[Mesa-dev] [PATCH] r600/sb: fix rotated register in while loop

Gert Wollny gw.fossdev at gmail.com
Mon Feb 19 13:37:15 UTC 2018


Am Montag, den 19.02.2018, 14:06 +1000 schrieb Dave Airlie:
> On 15 February 2018 at 01:26, Gert Wollny <gw.fossdev at gmail.com>
> wrote:
> > Am Mittwoch, den 14.02.2018, 17:18 +1000 schrieb Dave Airlie:
> > > From: Dave Airlie <airlied at redhat.com>
> > > 
> > > A bunch of CTS tests led me to write
> > > tests/shaders/ssa/fs-while-loop-rotate-value.shader_test
> > > which r600/sb always fell over on.
> > > 
> > > This patch fixes it, but I'll probably never be 100% sure why.
> 
> I've gone back to this approach, I've spot tested and I don't see
> as many regressions.

For me most of the assertion failures don't come up anymore, but in
total the numer of regressions is still quite high, i.e. 1 assertion
failure and 13 normal failures. 

  vs-struct-nonconst-non-opaque-members fails with           

  b/sb_sched.cpp:931:process_fetch: 
      Assertion `f->parent->count() ==  1' failed.

and the list of normal failures is very similar to you first patch, the
only change is that 
  glsl-1.10/execution/variable-indexing/vs-output-array-float-index-wr 
no longer fails, but instead 
  glsl-1.50/execution/geometry/dynamic_input_array_index
fails, 

Best, 
Gert




More information about the mesa-dev mailing list