Skip to content

intel/fs: fix incorrect register flag interaction with dynamic interpolator mode

Lionel Landwerlin requested to merge llandwerlin/mesa:review/anv-zink-interpolate-at-sample-fixes into main

What does this MR do and why?

intel/fs: fix incorrect register flag interaction with dynamic interpolator mode

Once NIR code is lowered and a few optimization passes have run, there
might be flag register interactions between instructions quite far
away from one another.

In the following case :

   f0 = and r0, r1
   ...
   fs_interpolate r2, r3
   ...
   if f0
      ...
   endif

If we lower fs_inteporlate while using the f0 register, we completely
garble the value meant for the if block.

To fix this, emit the predication for fs_interpolate in brw_fs_nir.cpp
when doing the NIR translation to the backend IR. This will guarantee
that the flag register interactions are visible to the optimization
passes, avoiding the problem above.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

Fixes: 68027bd3 ("intel/fs: implement dynamic interpolation mode for dynamic persample shaders") Closes: #9757 (closed)

Merge request reports