I think the plan was to add "struct pipe_query *render_condition" to pipe_blit_info. The value of NULL would mean the conditional rendering is disabled.
Marek On Sat, May 3, 2014 at 10:42 AM, Ilia Mirkin <imir...@alum.mit.edu> wrote: > Hello, > > I've noticed that nv50 fails the latest version of the conditional > rendering tests with glBlitFramebuffer. After looking at the changes > to the test, I just adjusted the nouveau blit code to not turn off the > current query. That fixed the failing test, but then the > copyteximage/etc tests started failing. This is probably because > pipe->blit is used for more than just the literal glBlitFramebuffer > (like glCopyTexImage). > > I briefly tested with llvmpipe which exhibits the same failure as nv50 > originally did, I'm guessing this is a general problem with gallium > rather than being nv50- (or nouveau-) specific. > > So... what's the right way of dealing with this? Adding a bit to > pipe_blit_info that indicates whether to leave the current query > enabled? Or perhaps pipe_query reference? Something else? > > Thanks, > > -ilia > > P.S. The failing test is > > bin/nv_conditional_render-blitframebuffer -auto -fbo > > And the tests that start failing after changing the logic to not > disable the query: > > bin/nv_conditional_render-copyteximage -auto > bin/nv_conditional_render-copytexsubimage -auto > bin/nv_conditional_render-generatemipmap -auto > _______________________________________________ > mesa-dev mailing list > mesa-dev@lists.freedesktop.org > http://lists.freedesktop.org/mailman/listinfo/mesa-dev _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev