[regression] Reboot after dpkg update set invalid Graphics resolutions (Dual Monitor) since ~7th/8th May

[Matthew Walker]

Now at Ubuntu 18.04.2 LTS, Gnome 3.28.2

Deduction. The user device settings menu for desktop control dual monitors offer the correct modes available. The Bionic Beaver update does something different.

Dual monitor configuration (Formerly Primary on Display Port 3840x2160 30Hz, Secondary on HDMI 3840x2160 30Hz on a three video output Intel HD620) for Cancer Research using LibreOffice Calc across both. Upgrades permitted manually rather than auto and usually done within a day. Since five months both monitors were operating HDMI (Display port not connected, not used) with Primary on HDMI2 and Secondary on HDMI1, both at 3840x2160 30Hz. Neither monitor is capable of 60Hz at UHD which is quite alright for office work. When installing such devices, strictly only the 3840x2160 30Hz mode is available on the settings page.

After the most recent update (7/8th may), the primary screen was blank on reboot (abnormal) and the secondary screen showed the extended desktop without menus (as usual). Rebooting to see the primary was unsuccessful and then the secondary monitor failed completely (confirmed the next day as graphics card hardware, broken). Cold disconnect of the secondary (awaiting repair), another boot brought up Primary display only yet flickering occasionally.

Primary Display. Considering the possibility of error in the graphics since the update - the Device settings (normal user page) revealed that the update had changed the output to HDMI 4096 lines (which the PC monitor can't do and the user can't select - outside specification). I was in such a hurry to save it also from damage that I put back the resolution to 3840x2160 30Hz immediately, without looking at which frequency had been set with the 4096 value. Now on rebooting, the display settings menu shows appropriately only 3840x2160 30 Hz as choice. It seems the Bionic Beaver update needed something else and set 4096 (which the user can't and notwithstanding the original user configuration).

Secondary Display. I don't know what it set, but it blew the monitor card. The linux sent either 3840 at 60 Hz or 4096 at any frequency without negociation, without being able to respect the user settings. It's also possible that it used a very low resolution for a few seconds which the secondary monitor can't do either; it's (well, it was) a real PC monitor with choices of DP or HDMI yet can jump into various modes (such as 60/75Hz for gamers). After some Ubuntu upgrades in the past (blank secondary screen), it was necessary to go back into the monitor menu to switch off again the gamer mode.

Something similar happened in 2017 with Light Ubuntu Update. The Secondary (very expensive) display making very alarming noises and staying blank for half a day. Now it's truly broken (I know it's the driver circuit and which parts). Waiting for repair or replacement, the system is in use now with the same Primary PC monitor and a UHD television as secondary to get by.

I worked 20 years on PC systems and monitors (for ship control) and accept that in the past we nearly always used a simple basic portable VGA monitor for system upgrades, then reconnected the user's gear later. I think I'll only do (and postpone) Linux upgrades from now using a spare single desktop monitor at 1920x1080. That was a very worrying experience.

ProblemType: Bug
DistroRelease: Ubuntu 18.04
Package: xorg 1:7.7+19ubuntu7.1
ProcVersionSignature: Ubuntu 4.15.0-48.51-generic 4.15.18
Uname: Linux 4.15.0-48-generic x86_64
ApportVersion: 2.20.9-0ubuntu7.6
Architecture: amd64
CompizPlugins: No value set for `/apps/compiz-1/general/screen0/options/active_plugins'
CompositorRunning: None
Date: Sat May 11 11:21:30 2019
DistUpgraded: Fresh install
DistroCodename: bionic
DistroVariant: ubuntu
ExtraDebuggingInterest: Yes, if not too technical
GraphicsCard:
 Intel Corporation HD Graphics 620 [8086:5916] (rev 02) (prog-if 00 [VGA controller])
   Subsystem: Holco Enterprise Co, Ltd/Shuttle Computer HD Graphics 620 [1297:4052]
InstallationDate: Installed on 2018-09-13 (239 days ago)
InstallationMedia: Ubuntu 18.04.1 LTS "Bionic Beaver" - Release amd64 (20180725)
Lsusb:
 Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
 Bus 001 Device 004: ID 0bda:0129 Realtek Semiconductor Corp. RTS5129 Card Reader Controller
 Bus 001 Device 003: ID 046d:c30a Logitech, Inc. iTouch Composite
 Bus 001 Device 002: ID 046d:c52b Logitech, Inc. Unifying Receiver
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
MachineType: Shuttle Inc. DS77U
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-4.15.0-48-generic root=UUID=9cdb8aeb-51e4-4cb5-95e7-b58572bf4dd7 ro quiet splash vt.handoff=1
SourcePackage: xorg
Symptom: display
UpgradeStatus: No upgrade log present (probably fresh install)
dmi.bios.date: 03/12/2018
dmi.bios.vendor: American Megatrends Inc.
dmi.bios.version: 1.08
dmi.board.asset.tag: Default string
dmi.board.name: FS77U
dmi.board.vendor: Shuttle Inc.
dmi.board.version: 1.0
dmi.chassis.asset.tag: Default string
dmi.chassis.type: 9
dmi.chassis.vendor: Default string
dmi.chassis.version: Default string
dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvr1.08:bd03/12/2018:svnShuttleInc.:pnDS77U:pvrV1.0:rvnShuttleInc.:rnFS77U:rvr1.0:cvnDefaultstring:ct9:cvrDefaultstring:
dmi.product.family: D
dmi.product.name: DS77U
dmi.product.version: V1.0
dmi.sys.vendor: Shuttle Inc.
version.compiz: compiz N/A
version.libdrm2: libdrm2 2.4.95-1~18.04.1
version.libgl1-mesa-dri: libgl1-mesa-dri 18.2.8-0ubuntu0~18.04.2
version.libgl1-mesa-glx: libgl1-mesa-glx 18.2.8-0ubuntu0~18.04.2
version.xserver-xorg-core: xserver-xorg-core 2:1.19.6-1ubuntu4.2
version.xserver-xorg-input-evdev: xserver-xorg-input-evdev N/A
version.xserver-xorg-video-ati: xserver-xorg-video-ati 1:18.0.1-1
version.xserver-xorg-video-intel: xserver-xorg-video-intel 2:2.99.917+git20171229-1
version.xserver-xorg-video-nouveau: xserver-xorg-video-nouveau 1:1.0.15-2

[Daniel van Vugt (vanvugt)]

Curious.

xorg-server on 18.04 hasn't changed since last year so that can't be the problem. The same goes for mutter, which would be the main package of interest if you were using a Wayland session.

There was a kernel update at the end of April, so that's the only candidate I can think of for this regression.

[Daniel van Vugt (vanvugt)]

Sorry, I misread the dates.

There was also a 'mutter' update at the end of April, so that may be the problem. To confirm this, please try reinstalling the .deb files of the prior version:

  https://launchpad.net/ubuntu/+source/mutter/3.28.3+git20190124-0ubuntu18.04.1/+build/16422451

and then reboot.

[Daniel van Vugt (vanvugt)]

Although 'mutter' isn't mentioned in the attached 'dpkg.log'. That may indicate the bug is not in mutter, or it may indicate that the assertion of 'dpkg.log' containing the regressed package might be wrong.

[Ubuntu Kernel Bot (ubuntu-kernel-bot)]

This change was made by a bot.

[Matthew Walker (techno286)]

For the confirmation, I'd need guidance. Which source/mutter link to click to reinstall the prior version (or is it all of them?).

Since writing the above report, by reflex more Ubuntu updates have been accepted (there have been very many this week - perhaps four or five) - new dpkg.log attached. After each new update the resolutions and frequencies sent to both monitors were correct; without problem. This morning, monitor 2 (the new one) followed the boot sequence in a mode I've never seen before - all pink and at ordinary HD 1920 (it seems). Then it went blank for half a second. Then it put up the usual message that it was changing to 4K at 30 Hz, then the desktop came up fine.

[Coincidentally, an email arrived that bug 1714178 (and my bug duplicate 1795439) have been fixed (released) to overcome the modesetting driver 8192x8192 limitation. I may try again using three 4K horizontal monitors - when the third is repaired. Are the patchset and merge to drm-tip before or after this particular incident?].

I feel to add (by experience) that the former damaged monitor 2 was perhaps more vulnerable than most to sudden fast mode changes. With a previous PC (ubuntu 16.04) with Intel HD600, that monitor had to receive a valid mode at a very specific moment (which meant the user had to learn to push on the power buttons precisely one second apart) else the monitor would revert immediately to high Hz modes at inferior resolutions or shut down. With this configuration Intel HD620 and Ubuntu Bionic Beaver, both monitors can be on, will move to sleep mode and wait - great! The user can switch on the PC when they're ready. Just to say that, perhaps monitor 2 was vulnerable to even a blip at 60Hz.

On the other hand, that doesn't explain how monitor 1 found itself attempting 4096 which is also part of this bug report.

Can I still help in confirmation work by going back to the prior version?

[Daniel van Vugt (vanvugt)]

It sounds like this bug is now fixed(?). If so then we don't need any further investigation thanks.