after disabling a display, enabling makes screens go black

[Jonas G. Drange]

Reproduce:
turn off then turn on some display (output.used = false then output.used = true)

What happens:
Turning off works fine, turning it back on does not

What should happen:
it should work as expected

USC log: http://pastebin.ubuntu.com/23420797/
Unity8 log: (see that mirserver: Disabled works fine) http://pastebin.ubuntu.com/23420803/
Unity8 log (w debug): http://pastebin.ubuntu.com/23451437/

[Daniel van Vugt (vanvugt)]

Your USC log shows USC is using the second display when you re-enable it:

[2016-11-03 14:51:24.574533] mirserver: 0.52: DisplayPort 24.9" 550x310mm
[2016-11-03 14:51:24.574558] mirserver: Disabled
...
[2016-11-03 14:51:29.648108] mirserver: 0.52: DisplayPort 24.9" 550x310mm
[2016-11-03 14:51:29.648126] mirserver: Current mode 2560x1440 59.95Hz
[2016-11-03 14:51:29.648137] mirserver: Preferred mode 2560x1440 59.95Hz
[2016-11-03 14:51:29.648145] mirserver: Logical position +1920+0

However Unity8 probably doesn't yet understand what that means. We have an agreement from last month that we would like Unity8 to just always clone to all displays. But your log shows we're still spanning (Logical position +1920+0). I would not expect Unity8 to understand what spanning means yet.

Since we have an agreement from last month that we'd like to just always clone in the absence of a design for spanning, we should enable that. USC already supports the option (--display-config=clone) so we just need to use it...

So long as your monitor's backlight is on and we're just not painting to it (hence black), this is not a Mir or USC bug. It just requires more work in Unity8 to paint across the span (once we have a design for that), or shorter term work in LightDM to add --display-config=clone for USC.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Daniel van Vugt (vanvugt)]

Is your backlight coming back on? If not then that part is a Mir bug.

[Gerry Boland (gerboland)]

QtMir/Unity8 is designed to deal with spanning displays - it is actually cloned displays it needs work to support!

@jonas Could you give us a means to test this behaviour?

[Gerry Boland (gerboland)]

Yeah the unity8 log makes me suspect qtmir, it deals with disabling the display ok, but enabling it doesn't yield 2 screens again

[Jonas G. Drange (jonas-drange)]

Silo 2140 as mentioned will give you a way to reproduce this. I am
currently unable to since I've lost the option to log into a unity8 desktop
session.

[Jonas G. Drange (jonas-drange)]

(Got my bugs mixed up, I didn't actually mention silo 2140. :))

[Jonas G. Drange (jonas-drange)]

> Is your backlight coming back on? If not then that part is a Mir bug.

The backlights are coming back online.

[Daniel van Vugt (vanvugt)]

Thanks. The logs show Mir knows about the display getting re-enabled and the backlight coming back on is the only part that Mir is responsible for. I think the relevant aspects of Mir are working correctly.

[Bill Filler (bfiller)]

@jgdx
Can you make sure that silo 2140 is up to date so this can be tested again?

thanks