ubuntu-x-swat team mailing list archive

[Chris Halse Rogers <raof@xxxxxxxxxx>]

This is actually pretty specific to my setup, but I don't think the
relevant code has changed for quite some time.

The sequence of events that's needed to trigger this is:
* Start up with two monitors plugged in.  X will clone them when it starts, and unity-greeter sets a cursor.
  - At this point both outputs are being driven by a single CRTC.
* On login, have g-s-d disable the LVDS display and change the HDMI display to native resolution.
  - At this point the first CRTC is disabled *without* having the pointer disabled, then the HDMI output is brought up on a new CRTC
* VT switch to a console.
  - Before the VT switch, X hides the cursor on all active outputs.  That means, it hides the cursor on the CRTC currently driving the HDMI output.  When KMS takes over to display the console, it needs to bring up 2 CRTCs, as the LVDS and HDMI outputs are driven at different resolutions.  The LVDS gets the CRTC which was driving the HDMI output - that doesn't have a cursor associated, because X hid it before VT switching.  The HDMI output gets the CRTC that was driving the cloned displays in unity-greeter; because nothing hid the cursor there, that's still got a cursor active.

To really fix this requires a bit of kernel refactoring - i915 should
hang cursors of the framebuffer object, not off the CRTC.

This could be worked-around *reasonably* easily - by having X hide the
cursor before it disables an output.  My naive patch to do that is
broken, though; it results in the cursor being hidden when coming from
lightdm :).

-- 
You received this bug notification because you are a member of Ubuntu-X,
which is subscribed to xserver-xorg-video-intel in Ubuntu.
https://bugs.launchpad.net/bugs/826508

Title:
  Leaves mouse pointer visible when changing VTs

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/826508/+subscriptions