On Tue, 10 May 2011 11:45:54 -0700, Eric Anholt <[email protected]> wrote: > On Mon, 9 May 2011 11:42:13 -0700, Jesse Barnes <[email protected]> > wrote: > This feels like a lot of churn to me, for the pre-gen4 chipsets at > least. gen2 is always is_830_display (which is overly specific), and > gen3 is always is_i915_display.
The argument is that we want to be consistent in how we test for capabilities so that adding new feature tests or new chipsets is easier and has no side-effects with existing capabilities. How the macros map on to the actual devices is another matter - a compromise between simplicity and runtime efficiency. If it doesn't reduce our long term maintenance burden, then there is no point making the distinction. I think more fine-grained checks will help in the future, (and consistency across all generations prevents silly bugs). I'm still skeptical that the segregation of capabilities between chipset/display/render is as clear cut as Jesse portrays. ;-) -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
