Created attachment 92108 [details] dmesg System Environment: -------------------------- Platform: Haswell(nv) Kernel: (drm-intel-nightly)07db3e0b8568d81226e2e17dbb1b340668cb2287 Bug detailed description: ------------------------- Many pm_pc8 subcases fail on haswell with NVIDIA card GT440. They fail on -nightly and -queued kernel, work well on -fixes kernel. The latest known good commit:d299cce76e001406df40c4ff712b33fccbce1222 The latest known bad commit: 534e3730732ee01e43cefa02240b0552a62781e4 run subcase i2c output: Runtime PM support: 1 PC8 residency support: 0 Test assertion failure function enable_one_screen, file pm_pc8.c:354: Last errno: 5, Input/output error Failed assertion: enable_one_screen_with_type(data, SCREEN_TYPE_ANY) Subtest i2c: FAIL Reproduce steps: ---------------- 1. ./pm_pc8 --run-subtest i2c
Do you have any outputs attached to the Intel card? It doesn't make sense to not have any, but I could try to SKIP the whole suite if that's the case.
Also, could you please finish bisecting? Thanks, Paulo
Test on latest -queued kernel. The fail goes away. output: IGT-Version: 1.5-gd1b5823 (x86_64) (Linux: 3.13.0-rc8_drm-intel-next-queued_369a13_20140129_debug+ x86_64) Runtime PM support: 1 PC8 residency support: 0 Subtest i2c: SUCCESS
(In reply to comment #3) > Test on latest -queued kernel. > The fail goes away. > output: > IGT-Version: 1.5-gd1b5823 (x86_64) (Linux: > 3.13.0-rc8_drm-intel-next-queued_369a13_20140129_debug+ x86_64) > Runtime PM support: 1 > PC8 residency support: 0 > Subtest i2c: SUCCESS I'm a little confused, but I'm going to assume it works now because either a monitor was connected to the Intel graphics card or whatever was just in -fixes got merged to -nightly and -queued. So closing bug. If you think the bug is still not closed, please reopen it and explain the remaining problem. Thanks, Paulo
> I'm a little confused, but I'm going to assume it works now because either a > monitor was connected to the Intel graphics card or whatever was just in > -fixes got merged to -nightly and -queued. So closing bug. > > If you think the bug is still not closed, please reopen it and explain the > remaining problem. > > Thanks, > Paulo Paulo, pm_pc8 should skip except ULT? This pass happens on desktop(haswell with nv), If it's expected, I will close it.
This fail happens with headless. Comment 3's pass due to connect HDMI. Reopen it.
It doesn't cause by NV. It fails with headless. Bisect shows: df4547d82589714f1b0cecd93569130a452cbf46 is the first bad commit commit df4547d82589714f1b0cecd93569130a452cbf46 Author: Paulo Zanoni <paulo.r.zanoni@intel.com> Date: Fri Dec 13 15:22:32 2013 -0200 drm/i915: add runtime PM support on Haswell Now that all the infrastructure is in place and all the tests from pm_pc8 pass, we can finally enable the feature. Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
I guess we need to skip some (or even all) of the pc8 tests if there's no output. And add a big fixme that we need to revisit this once we have working output config injection support for igt kms tests. Paulo, can you pls take a look at this again?
Hi I believe this is fixed by commit "tests/pm_pc8: skip tests when no screens are connected", which was committed to the intel-gpu-tools repository today. If the bug still persists, please reopen. Thanks, Paulo
Verified.Fixed.
Closing old verified.
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.