The shader runner test fails as a result of inability to spill enough registers. The theory is that gen9 needs the message header for simd4x2. This causes two problems, first is extra register pressure, and the second is our vec4 spilling code cannot spill registers with length greater than 1 (the header is 2). I played around with fixing this, but I didn't yet succeed (I don't remember what's in these branches, they're just for posterity). http://cgit.freedesktop.org/~bwidawsk/mesa/log/?h=vec4-more-spill http://cgit.freedesktop.org/~bwidawsk/mesa/log/?h=vec4-more-spill2 http://cgit.freedesktop.org/~bwidawsk/mesa/log/?h=vec4-more-spill3
This was fixed a while ago I believe by Iago Toral's live range changes. Now NIR is default, so any other issues with this test should get a new bug.
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.