[directfb-dev] Re: Bug in gfxcard.c?

[simon.willcocks@t-online.de (Simon Willcocks)]

In message <20031127171003.A14515@sci.fi>
          Ville Syrjälä <syrjala@sci.fi> wrote:

> On Thu, Nov 27, 2003 at 02:48:48PM +0100, Simon Willcocks wrote:
> > The subprocess got started but failed in the call to DirectFB::Create; I
> > tracked this down to the dfb_gfxcard_load_driver routine doing a strcmp of
> > module->name with a NULL valued card->shared->module_name.
> > 
> > A quick and dirty hack to modify the if statement to fail if the module_name
> > is NULL allows the program to continue (display a window, get events, etc.).
> 
> Fixed in 0.9.20.

Apologies, I didn't even notice there was a newer version!  I'll upgrade.

> Your fix is fine but even more simple is to return before
> fusion_list_foreach(). That's how it's done in >=0.9.20.
> 
> > However, the card->shared->module_name field isn't set for any of the
> > cards. which means the dfb_gfxcard_load_driver routine is apparently
> > pointless.
> 
> ??
> 
> It is set by the master if there's a gfxdriver for your card. Master
> uses dfb_gfxcard_find_driver() and slaves use dfb_gfxcard_load_driver().

Well, that's why I said "apparently"!  :-)

I'll have to work out why I've not got an Nvidia driver (but I'm getting a
display, which is good enough for the moment)!

Thanks,
Simon


--
Info:  To unsubscribe send a mail to ecartis@directfb.org with
"unsubscribe directfb-dev" as subject.