Re: [v4l-dvb-maintainer] [2.6 patch] dvb_frontend_ioctl(): fix check-after-use

[Trent Piepho <xyzzy@speakeasy.org>]

On Fri, 10 Aug 2007, Markus Rechberger wrote:
> On 8/1/07, Manu Abraham <abraham.manu@gmail.com> wrote:
> > On 7/31/07, Adrian Bunk <bunk@stusta.de> wrote:
> > > The Coverity checker spotted that we have already oops'ed if "fe" was
> > NULL.
> > >
> > > --- linux-2.6.23-rc1-mm1/drivers/media/dvb/dvb-core/dvb_frontend.c.old
> > > +++ linux-2.6.23-rc1-mm1/drivers/media/dvb/dvb-core/dvb_frontend.c
> > > @@ -706,11 +706,11 @@ static int dvb_frontend_ioctl(struct ino
> > > -       if (!fe || fepriv->exit)
> > > +       if (fepriv->exit)
> > >                 return -ENODEV;
>
> This issue has been known for a while including some other problems at
> that part.
>
> http://article.gmane.org/gmane.linux.drivers.dvb/35351/match=patch+dvb_net+hotplugging+support
>
> this includes a link where this and more got discussed in May.

For dvb_net_close, I like the patch I already posted better.  To fix the
check-after-use, it's not "use" part that's the problem, it's the "check" part
that isn't necessary.

I traced the dvb-net code, http://article.gmane.org/gmane.linux.kernel/543689,
and I'm sure that dvbdev can't be NULL.

My patch also deletes a few pieces of duplicated code by calling
dvb_generic_release().

The only problem is that practically no one uses dvb-net, so it's very hard to
test these patches.

In all the dvb code, were is the locking for device open and release?  I don't
see it.  What is preventing two threads from trying to open and/or close the
same dvb device at the same time?