piglit.spec.arb_shader_ballot.execution.fs-readinvocation-uint Standard Output /tmp/build_root/m64/lib/piglit/bin/shader_runner /tmp/build_root/m64/lib/piglit/tests/spec/arb_shader_ballot/execution/fs-readInvocation-uint.shader_test -auto -fbo piglit: debug: Requested an OpenGL 3.2 Core Context, and received a matching 4.5 context Probe color at (4,244) Expected: 0 255 0 255 Observed: 255 255 10 8 Test failure on line 37 This is a newly enabled feature that is broken on atom platforms. It is not a regression, but the Intel Mesa engineers would like to fix it before the 17.2 release.
I looked at this with Curro a few weeks ago, and we noticed that commenting out if ((active_set & (1ul << i)) == 0ul) continue; makes the test pass. Poking a little at it, I think the cause is shl(8) g8<1>Q g4<4,4,1>Q g2<4,4,1>UD { align1 1Q }; CHV/BXT require that the "Source and Destination horizontal stride must be aligned to the same qword." for 64-bit operations. I'll see if fixing that sorts this out.
The series I sent last night fixes this and adds EU validation code to that would catch the problem.
Fixed by: commit b541945c2027990ac571184bbf8e01285be0e33a Author: Matt Turner <mattst88@gmail.com> Date: Thu Sep 21 16:42:02 2017 -0700 i965/fs: Unpack count argument to 64-bit shift ops on Atom
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.