[Mesa-dev] [PATCH] Don't set defaults for gallium and dri drivers

Tom Stellard thomas.stellard at amd.com
Mon Jan 23 09:48:59 PST 2012


On Mon, 2012-01-23 at 12:37 -0500, Matt Turner wrote:
> On Mon, Jan 23, 2012 at 9:01 AM, Brian Paul <brian.e.paul at gmail.com> wrote:
> > On Sat, Jan 21, 2012 at 8:47 PM, Matt Turner <mattst88 at gmail.com> wrote:
> >> If you're building mesa, you know what drivers you want.
> >> ---
> >> I mentioned this on IRC and a couple of people agreed, we'd rather
> >> not have to specify an empty list of drivers to not build either
> >> gallium or dri drivers.
> >
> > I guess I don't follow what this means.  What drivers are build by
> > default if nothing special is specified?
> >
> > -Brian
> 
> For Gallium, r300, r600, and swrast are built.
> 
> For DRI, configure selects what drivers it thinks should be built,
> given the platform. It's an okay idea, but it results in a lot of
> unnecessary code. Packagers surely know not to build i965 on PowerPC,
> for instance. I don't think we need to encode this logic.
> 
> For instance, if I'm working on i965, one would think to configure
> with ./configure --with-dri-drivers=i965, but this actually builds
> i965, Gallium3D, r300g, r600g, and softpipe as well. Similarly if I
> configured with ./configure --with-gallium-drivers=r300, I'd get
> r300g, radeon, r200, i915, i965, nouvuea_vieux, and swrast.
> 

I agree, this behavior is not very friendly to users.  If I build
with ./configure --with-gallium-drivers=r300, then configure will fail
with an error saying I need to install libdrm_intel, which is very
confusing for users trying to build for the first time.

I think Jakob's suggestion to merge the --with-gallium-drivers and
--with-dri-drivers into a single option is a good solution to this
problem.

-Tom


> Some people on IRC agreed that this behavior isn't what they want.
> 
> Matt
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/mesa-dev
> 





More information about the mesa-dev mailing list