Re: [GIT PULL v4 for 4.21] META_OUTPUT buffer type and the ipu3 staging driver

[Mauro Carvalho Chehab <mchehab+samsung@kernel.org>]

Em Tue, 8 Jan 2019 00:51:25 +0200
Sakari Ailus <sakari.ailus@iki.fi> escreveu:

> Hi Mauro,
> 
> On Mon, Jan 07, 2019 at 04:11:34PM -0200, Mauro Carvalho Chehab wrote:
> > Em Mon, 7 Jan 2019 16:01:07 -0200
> > Mauro Carvalho Chehab <mchehab+samsung@kernel.org> escreveu:
> >   
> > > Hi Sakari/Bingbu,
> > > 
> > > Em Thu, 13 Dec 2018 14:03:40 +0200
> > > sakari.ailus@iki.fi escreveu:
> > >   
> > > > Hi Mauro,
> > > > 
> > > > Here's the ipu3 staging driver plus the META_OUTPUT buffer type needed to
> > > > pass the parameters for the device. If you think this there's still time to
> > > > get this to 4.21, then please pull. The non-staging patches have been
> > > > around for more than half a year and they're relatively simple.    
> > > 
> > > I'm now getting a lot of new warnings when building it:
> > > 

<snip>

> > > Could you please send ASAP a patch series fixing them?
> > > 
> > > Thanks,
> > > Mauro  
> > 
> > In time, I fixed a few really trivial warnings there, due to the
> > lack of an #include directive.
> > 
> > As this patch is trivial enough, I'll go ahead and just apply it.
> > I'll let the others for you to handle.  
> 
> Interestingly enough, I haven't seen these warnings here. I presume you use
> W=1 when compiling? 

Yes, that's the case. I always build here with:

	W=1 CF=-D__CHECK_ENDIAN__ CONFIG_DEBUG_SECTION_MISMATCH=y

Most of the time, I use ARCH=i386 on my builds, in order to get errors
about div64() stuff (also, gcc warnings seem to work better on x86).

Btw, are you compiling against 5.0-rc1 and using an updated gcc compiler?

Please notice that there were a patchset merged recently changing
the enabled warning flags for W=0 and W=1. Those got merged (I think)
on 5.0-rc1.

Here, As I use Fedora, with gets updated every 6 months, I'm usually
not far behind gcc upstream. Right now (Fedora 29), I'm using gcc
version 8.2.1:

	gcc (GCC) 8.2.1 20181215 (Red Hat 8.2.1-6)

> Some of these are worth more attention than just trying
> to squash them by including the necessary headers. 

Agreed. Several of them are not trivial to solve, yet seem relevant
enough to be fixed.

Thanks,
Mauro