Description of problem: Latest upower causing black flashing screen. When screen flashes, I get: "[drm] nouveau 0000:01:00.0: Load detected on output A" Version-Release number of selected component (if applicable): upower-0.9.8-1.fc15 causes problems, upower-0.9.7-2.fc15 works fine How reproducible: Everytime Steps to Reproduce: 1.Update upower 2.Screen starts flashing black 3.Downgrade upower, screen works fine Actual results: Screen flashes black for a couple of seconds, then again less than a minute later. Expected results: Steady screen Additional info:
upower uses the drm state to detect if the laptop is docked or not. I'm guessing you also get the blanking when you do: cat /sys/class/drm/*/status as a normal user. If so, that's a nouveau bug, and a fairly serious DOS if that's the case.
Yes, I get that blanking screen with cat /sys/class/drm/*/status and it says: disconnected disconnected connected I get this error on an old desktop.
-- Fedora Bugzappers volunteer triage team https://fedoraproject.org/wiki/BugZappers
I am seeing what seems to be the same thing on a desktop system with an nv28. The screen blanks briefly about every 30 seconds.
This is also happening on my rv280 system. It's just that the blink is much shorter. It happens every 30 seconds.
Related to this is that the nouveau driver (but not radeon) will write a console message for each of these checks. This makes using vts a pain.
This is still happening with kernel-PAE-2.6.38-0.rc7.git2.3.fc15.i686 and xorg-x11-drv-nouveau-0.0.16-20.20110224gitbc5dec2.fc15.i686.
FWIW, I applied this the other day: commit 38199e44ff1e87b586d51d30b69d1f70e40d27f1 Author: Richard Hughes <richard> Date: Thu Mar 3 14:02:27 2011 +0000 Add an option for polling dock devices in UPower.conf, defaulting to false So now we're relying on the kernel to do the polling for us. If that doesn't work the user can re-enable the userspace poll in a config file.
That may help, but it still seems like there is a bug in the drivers.
The last upower build is from before that change. Is there going to be a new one soon?
It's not a bug in the drivers, such blinking is unavoidable with VGA load detection. I don't know about radeon, but nouveau tries its very best to avoid causing the screen to flicker when doing load detection. It does this by: 1. using an unused CRTC (if there's any available) while doing the probe, and 2. not polling for VGA disconnects when a mode is set on a VGA output If upower is sitting there explicitly asking us to reprobe whether a connected VGA monitor has disappeared, there's *nothing* we can do. So, reassigning to upower until the upstream fix has made it into the fedora package.
*** Bug 678381 has been marked as a duplicate of this bug. ***
This no longer happens with upower-0.9.9-1.fc15.i686. Thanks.