On Fri, Nov 21, 2014 at 07:01:54PM +0000, Dave Gordon wrote: > On 20/11/14 08:45, Daniel Vetter wrote: > > We need to do that every time we resume the rings, not just at load. > > I've overlooked this in my untangling of the ring init code. > > Hi Daniel, > > another thing that needs untangling in the general maze of init code is > the initialisation of the active and request lists -- Thomas Daniel's > complaint about 11/28 of the s/seqno/request/ patchset was essentially > because John was adding more lists that appear to be redundantly > initialised in multiple places. Please see my followup at > http://lists.freedesktop.org/archives/intel-gfx/2014-November/055856.html > > As it looks like you're getting rid of intel_render_ring_init_dri(), if > we could also resolve whether init_ring_lists() is also now redundant, > that would mean there were no duplicated list initialisations :) Actually I wanted to feature my little series here in your thread as one step closer to untangling this stuff too, so ... care to review (except for the one patch that is dropped already)? Wrt the lists I think we should first untangle the execlit/request story, since I expect that a few of them need to be moved to different structs. After that it should be a bit clearer what needs to be moved where. And then there's also the golden context and render wa init code which also needs to be shuffled a bit (similar to this series). -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
