Created attachment 114684 [details] 20150328-2050_4.0.0-rc5+-start_dmesg-ignorelid_wrong-res.log This is a followup to bug 58876. My setup is a Thinkpad X200s (lid closed) sitting on an Ultrabase docking station, connected to a 1920x1200 DVI monitor by means of a DP-DVI dongle. 4.0.0 is the first kernel version since 3.7 that always seems to detect my external display: What previously most of the time ended with a black screen / standby now seems to have been replaced with KMS not setting the correct resolution. In the end KDE adjusts this correctly anyway, so this behaviour is already a big improvement, but it would be nice to get this ironed out as well. I usually have to set i915.panel_ignore_lid=0 so that KMS is able to use all of the external screen estate and make it not align a limited output to the upper and left side, this setting also affects outcome with this bug: If i915.panel_ignore_lid=0, then some very low standard resolution is set. else, the LVDS panel's native resolution seems to be set for the external screen.
Created attachment 114685 [details] 20150328-2050_4.0.0-rc5+-start_dmesg-ignorelid_right-res.log
Created attachment 114686 [details] 20150328-2050_4.0.0-rc5+-start_dmesg-wrong-res.log (really low resolution mode here)
Created attachment 114687 [details] 20150328-2050_4.0.0-rc5+-start_dmesg-right-res.log
Please grab the logs with drm.debug=0xe. drm.debug=1 doesn't really provide usefule debug information for modeset issues.
This seems to be expected behaviour. The external outputs will not be available when KMS takes over, and so they will be hotplugged in later when the dock is enabled. As a result the initial resolution is computed solely for the eDP, which will then be mirrored onto the new outputs when they arrived. If they are all available, then the mode chosen will be one common for all outputs (probably still end up with some mismatch).
(In reply to Chris Wilson from comment #5) > The external outputs will not be available when KMS takes over, and so they > will be hotplugged in later when the dock is enabled. As a result the initial > resolution is computed solely for the eDP, which will then be mirrored onto the > new outputs when they arrived. Except that 3.4.x on that particular setup and any other kernel on the same dock (but connected to a DP screen), together with i915.panel_ignore_lid=0 is always able to set the correct initial resolution - the dock is enabled from the start. It only happens in conjunction with the DP-DVI connector that it does not behave as expected (but much better than pre-4.0.0). I can send new debug logs on the weekend.
Timeout, closing. Please reopen if the problem persists with latest kernels.
(In reply to Jani Nikula from comment #7) > Timeout, closing. Please reopen if the problem persists with latest kernels. This is indeed always working now in 4.2. Sorry about forgetting to report.
(In reply to andreas.sturmlechner from comment #8) > (In reply to Jani Nikula from comment #7) > > Timeout, closing. Please reopen if the problem persists with latest kernels. > > This is indeed always working now in 4.2. Sorry about forgetting to report. No worries, thanks for the follow-up!
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.