On Fri, Nov 1, 2013 at 10:00 AM, Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> wrote: > On Fri, Nov 01, 2013 at 09:54:39AM +0100, Daniel Vetter wrote: >> On Fri, Nov 1, 2013 at 9:38 AM, <ville.syrjala@xxxxxxxxxxxxxxx> wrote: >> > From: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> >> > >> > i915 depends on CONFIG_AGP and CONFIG_AGP_INTEL. Since AGP is a thing >> > of the past, it's somewhat likely that a user would not realize that >> > they need to enable those to get i915. So let's just select CONFIG_AGP >> > and CONFIG_AGP_INTEL instead of depending on them. >> > >> > Signed-off-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> >> >> We don't actually need agp, at least for the kms driver. So I'd go the >> other direction and make it fully optional. UMS won't work if someone >> misconfigures their kernel, but meh. > > Don't we still need the AGP backend code for GTT frobbing on older gens? We have the intel-gtt.ko module, which we need on pre-gen6. We could add a hidden config symbol for that which is selected by both i915.ko and the intel agp driver. On i945g we also need the real apg driver in kms mode to keep really old XvMC working. But imo we could switch that to a flexible depency (i.e. only enforce it if CONFIG_AGP_INTEL is defined). Iirc there's 1-2 other places that could be treated similarly. And of course without the AGP driver the ums driver will fail to load, but that's already enforced in the drm core with the AGP_REQUIRE flag. -Daniel -- Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
