Re: [PATCH 1/1] docs: kernel-doc: fix parsing of function pointers

[Joe Perches <joe@perches.com>]

On Mon, 2018-09-03 at 15:38 -0600, Jonathan Corbet wrote:
> On Mon, 3 Sep 2018 22:29:00 +0200
> Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> 
> > > >  	    # pointer-to-function
> > > >  	    $arg =~ tr/#/,/;
> > > > -	    $arg =~ m/[^\(]+\(\*?\s*([\w\.]*)\s*\)/;  
> > 
> > m/[^\(]+\(\*?\s*([\w\.]*)\s*\)/;
> >            ^
> > Here we allow for 0..1 asterixes.
> > 
> > If there is no asterix it is not a function pointer. Why should we care
> > for this case?
> 
> GCC seems to allow that asterisk (asterix is an indomitable Gaul :) to be
> missing; not sure if that's officially allowed by the language or not.  I
> also don't know if any code in the kernel elides it,

Many typedefs for function pointers do not use the *
Dunno if there are many others, I didn't look hard.

$ git grep -P '\w+\s*\*?\s*\(\s*\w+\w*\)\s*\(\w+' | \
  grep -w typedef