[Mesa-dev] [PATCH 0/8] [RFC] improve driconf support for gallium

[Jose Fonseca]

----- Original Message -----
> On Mit, 2012-04-11 at 08:43 +0200, Michel Dänzer wrote:
> > On Die, 2012-04-10 at 22:04 +0400, Vadim Girlin wrote:
> > > On Tue, 2012-04-10 at 09:56 +0200, Michel Dänzer wrote:
> > > > On Mon, 2012-04-09 at 19:32 +0400, Vadim Girlin wrote:
> > > > > These patches allow to use driver-specific driconf settings,
> > > > > [...]
> > > > 
> > > > How does it allow that? The list of supported driconf options
> > > > is still
> > > > in st/dri, isn't it?
> > > 
> > > Yes, it seems I used the wrong word. The list of options is still
> > > the
> > > same for all gallium drivers.
> > > 
> > > Anyway, my primary goal with these patches is to handle
> > > force_glsl_extensions_warn, to make unigine apps work correctly.
> > > The
> > > idea about pipe_screen::get_driver_name is more a question than a
> > > proposal, probably we can drop it for now.
> > 
> > On second thought, it might still make sense to be able to set
> > different
> > settings for different Gallium drivers in one drirc file. However,
> > I
> > wonder if it wouldn't be better to use the DRI driver name (the
> > output
> > of xdriinfo driver <screen>), i.e. e.g. 'r600' instead of 'r600g'.
> > Isn't
> > there a discrepancy otherwise between the settings shown by the
> > driconf
> > GUI and those actually taking effect?
> 
> Which could probably even be handled by st/dri internally, without
> the
> need for a new driver hook.

Yep. Either pipe_screen::get_driver_name is something that has meaning and use beyond linux/driconf, or that bit of info is better stored somewhere with more knowledge about the platform (e.g., target).

Jose