Re: [PATCH v3] media: add virtio-media driver

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

Em Sun, 1 Jun 2025 12:01:22 +0200
Ricardo Ribalda <ribalda@chromium.org> escreveu:

> Hi Mauro
> 
> On Sun, 1 Jun 2025 at 11:34, Mauro Carvalho Chehab
> <mchehab+huawei@kernel.org> wrote:
> >
> > Em Wed, 28 May 2025 18:23:02 +0200
> > Ricardo Ribalda <ribalda@chromium.org> escreveu:
> >  
> > > > +static int scatterlist_builder_add_userptr(struct scatterlist_builder *builder,
> > > > +                                          unsigned long userptr,
> > > > +                                          unsigned long length)
> > > > +{
> > > > +       int ret;
> > > > +       int nents;  
> > > Could you initialize nents and sg_list?
> > > old versions of gcc are a bit picky
> > > https://gitlab.freedesktop.org/linux-media/users/ribalda/-/jobs/77042562#L4381  
> >
> > Please don't. In this specific case, ret is always initialized:
> >  
> > > +       struct virtio_media_sg_entry *sg_list;
> > > +
> > > +       ret = __scatterlist_builder_add_userptr(builder, userptr, length,
> > > +                                               &sg_list, &nents);  
> >
> > nents and sg_list may or may not be initialized at the function,
> > but initializing it is wrong, as, when they are not initialized, the
> > ret code shall catch it (and if not, we *do* want gcc to warn).
> >
> > So, if our CI is warning about that due to an old version, please upgrade
> > the version at the CI runner.  
> 
> The main version of gcc works fine. It is the minimal version (8.1) required by
> https://www.kernel.org/doc/html/next/process/changes.html
> that  complains.

Ricardo, 

gcc 8.1 was released in May 2, 2018. I don't think it makes sense to
address bogus warnings with that old gcc versions. I would just disable
WERROR for such versions on our CI tests.

---

Alexandre/Michael,

I need a couple of full days to properly review virtio-media. 
I was planning to do it during this Kernel cycle, but I ended
allocating too much time just to be able to create a crossvm
that would allow testing it. Afterwards, I got sidetracked with other 
issues. I won't be able to review it in time for this merge window.

I'm planning to do it at the beginning of the next merge cycle.

Thanks,
Mauro