Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> writes: > On Mon, Mar 16, 2015 at 04:20:46PM +0200, Mika Kuoppala wrote: >> Perhaps not in the golden state, but do we push any workarounds through >> render ring init, that are global in scope and required by the other rings? > > No. Afaict the w/a only apply to render. Or else we have been horribly > misinformed. I still suspect it is something due to the ordering of pd > loads between rings. Serialising the render context init should not fix > the similar hangs we see at runtime. Ok please ignore this patch then. I will send another that does the immediate tail updates on ring->mm_switch. -Mika _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
