On Wed, Jun 22, 2016 at 05:21:01PM +0100, Tvrtko Ursulin wrote: > > On 22/06/16 17:10, Chris Wilson wrote: > >On Wed, Jun 22, 2016 at 04:55:51PM +0100, Tvrtko Ursulin wrote: > >>+ BUILD_BUG_ON((1 << RCS) != RENDER_RING); > >>+ BUILD_BUG_ON((1 << BCS) != BLT_RING); > >>+ BUILD_BUG_ON((1 << VCS) != BSD_RING); > >>+ BUILD_BUG_ON((1 << VCS2) != BSD2_RING); > >>+ BUILD_BUG_ON((1 << VECS) != VEBOX_RING); > > > >Heh, isn't that the very definition of those in the header. > >Planning for some array compaction? > > No I was trying to protect against someone changing the definitions > of RENDER_RING & co since the loop below this depends on that. Maybe > it was too paranoid. Or maybe better, I could add HAS_ENGINE(id) and > cement that in one place instead of this many BUILD_BUG_ONs.
Hmm, the logical_rings[] table is ordered by id, so it should be in the same order as the mask. It is probably going to be safer to remove the RENDER_RING et al and replace them with BIT(RCS) when making the masks. -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
