Unity8 never loads and just stays on a black screen [unity8.log: ERROR: QMirServer - Mir failed to start]

When this happens switch to TTY1 (Ctrl+Alt+F1) and restart lightdm.

sudo systemctl restart lightdm

Unity 8 should start after this.

There is another bug reported about this issue. Either mark that other one (Bug #1539811) as duplicate or this one.

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

looks to be the same in released xenial

I think this bug affects me as well.

It seems that the system doesn't switch to tty8 properly. If desktop fails to load I can sometimes get it to appear by first switching to tty1 then tty8. A bit more reliable way seems to switch to tty1 and delete ~/.cache before logging into unity8.

Same problem. Neither restarting lightdm nor deleting ~/.cache makes any difference.

How do I switch back to Unity7, it sucks!

I tried on my desktop (nvidia graphics card) and within a Fusion VM on my MacBook. I have a black screen in both cases, though it says something on my desktop before.

The message is related to "fifo engine fault". I didn't manage to capture the whole thing as it disappeared pretty fast.

In both cases I can't get to TTY1, though it may be related to some badly configured keymap in the VM.

I forgot to add I tried with the free driver (I think it's Nouveau) on my desktop with the nvidia graphics.

I get a black screen too, if I SSH in and dmesg I get this in the log:

[ 8.025978] [drm:atom_op_jump [radeon]] *ERROR* atombios stuck in loop for more than 5secs aborting
[ 8.025995] [drm:atom_execute_table_locked [radeon]] *ERROR* atombios stuck executing E4DE (len 2650, WS 0, PS 8) @ 0xEAE6
[ 8.121017] [drm:radeon_dp_link_train [radeon]] *ERROR* clock recovery reached max voltage
[ 8.121038] [drm:radeon_dp_link_train [radeon]] *ERROR* clock recovery failed

Over ssh I added nomodeset and it boots into the desktop, although the plymouth is not showing properly and it looks like the res is wrong.

All: We have a few different black screen bugs around Unity8/Mir. Please be sure to log a bug of your own detailing your system configuration...

General bugs like this which have no involvement from the original reporter tend to never get definitively resolved.

nvidia nouveau users: try bug 1575516
nvidia proprietary driver users: may see bug 1553549
intel pineview users: see bug 1549455
virtual machine users: see bug 1118903

ferenos:
Please attach more details about your system like running 'lspci'.
Also, please try this command which will upload more information about your system to this bug:
    apport-collect 1537389

and the last updates broke it for me too (black screen on tty8, Num Lock doesn't work on tty8)

unity8.log
android_dlopen called but not implemented!
unity8: linker.c:27: android_dlopen: Assertion `0' failed.
()
qtmir.screens: ScreensModel::ScreensModel
[2016-05-06 10:42:04.970417] mirplatform: Found graphics driver: mir:mesa-x11 (version 0.22.0)
[2016-05-06 10:42:04.972690] mirplatform: Found graphics driver: mir:mesa-kms (version 0.22.0)
qtmir.mir: MirServer created
qtmir.mir: Command line arguments passed to Qt: ("unity8")
[2016-05-06 10:42:04.989166] mirserver: Starting
ERROR: QMirServer - Mir failed to start

unity8-dash.log
Loading module: 'libubuntu_application_api_desktop_mirclient.so.3.0.0'
Loading module: 'libubuntu_application_api_desktop_mirclient.so.3.0.0'
UbuntuClientIntegration: connection to Mir server failed. Check that a Mir server is
running, and the correct socket is being used and is accessible. The shell may have
rejected the incoming connection, so check its log file
UbuntuClientIntegration: connection to Mir server failed. Check that a Mir server is
running, and the correct socket is being used and is accessible. The shell may have
rejected the incoming connection, so check its log file

i do have the overlay ppa so in my case is probably this bug https://bugs.launchpad.net/mir/+bug/1578461

dinamic:

Indeed comment #15 confirms you are hitting bug 1578461:

android_dlopen called but not implemented!
unity8: linker.c:27: android_dlopen: Assertion `0' failed.
()

right, which is a hybris bug, so 'incomplete' (or making this bug a dup of 1578461) seems appropriate

Incorrect. dinamic is not the reporter of this bug so what dinamic experiences is not relevant here. This bug is owned by ~ferenos.

I have the same configuration as the OP it seems so if they have abandoned the report I am happy to take over. The error in the log was: Failed to initialize EGL

I have a mobile AMD GPU with open source drivers: Gallium 0.4 on AMD RV620
Full unity 8 log here: http://paste.ubuntu.com/16581290/

The comment suggesting $ sudo sysctl restart lightdm
fails with
sysctl: cannot stat /proc/sys/restart: No such file or directory
sysctl: cannot stat /proc/sys/lightdm: No such file or directory

Switching TTYs and back again doesn't help. I still have a completely black screen.

Thanks for that.

Please just log a new bug with your system details so we can treat it separately:
https://bugs.launchpad.net/ubuntu/+source/unity8-desktop-session/+filebug

Found a duplicate bug and also my xenial laptop experiencing the same failure regularly...

unity8.log:
()
qtmir.screens: ScreenController::ScreenController
[2016-06-07 09:13:14.320014] mirplatform: Found graphics driver: mir:mesa-x11 (version 0.21.0)
[2016-06-07 09:13:14.320952] mirplatform: Found graphics driver: mir:mesa-kms (version 0.21.0)
qtmir.mir: MirServer created
qtmir.mir: Command line arguments passed to Qt: ("unity8")
[2016-06-07 09:13:14.327471] mirserver: Starting
[2016-06-07 09:13:14.337275] mirserver: Using nested cursor
ERROR: QMirServer - Mir failed to start

I have unity-system-compositor and unity8-dash running fine. Only unity8 is missing.

Come to think of it, duplicate bug 1543133 might be a better report of this problem.

Interesting. Here's the unity8.log from one machine when it's working and (randomly) not working. This might help us narrow down the point of failure:

WORKING:
qtmir.mir: MirServer created
qtmir.mir: Command line arguments passed to Qt: ("unity8")
[2016-06-07 09:39:06.463750] mirserver: Starting
[2016-06-07 09:39:06.472754] mirserver: Using nested cursor
[2016-06-07 09:39:06.508085] mirserver: Initial display configuration:
[2016-06-07 09:39:06.518572] mirserver: 0.31: LVDS 12.7" 280x160mm
[2016-06-07 09:39:06.518613] mirserver: Current mode 1366x768 60.01Hz
[2016-06-07 09:39:06.518661] mirserver: Preferred mode 1366x768 60.01Hz
[2016-06-07 09:39:06.518803] mirserver: Logical position +0+0

NOT WORKING:
qtmir.mir: MirServer created
qtmir.mir: Command line arguments passed to Qt: ("unity8")
[2016-06-07 09:13:14.327471] mirserver: Starting
[2016-06-07 09:13:14.337275] mirserver: Using nested cursor
ERROR: QMirServer - Mir failed to start

Comment #5 agrees with bug 1589550, and they might explain how Mir might be hung on startup. QtMir only waits 10 seconds before giving up and committing suicide so if we're blocked by being on the wrong VT that could explain everything...

Can you try lightdm 1.19.1 / 1.18.2 and see if that solves this?

This might have been resolved as bug 1576260 in Mir 0.23.0

Does anyone still see 'ERROR: QMirServer - Mir failed to start' in Ubuntu 16.10 yakkety?

De-prioritized per comment #27.

[Expired for unity8-desktop-session (Ubuntu) because there has been no activity for 60 days.]

[Expired for unity8 (Ubuntu) because there has been no activity for 60 days.]

[Expired for mir (Ubuntu) because there has been no activity for 60 days.]

[Expired for Mir because there has been no activity for 60 days.]