On 01/08/2014 04:27 PM, Chris Forbes wrote:
>> + OUT_BATCH(0);
>> + ADVANCE_BATCH();
>> + } else {
>> + BEGIN_BATCH(4);
>> + OUT_BATCH(_3DSTATE_PIPE_CONTROL | flags | (4 - 2));
>
> If the PRM is to believed, then ILK has two of its flag bits in dw1
> [Stall at scoreboard, and the nonfunctional depth cache flush
> inhibit]; it looks as if we never try to actually use stall at
> scoreboard pre-Gen6 though, so this works.
>
>
> Patches 1-3 are
>
> Reviewed-by: Chris Forbes <chr...@ijw.co.nz>
Oh yeah...I don't think we ever use those bits on ILK. Not sure whether
the PRM is to be believed or not. It's probably right.
Suggestions? I could try to add defines for those flags and make it
actually work, but...kind of loathe to do that. I could also add a
comment saying that it's technically wrong for ILK...
--Ken
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
