https://bugs.freedesktop.org/show_bug.cgi?id=76531
Priority: medium Bug ID: 76531 Assignee: dri-devel@lists.freedesktop.org Summary: linux-3.14-rc7 radeon/nouveau runtime-suspend never coming back Severity: normal Classification: Unclassified OS: Linux (All) Reporter: bonbons67@internet.lu Hardware: Other Status: NEW Version: XOrg CVS Component: DRM/other Product: DRI
Most recent testing: 3.14-rc7 mainline.
As suggested on IRC by imirkin, returning -EBUSY instead of -EINVAL from nouveau_pmops_runtime_suspend() avoids driver to fall too far to the ground.
When returning -EINVAL connectors get being seen as "disconnected" and there is no way to get them connected anymore.
When returning -EBUSY the connectors stay functional but DE (enlightenment) complains about missing crtc and all that is visible is black background and a cursor that response to mouse/trackpad.
Way to reproduce: Have xorg running, then for each avtive output call: xrandr --output $OUTPUT --off
When last output got marked off runtime-suspend kicks in.
Same/similar behavior has been observed on radeon (I don't remember the exact kernel releases, but older ones got console to display when switching VTs while newer failed to do so).