This test seems to reliably cause a gpu hang on ICL. Test output: /tmp/build_root/m64/lib/piglit/bin/shader_runner /tmp/build_root/m64/lib/piglit/tests/spec/arb_compute_shader/execution/shared-atomicMax-uint.shader_test -auto -fbo piglit: debug: Requested an OpenGL 3.3 Core Context, and received a matching 4.5 context Atomic counter 0 test failed: Reference == Observed Reference: 0 Observed: 60 Test failure on line 137
Created attachment 144590 [details] error state
This test seems to pass reliably for me, I have PCI ID 0x8a56 using Mesa @ a7f75377aba and drm-tip 5.2.0-rc5+
(In reply to Tapani Pälli from comment #2) > This test seems to pass reliably for me, I have PCI ID 0x8a56 using Mesa @ > a7f75377aba and drm-tip 5.2.0-rc5+ Argh sorry, this was on i965.
(In reply to Tapani Pälli from comment #3) > (In reply to Tapani Pälli from comment #2) > > This test seems to pass reliably for me, I have PCI ID 0x8a56 using Mesa @ > > a7f75377aba and drm-tip 5.2.0-rc5+ > > Argh sorry, this was on i965. Having said that, this test seems to reliably pass for me on Iris as well, Mesa @ 946193ae008.
Clayton, what kernel version was used? I see no hangs even with quite old Mesa so I'm suspecting something might have been fixed at kernel side?
I was using a drm-tip kernel from March of this year. I am hoping to update the kernel on the ICL systems in CI, along with the bios, so maybe this (plus some of the other gpu hangs) will be resolved. It's good to know you cannot reproduce this on a newer kernel though, it gives me hope!
I'm not able to reproduce the gpu hang using drm-tip kernel + newer platform BIOS.
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.