Re: [PATCH 09/10] misc: fix returning void-valued expression warnings

[David Miller <davem@davemloft.net>]

From: Peter Zijlstra <a.p.zijlstra@chello.nl>
Date: Thu, 01 May 2008 14:17:06 +0200

> On Thu, 2008-05-01 at 13:00 +0100, Al Viro wrote:
> > On Thu, May 01, 2008 at 02:43:50PM +0300, Boaz Harrosh wrote:
> > 
> > > I don't know who invented sparse, but I like this form of return.
> > > 1 - It saves me the curly brackets and extra return line. But mainly
> > > 2 - It is a programing statement that says: "Me here I'm an equivalent 
> > >   to that other call". So if in the future that inner function starts
> > >   to return, say, an error value, with the first style the compiler will
> > >   error. But with the second style the new error return will be silently
> > >   ignored. So these are not equivalent replacements. The former is a much
> > >   stronger bond between the caller and the callie.
> > 
> > 3.  6.8.6.4(1): A return statement with an expression shall not appear in
> > a function whose return type is void.
> > 
> > Write in C, please.
> 
> We use GNU99 all over the place, or are you going to clear up all the
> statement expressions and such other fancy gnu extensions to the
> language as well?
> 
> I'm really not seeing why this would be wrong, other than the standard
> saying it is, ie. I think the standard got it wrong here.
> 
> Harvey can just use -Wno-return-void, or someone can modify sparse to
> have that default disabled for STANDARD_GNU[89]9.

Even Linus thinks this construct is fine and said he had a patch to
make sparse allow it.  This discussion is pointless especially since
GCC has allowed this since basically day one, and it's even self
consistent.