On Fri, Apr 22, 2016 at 10:39:37AM -0700, Kenneth Graunke wrote: > On Friday, April 22, 2016 3:01:27 PM PDT Juha-Pekka Heikkila wrote: > > base_mrf is unsigned, checking if its greater or equal to zero will > > not fail. > > > > Signed-off-by: Juha-Pekka Heikkila <juhapekka.heikk...@gmail.com> > > --- > > src/mesa/drivers/dri/i965/brw_blorp_clear.cpp | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/src/mesa/drivers/dri/i965/brw_blorp_clear.cpp b/src/mesa/ > drivers/dri/i965/brw_blorp_clear.cpp > > index 51f915d..06ac69e 100644 > > --- a/src/mesa/drivers/dri/i965/brw_blorp_clear.cpp > > +++ b/src/mesa/drivers/dri/i965/brw_blorp_clear.cpp > > @@ -314,7 +314,7 @@ brw_blorp_const_color_program::compile(struct > brw_context *brw, > > /* Now write to the render target and terminate the thread */ > > brw_fb_WRITE(&func, > > 16 /* dispatch_width */, > > - base_mrf >= 0 ? brw_message_reg(base_mrf) : mrf_rt_write, > > + brw_message_reg(base_mrf), > > brw_null_reg() /* header */, > > msg_type, > > BRW_BLORP_RENDERBUFFER_BINDING_TABLE_INDEX, > > > > Yeah, this is definitely dead - base_mrf seems to always equal 2. > Thanks for catching this :) > > The two parts of the conditional should be equivalent. One has an > <8,8,1> region and the other <16,16,1> but those refer to the same > region, AFAIK. > > I'd be tempted to just use mrf_rt_write unconditionally. If we go with > your patch, we should delete mrf_rt_write because it becomes dead. > > Topi, thoughts?
I agree. I think I we can drop it also in fs_generator::generate_blorp_fb_write(), just need to adjust brw_blorp_blit_program::render_target_write() accordingly. _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
