you should probably go back to chapter 6 for a while.  You can make
 436small comments to note or warn about something particularly clever (or
 437ugly), but try to avoid excess.  Instead, put the comments at the head
 438of the function, telling people what it does, and possibly WHY it does
 439it.
 440
 441When commenting the kernel API functions, please use the kernel-doc format.
 442See the files Documentation/kernel-doc-nano-HOWTO.txt and scripts/kernel-doc
 443for details.
 444
 445Linux style for comments is the C89 "/* ... */" style.
 446Don't use C99-style "// ..." comments.
 447
 448The preferred style for long (multi-line) comments is:
 449
 450        /*
 451         * This is the preferred style for multi-line
 452         * comments in the Linux kernel source code.
 453         * Please use it consistently.
 454         *
 455         * Description:  A column of asterisks on the left side,
 456         * with beginning and ending almost-blank lines.
 457         */
 458
 459It's also important to comment data, whether they are basic types or derived
 460types.  To this end, use just one data declaration per line (no commas for
 461multiple data declarations).  This leaves you room for a small comment on each
 462item, explaining its use.
 463
 464
 465                Chapter 9: You've made a mess of it
 466
 467That's OK, we all do.  You've probably been told by your long-time Unix
 468user helper that "GNU emacs" automatically formats the C sources for
 469you, and you've noticed that yes, it does do that, but the defaults it
 470uses are less than desirable (in fact, they are worse than random
 471typing - an infinite number of monkeys typing into GNU emacs would never
 472make a good program).
 473
 474So, you can either get rid of GNU emacs, or change it to use saner
 475values.  To do the latter, you can stick the following in your .emacs file:
 476
 477(defun c-lineup-arglist-tabs-only (ignored)
 478  "Line up argument lists by tabs, not spaces"
 479  (let* ((anchor (c-langelem-pos c-syntactic-element))
 480         (column (c-langelem-2nd-pos c-syntactic-element))
 481         (offset (- (1+ column) anchor))
 482         (steps (floor offset c-basic-offset)))
 483    (* (max steps 1)
 484       c-basic-offset)))
 485
 486(add-hook 'c-mode-common-hook
 487          (lambda ()
 488            ;; Add kernel style
 489            (c-add-style
 490             "linux-tabs-only"
 491             '("linux" (c-offsets-alist
 492                        (arglist-cont-nonempty
 493                         c-lineup-gcc-asm-reg
 494                         c-lineup-arglist-tabs-only))))))
 495
 496(add-hook 'c-mode-hook
 497          (lambda ()
 498            (let ((filename (buffer-file-name)))
 499              ;; Enable kernel mode for the appropriate files
 500              (when (and filename
 501                         (string-match (expand-file-name "~/src/linux-trees")
 502                                       filename))
 503                (setq indent-tabs-mode t)
 504                (c-set-style "linux-tabs-only")))))
 505
 506This will make emacs go better with the kernel coding style for C
 507files below ~/src/linux-trees.
 508
 509But even if you fail in getting emacs to do sane formatting, not
 510everything is lost: use "indent".
 511
 512Now, again, GNU indent has the same brain-dead settings that GNU emacs
 513has, which is why you need to give it a few command line options.
 514However, that's not too bad, because even the makers of GNU indent
 515recognize the authority of K&R (the GNU people aren't evil, they are
 516just severely misguided in this matter), so you just give indent the
 517options "-kr -i8" (stands for "K&R, 8 character indents"), or use
 518"scripts/Lindent", which indents in the latest style.
 519
 520"indent" has a lot of options, and especially when it comes to comment
 521re-formatting you may want to take a look at the man page.  But
 522remember: "indent" is not a fix for bad programming.
 523
 524
 525                Chapter 10: Kconfig configuration files
 526
 527For all of the Kconfig* configuration files throughout the source tree,
 528the indentation is somewhat different.  Lines under a "config" definition
 529are indented with one tab, while help text is indented an additional two
 530spaces.  Example:
 531
 532config AUDIT
 533        bool "Auditing support"
 534        depends on NET
 535        help
 536          Enable auditing infrastructure that can be used with another
 537          kernel subsystem, such as SELinux (which requires this for
 538          logging of avc messages output).  Does not do system-call
 539          auditing without CONFIG_AUDITSYSCALL.
 540
 541Features that might still be considered unstable should be defined as
 542dependent on "EXPERIMENTAL":
 543
 544config SLUB
 545        depends on EXPERIMENTAL && !ARCH_USES_SLAB_PAGE_STRUCT
 546        bool "SLUB (Unqueued Allocator)"
 547        ...
 548
 549while seriously dangerous features (such as write support for certain
 550filesystems) should advertise this prominently in their prompt string:
 551
 552config ADFS_FS_RW
 553        bool "ADFS write support (DANGEROUS)"
 554        depends on ADFS_FS
 555        ...
 556
 557For full documentation on the configuration files, see the file
 558Documentation/kbuild/kconfig-language.txt.
 559
 560
 561                Chapter 11: Data structures
 562
 563Data structures that have visibility outside the single-threaded
 564environment they are created and destroyed in should always have
 565reference counts.  In the kernel, garbage collection doesn't exist (and
 566outside the kernel garbage collection is slow and inefficient), which
 567means that you absolutely _have_ to reference count all your uses.
 568
 569Reference counting means that you can avoid locking, and allows multiple
 570users to have access to the data structure in parallel - and not having
 571to worry about the structure suddenly going away from under them just
 572because they slept or did something else for a while.
 573
 574Note that locking is _not_ a replacement for reference counting.
 575Locking is used to keep data structures coherent, while reference
 576counting is a memory management technique.  Usually both are needed, and
 577they are not to be confused with each other.
 578
 579Many data structures can indeed have two levels of reference counting,
 580when there are users of different "classes".  The subclass count counts
 581the number of subclass users, and decrements the global count just once
 582when the subclass count goes to zero.
 583
 584Examples of this kind of "multi-level-reference-counting" can be found in
 585memory management ("struct mm_struct": mm_users and mm_count), and in
 586filesystem code ("struct super_block": s_count and s_active).
 587
 588Remember: if another thread can find your data structure, and you don't
 589have a reference count on it, you almost certainly have a bug.
 590
 591
 592                Chapter 12: Macros, Enums and RTL
 593
 594Names of macros defining constants and labels in enums are capitalized.
 595
 596#define CONSTANT 0x12345
 597
 598Enums are preferred when defining several related constants.
 599
 600CAPITALIZED macro names are appreciated but macros resembling functions
 601may be named in lower case.
 602
 603Generally, inline functions are preferable to macros resembling functions.
 604
 605Macros with multiple statements should be enclosed in a do - while block:
 606
 607#define macrofun(a, b, c)                       \
 608        do {                                    \
 609                if (a == 5)                     \
 610                        do_this(b, c);          \
 611        } while (0)
 612
 613Things to avoid when using macros:
 614
 6151) macros that affect control flow:
 616
 617#define FOO(x)                                  \
 618        do {                                    \
 619                if (blah(x) < 0)                \
 620                        return -EBUGGERED;      \
 621        } while(0)
 622
 623is a _very_ bad idea.  It looks like a function call but exits the "calling"
 624function; don't break the internal parsers of those who will read the code.
 625
 6262) macros that depend on having a local variable with a magic name:
 627
 628#define FOO(val) bar(index, val)
 629
 630might look like a good thing, but it's confusing as hell when one reads the
 631code and it's prone to breakage from seemingly innocent changes.
 632
 6333) macros with arguments that are used as l-values: FOO(x) = y; will
 634bite you if somebody e.g. turns FOO into an inline function.
 635
 6364) forgetting about precedence: macros defining constants using expressions
 637must enclose the expression in parentheses. Beware of similar issues with
 638macros using parameters.
 639
 640#define CONSTANT 0x4000
 641#define CONSTEXP (CONSTANT | 3)
 642
 643The cpp manual deals with macros exhaustively. The gcc internals manual also
 644covers RTL which is used frequently with assembly language in the kernel.
 645
 646
 647                Chapter 13: Printing kernel messages
 648
 649Kernel developers like to be seen as literate. Do mind the spelling
 650of kernel messages to make a good impression. Do not use crippled
 651words like "dont"; use "do not" or "don't" instead.  Make the messages
 652concise, clear, and unambiguous.
 653
 654Kernel messages do not have to be terminated with a period.
 655
 656Printing numbers in parentheses (%d) adds no value and should be avoided.
 657
 658There are a number of driver model diagnostic macros in <linux/device.h>
 659which you should use to make sure messages are matched to the right device
 660and driver, and are tagged with the right level:  dev_err(), dev_warn(),
 661dev_info(), and so forth.  For messages that aren't associated with a
 662particular device, <linux/kernel.h> defines pr_debug() and pr_info().
 663
 664Coming up with good debugging messages can be quite a challenge; and once
 665you have them, they can be a huge help for remote troubleshooting.  Such
 666messages should be compiled out when the DEBUG symbol is not defined (that
 667is, by default they are not included).  When you use dev_dbg() or pr_debug(),
 668that's automatic.  Many subsystems have Kconfig options to turn on -DDEBUG.
 669A related convention uses VERBOSE_DEBUG to add dev_vdbg() messages to the
 670ones already enabled by DEBUG.
 671
 672
 673                Chapter 14: Allocating memory
 674
 675The kernel provides the following general purpose memory allocators:
 676kmalloc(), kzalloc(), kcalloc(), and vmalloc().  Please refer to the API
 677documentation for further information about them.
 678
 679The preferred form for passing a size of a struct is the following:
 680
 681        p = kmalloc(sizeof(*p), ...);
 682
 683The alternative form where struct name is spelled out hurts readability and
 684introduces an opportunity for a bug when the pointer variable type is changed
 685but the corresponding sizeof that is passed to a memory allocator is not.
 686
 687Casting the return value which is a void pointer is redundant. The conversion
 688from void pointer to any other pointer type is guaranteed by the C programming
 689language.
 690
 691
 692                Chapter 15: The inline disease
 693
 694There appears to be a common misperception that gcc has a magic "make me
 695faster" speedup option called "inline". While the use of inlines can be
 696appropriate (for example as a means of replacing macros, see Chapter 12), it
 697very often is not. Abundant use of the inline keyword leads to a much bigger
 698kernel, which in turn slows the system as a whole down, due to a bigger
 699icache footprint for the CPU and simply because there is less memory
 700available for the pagecache. Just think about it; a pagecache miss causes a
 701disk seek, which easily takes 5 milliseconds. There are a LOT of cpu cycles
 702that can go into these 5 milliseconds.
 703
 704A reasonable rule of thumb is to not put inline at functions that have more
 705than 3 lines of code in them. An exception to this rule are the cases where
 706a parameter is known to be a compiletime constant, and as a result of this
 707constantness you *know* the compiler will be able to optimize most of your
 708function away at compile time. For a good example of this later case, see
 709the kmalloc() inline function.
 710
 711Often people argue that adding inline to functions that are static and used
 712only once is always a win since there is no space tradeoff. While this is
 713technically correct, gcc is capable of inlining these automatically without
 714help, and the maintenance issue of removing the inline when a second user
 715appears outweighs the potential value of the hint that tells gcc to do
 716something it would have done anyway.
 717
 718
 719                Chapter 16: Function return values and names
 720
 721Functions can return values of many different kinds, and one of the
 722most common is a value indicating whether the function succeeded or
 723failed.  Such a value can be represented as an error-code integer
 724(-Exxx = failure, 0 = success) or a "succeeded" boolean (0 = failure,
 725non-zero = success).
 726
 727Mixing up these two sorts of representations is a fertile source of
 728difficult-to-find bugs.  If the C language included a strong distinction
 729between integers and booleans then the compiler would find these mistakes
 730for us... but it doesn't.  To help prevent such bugs, always follow this
 731convention:
 732
 733        If the name of a function is an action or an imperative command,
 734        the function should return an error-code integer.  If the name
 735        is a predicate, the function should return a "succeeded" boolean.
 736
 737For example, "add work" is a command, and the add_work() function returns 0
 738for success or -EBUSY for failure.  In the same way, "PCI device present" is
 739a predicate, and the pci_dev_present() function returns 1 if it succeeds in
 740finding a matching device or 0 if it doesn't.
 741
 742All EXPORTed functions must respect this convention, and so should all
 743public functions.  Private (static) functions need not, but it is
 744recommended that they do.
 745
 746Functions whose return value is the actual result of a computation, rather
 747than an indication of whether the computation succeeded, are not subject to
 748this rule.  Generally they indicate failure by returning some out-of-range
 749result.  Typical examples would be functions that return pointers; they use
 750NULL or the ERR_PTR mechanism to report failure.
 751
 752
 753                Chapter 17:  Don't re-invent the kernel macros
 754
 755The header file include/linux/kernel.h contains a number of macros that
 756you should use, rather than explicitly coding some variant of them yourself.
 757For example, if you need to calculate the length of an array, take advantage
 758of the macro
 759
 760  #define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
 761
 762Similarly, if you need to calculate the size of some structure member, use
 763
 764  #define FIELD_SIZEOF(t, f) (sizeof(((t*)0)->f))
 765
 766There are also min() and max() macros that do strict type checking if you
 767need them.  Feel free to peruse that header file to see what else is already
 768defined that you shouldn't reproduce in your code.
 769
 770
 771                Chapter 18:  Editor modelines and other cruft
 772
 773Some editors can interpret configuration information embedded in source files,
 774indicated with special markers.  For example, emacs interprets lines marked
 775like this:
 776
 777-*- mode: c -*-
 778
 779Or like this:
 780
 781/*
 782Local Variables:
 783compile-command: "gcc -DMAGIC_DEBUG_FLAG foo.c"
 784End:
 785*/
 786
 787Vim interprets markers that look like this:
 788
 789/* vim:set sw=8 noet */
 790
 791Do not include any of these in source files.  People have their own personal
 792editor configurations, and your source files should not override them.  This
 793includes markers for indentation and mode configuration.  People may use their
 794own custom mode, or may have some other magic method for making indentation
 795work correctly.
 796
 797
 798
 799                Appendix I: References
 800
 801The C Programming Language, Second Edition
 802by Brian W. Kernighan and Dennis M. Ritchie.
 803Prentice Hall, Inc., 1988.
 804ISBN 0-13-110362-8 (paperback), 0-13-110370-9 (hardback).
 805URL: http://cm.bell-labs.com/cm/cs/cbook/
 806
 807The Practice of Programming
 808by Brian W. Kernighan and Rob Pike.
 809Addison-Wesley, Inc., 1999.
 810ISBN 0-201-61586-X.
 811URL: http://cm.bell-labs.com/cm/cs/tpop/
 812
 813GNU manuals - where in compliance with K&R and this text - for cpp, gcc,
 814gcc internals and indent, all available from http://www.gnu.org/manual/
 815
 816WG14 is the international standardization working group for the programming
 817language C, URL: http://www.open-std.org/JTC1/SC22/WG14/
 818
 819Kernel CodingStyle, by greg@kroah.com at OLS 2002:
 820http://www.kroah.com/linux/talks/ols_2002_kernel_codingstyle_talk/html/
 821
 822--
 823Last updated on 2007-July-13.
 824
 825