saltyeggs's questions

Only since adding a second monitor have I been experiencing soft lockups on suspend and roughly 20% of logins on Ubuntu 20.04 - I do not experience either of these issues with just a single monitor connected.

Hardware Details:

My existing Acer XB270HU only has a single DisplayPort input (and no other inputs at all) and my GTX 970 only has one Displayport output and one HDMI output. A new MSI MAG274QRF-QD is now instead using the DisplayPort output and I have connected the Acer back to the graphics card using this HDMI to DisplayPort adapter. The motherboard is a Gigabyte GA-Z97MX-Gaming 5. CPU: i5-4690K.

Intermittent Soft Lockup on Login

Even when it does not crash it can take between 40 seconds and 90 seconds after login for my desktop to appear though when just using one monitor this happens in seconds. When it crashes the screen (or screens as it varies between one or both having signal immediately after login) will display a black screen and then the cursor will load either frozen from the beginning or almost immediately after appearing. When this occurs I am unable to get to a TTY and REISUB works at least sometimes.

The login screen mostly loads on the old Acer but sometimes it loads on the MSI, although there seems to be no pattern between this and the frozen cursor.

journalctl syslog Xorg.0.log.old Xorg.1.log.old: Nothing for the same time period

Soft Lockup on Suspend

When I try to suspend the monitors stop receiving the signal and turn off, the keyboard lights flash at least sometimes like the system is trying to suspend but the fans in my system keep going indefinitely. I cannot restore the display by typing on the keybaord, or trying a TTY, and REISUB seems to work at least sometimes.

journalctl syslog Xorg.0.log.old Xorg.1.log.old

Separate to Ubuntu but may be relevant is that POST and the prompt to unlock my encrypted volume (before the Ubuntu login prompt) appear only on the Acer but in low resolution in the form of a miniature screen in the middle of the display - I expect this is down to the adapter, which works fine once I am logged into Ubuntu successfully. I would prefer this all to happen on my new MSI and in fullscreen but I have found no setting in the BIOS to do this, and I have tried enabling CSM.

Is there something that might resolve these two issues?

A few weeks ago I clean installed 20.04.1 and at the same time I switched to the Brave Browser. As I was frequently experiencing my system freezing, and by freezing I mean I could still move the mouse cursor but nothing would respond, I assumed the issue was the Brave Browser and I found on their git a history of a similar issue. I disabled hardware acceleration in Brave, which is a reported workaround, but when that did not help I switched back to Chrome.

Unfortunately even using the latest version of Chrome, the issue is ongoing. When the system becomes unresponsive it stays so for approximately 30 seconds to 1 minute. A few times I have had to use REISUB or when that did not work cold reboot, but the majority of the time it just starts responding again. I have tried to completely mirror the same clean install setup and packages on my Dell XPS 13 9360, which does not experience the same issue.

As I post this morning some of the lockups I have experienced today have allowed me to switch window, for instance terminal would freeze but I could switch to a non-responsive Chrome or vice versa.

Please find several logs below which were created after the issue occurred and immediately after my system started responding again. Often I am not doing anything especially heavy, just Chrome open with 4 or 5 tabs. My system is an i5-4690K, GTX 970, 8GB of RAM, Samsung SSD 840 EVO Basic SATA. I am using the Nvidia driver 450.66.

Searching for a similar error to one of those that occurred a lot this morning: 06:13:58 fuser: Cannot stat file /proc/2733/fd/1023: Permission denied (the path after /proc/ varies) I found this thread Getting strange errors while downloading mysql and although I was not installing MySQL and I do not have it installed but in case it could help I followed some of the troubleshooting steps: libaio1 was installed but I input sudo apt-get install libaio1 libaio-dev, sudo apt install aptitude, sudo aptitude update, sudo aptitude upgrade, and rebooted. Unfortunately the same issue occurs, and as you will see from the latter 4 logs this happened as many times over 34 minutes, and those are just the ones I generated logs for and the count is more like twice that. Since rebooting about 30 minutes ago it has only happened twice so far but I expect it to continue throughout the day as normal.

https://paste.ubuntu.com/p/T8mfWwsyHS/

https://pastebin.ubuntu.com/p/hKNtVqdkbz/

https://pastebin.ubuntu.com/p/dCBjJqJRzs/

https://pastebin.ubuntu.com/p/vM8X6TbCC6/

https://pastebin.ubuntu.com/p/WnmRmX65c8/

https://pastebin.ubuntu.com/p/jnhGPFgNpX/

Could you please review my logs and let me know if there is something that may suggest how to resolve this?

I have hibernate working, and I can also trigger 'true' hybrid-suspend from the command line or my power menu, but I want to configure Ubuntu 20.04 to hybrid-suspend when idle rather than just suspend or hibernate.

When I say 'true' hyrbrid-suspend I mean 'true' hybrid-sleep, suspend-and-hibernate, suspend-to-both, where my desktop will suspend and hibernate at the same time - if it has power when I turn it back on it will resume from RAM, if not it will resume from disk. There is some confusion around this as hybrid-sleep is sometimes used to refer to suspend-then-hibernate, suspend first and hibernate after a set period of time, and why the former, and my desired setup, is often denoted 'true'.

There are several articles explaining how to do this, like How to Use Hybrid Suspend in Ubuntu but that method references a directory that does not exist: /etc/pm/config.d/

Similarly, How do I use pm-suspend-hybrid by default instead of pm-suspend? has a section for 18.04 but reports:

In the hybrid-sleep mode, the hibernate part gets effective only when the battery is critically low and the system shuts down.

Whereas I want it to suspend to both from the beginning. In the same post there is a section for 16.04 which references hybrid-sleep but it has already said that hybrid-sleep but as quoted this hybrid-sleep will only hibernate when the battery is critically low.

How can I set this up please?

I am trying to install *20.04.01 on my desktop with full (or almost full as Tj puts it) disk encryption including encrypted swap partition that allows hibernation, and btrfs.

The last few years I am using ManualFullSystemEncryption but even being careful to make sure I do not turn my system off after an update before the grub fix script automatically runs, these updates or *Windows updates if dual booting, have broken grub and I have often had to boot to a live USB and run fix-grub.sh.

Yesterday I followed Tj's Full_Disk_Encryption_Howto_2019 but when I got to the point of formatting /dev/mapper/ubuntu--vg-root I searched to see if I needed to do anything extra to use btrfs and ext3. I have since learned that using btrfs is not as simple as just selecting another filesystem from the dropdown.

This next brought me to Willi Mutschler's Ubuntu 20.04 with btrfs-luks full disk encryption including /boot and auto-apt snapshots with Timeshift which is very similar to Tj's, although it includes many extra steps for optimising btrfs and SSD. This is almost perfect for my intention, the only issue is that Willi sets it up so that swap is encrypted with a random password as he does not use hibernation:

export SWAPUUID=$(blkid -s UUID -o value /dev/vda2)echo "cryptswap UUID=${SWAPUUID} /dev/urandom swap,offset=1024,cipher=aes-xts-plain64,size=512" >> /etc/crypttab
cat /etc/crypttab
# cryptdata UUID=8e893c0f-4060-49e3-9d96-db6dce7466dc none luks
# cryptswap UUID=9cae34c0-3755-43b1-ac05-2173924fd433 /dev/urandom swap,offset=1024,cipher=aes-xts-plain64,size=512

He references Archlinux's dm-crypt/Swap encryption which I have looked through, and it would seem I need to set up some kind of hook for swap, but I have not been able to make much more sense of it than that.

Last night I found Félix Saparelli's Full-disk encryption with Btrfs, swap, and hibernation. It gives commands for setting up encrypted swap but not the rest of the install, so I had planned today to try and use it in conjunction with Willi's guide to achieve my desired install.

I wanted to post and ask if this is the best approach, or if there is better way to do this or a more complete guide?

Edit: HOWTO - GPT/UEFI install with full disk encryption: BTRFSonLUKS with separate root, home and pkg subvolumes; hibernation with a swapfile; auto-snapshots with easy system rollback (GUI); boot into snapshots describes the kind of install I am trying to get to although I am not sure exactly how to change it to work with Ubuntu.

*I would prefer to install Ubuntu Server edition on my desktop and then manually install my desktop environment as I have read this is an even more stripped down version of Ubuntu than the minimal install. Unfortunately when following Tj's guide, which is not intended for Server, after selecting the partitions the installer errored. Willi's guide references files for optimising btrfs in /usr/lib/partman which is not contained in the server installer and the same files do not exist anywhere else and it does not seem I can just install a package for partman.

*I was running a setup with Windows 10 on the same drive encrypted with VeraCrypt. This time I am not dual booting.

I am happy to use either gdm3 or lightdm, but neither are stable on my Ubuntu 18.04 system, they will work for either one or several boots after I take corrective action, but inevitably this results in the greeter not loading - the last few times (at least) I noticed this was immediately after I did a sudo apt update. The graphics card on my Intel Core i5 system is an Nvidia GTX 970, and I manually installed nvidia-driver-415 using apt when I could not load a greeter as soon as I had installed.

Searching here and elsewhere, I have tried many things to resolve this, which I will outline below. If you know something else I can try - to get either greeter stable - please let me know.

gdm3

Even with WaylandEnable=false uncommented in /etc/gdm3/custom.conf, I could never login, instead of the greeter I would see either a black screen or graphic distortion like a black screen with some pink lines across the top, and I am only able to login setting nomodeset.

gdm3 sample distortion

With gd3 not working, I tried lightdm, and on encountering similar issues switched back to gdm3 again:

sudo apt-get install gdm3
sudo dpkg-reconfigure gdm3

This seemed to work - kind of, for the first few days I could boot most of the time. I have quiet splash permanently removed from my boot options, and boot was sometimes hanging on Started NVIDIA Persistence Daemon as below:

gdm3 fail 1

Another time with gdm3 I saw the following output instead of the greeter:

gdm3 fail 2

gdm3 fail 3

lightdm

When I use lightdm instead, and the greeter does not load, I see a flashing black screen with a single white underscore top left. I can never get to a TTY using ctrl + alt + f3/f4/f5, REISUB does not work, so I hard rebooted, the greeter loads with nomodeset, and from there without logging in I switch to a TTY and I have tried what I outline below. Sometimes if I just reboot again without nomodeset the greeter will just load properly, without my changing anything.

lightdm underscore

Observing these permissions:

-rw-------  1 dusf dusf      56 Jan  7 17:38 .Xauthority
-rw-------  1 dusf dusf   15886 Jan  7 17:38 .ICEauthority

drwxrwxrwt 16 root root 4096 Jan  7 17:45 /tmp

I input the following, and could login normally next reboot:

sudo chown dusf:dusf .Xauthority
sudo chown dusf:dusf .ICEauthority
sudo chmod a+wt /tmp

The next time the issue occurred I input the following, and could login next reboot:

sudo dpkg-reconfigure lightdm

In cat .xsession-errors, I noticed the following line, which I thought could be relevant considering I see an underscore instead of a greeter?

_=/usr/bin/dbus-update-activation-environment

Next temporary fix:

mv .Xauthority .Xauthority.bak

Again:

sudo apt-get purge lightdmsudo apt-get install --reinstall lightdm
sudo apt-get install gnome

After a sudo apt upgrade a few days ago with the same issue recurring I input all of the following and everything seemed fine for a few days:

sudo apt upgradesudo apt purge lightdm
sudo apt purge gnome
sudo apt purge nvidia*
sudo apt remove nvidia-driver-415
reboot
sudo apt purge nvidia-*
sudo apt install nvidia-415
sudo apt install nvidia-driver-415
reboot
sudo apt purge xorg-* xserver-xorg
sudo apt install xorg xserver-xorg
sudo dpkg-reconfigure xorg
sudo apt purge ubuntu-gnome-desktop
sudo apt purge gnome-desktop-environment
apt install ubuntu-gnome-desktop
sudo apt install ubuntu-gnome-desktop
sudo apt purge lightdm
sudo apt install --reinstall lightdm

The next time the issue occurred I noticed in my .bash_history the last commands I input were:

sudo apt update
sudo apt upgrade

and I uploaded the following logs:

/var/log/lightdm/lightdm.log

cat /var/log/kern.log

cat .xsession-errors

I read adding an xrandr line to some files may be the next step, but the actual lines seem to be system specific, and their are recommendations on adding them to different files.

How can I begin to resolve the issue, either with gdm3 or lightdm please?