This is a rather difficult issue to reproduce. After leaving my PC idle overnight (monitor turned off) and waking up the next day, I noticed an odd issue started happening: The monitor started randomly cutting to black for a second. It was not entirely unpredictable: it seemed to happen especially when I triggered a seek in `mpv` -perhaps because this either recreated the framebuffer or recreated some video textures, I'm not entirely sure. Restarting X seems to have fixed it. The GPU was in power mode 'auto' overnight. It sort of reminds me of similar issues I had with the nvidia drivers some years ago, where leaving them in 'adaptive' performance mode overnight caused them to fail switching frequencies the next day. Leaving them in 'performance' mode overnight solved that issue; but it has also been since fixed. But that was an issue actually related to the performance scaling, whereas this seems more like an issue with the display connection in general. I'm not sure where else to take this - I can try reproducing it the next time I go to sleep, I guess, just to see if it will happen again. Apart from that, I'm pretty much blind debuggin here. Xorg version is 1.19.3, mesa version is 17.2, kernel version is 4.12.4, device is a Sapphire RX 560.
Can't reproduce the cut-to-black issue, after several attempts (with either 'auto' or 'high' performance mode). I did notice another issue while testing though, which is sort of related: If leaving the GPU in 'high' performance mode while the display is turned off, it still goes down to ~300 MHz core clock - except that it never goes back up to high performance mode. This bug seems extremely similar to the one I had with nvidia drivers (although nvidia drivers required an X restart, whereas simply going from 'high' to 'auto' and back clears this issue).
Bug is possibly fixed in the meantime. I think I haven't noticed it for a while. Closing pre-emptively.
-- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/drm/amd/issues/239.
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.