linux kernel coding style and script

Greg KH greg at
Thu Mar 26 09:17:32 EDT 2020

On Thu, Mar 26, 2020 at 04:01:18PM +0300, Konstantin Andreev wrote:
> Valdis Klētnieks, 26 Mar 2020 07:13 MSK:
> > Don't split literal strings, it means that grepping the source tree for "already registered" fails. Making grep for a string work is more important than shutting up checkpatch.
> On Thu, Mar 26, 2020 at 02:36:23PM +0300, Konstantin Andreev wrote:
> > Sic! Grepping is important. Given that, why are kernel functions coded in a
> > 
> > | static int __init loglevel(char *str)
> > | {
> > 
> > way, but not old decent unix way:
> > 
> > | static int __init
> > | loglevel(char *str)
> > | {
> Greg KH, 26 Mar 2020 15:06 MSK:
> > Documentation/process/coding-style.rst
> This document does not answer my question. It does not even require
> | static int __init loglevel(char *str)
> | {
> style. Here is a relevant part of document: "separate functions with one blank line... EXPORT macro should follow closing function brace ... In function prototypes, include parameter names ... E.g.:"
> | int system_is_up(void)
> | {
> |         return system_state == SYSTEM_RUNNING;
> | }
> | EXPORT_SYMBOL(system_is_up);
> That's all. Have I overlooked something? Could you, please, share your own point of view?

If it's not a specific rule in there, then that means you could do
someething crazy like:
	static int __init
	loglevel(char *str)
if you were so loony to do so.

But, it turns out that when you write your code like that, it's harder
to actually find where the function is defined vs. where it is called
when greping a codebase.

That's why those of us who have been at this for a long time prefer:
	static int __init loglevel(char *str)

much easier to pick out of a list of "where is this defined" output of
'git grep'.

And yes, I know all about ctags and the like, but sometimes you don't
have access to that, or don't want to fire it up and have it pre-process
things just to look at a specific git tree at the moment.

Also, big shoutout to 'vgrep' which
makes using 'git grep' a zillion times easier and is what I rely on all
the time.

	either way is fine, but putting the return value on the function
	name line usually makes it easier for others to find your code.

Hope this helps,

greg k-h

More information about the Kernelnewbies mailing list