Re: General protection fault on rmmod cx8800

[Jean Delvare <khali@linux-fr.org>]

On Tue, 3 Mar 2009 12:14:06 -0800 (PST), Trent Piepho wrote:
> On Tue, 3 Mar 2009, Jean Delvare wrote:
> > On Tue, 3 Mar 2009 01:40:00 -0800 (PST), Trent Piepho wrote:
> > > On Mon, 2 Mar 2009, Jean Delvare wrote:
> > > In 2.6.20 delayed_work was split from work_struct.  The concept of delayed
> > > work was already there and schedule_delayed_work() hasn't changed.  I think
> > > this can also be handled with a compat.h change that defines delayed_work
> > > to work_struct.  That will only be a problem on pre 2.6.20 kernels if some
> > > code decides to define identifiers named work_struct and delayed_work in
> > > the same scope.  There are currently no identifier named delayed_work in
> > > any driver and one driver (sq905) has a structure member named
> > > work_struct.  So I think it'll be ok.
> >
> > Wow, I didn't expect that many different compatibility issues. This
> > goes beyond the time I am ready to spend on it, I'm afraid.
> 
> I already have a patch for compat.h that handles the last remaining issue.
> You don't have to do anything.

Ah, very nice then. Please push it to the v4l-dvb repository so that I
can send my own patch. I will also fix the 3 other drivers that have
the same bug (ir-kbd-i2c, saa6588 and em28xx-input).

-- 
Jean Delvare