I am trying to upgrade from Ubuntu 18.04.5 to 20.04.1 via:

$ sudo do-release-upgrade

This fails as follows:

...

Could not calculate the upgrade

An unresolvable problem occurred while calculating the upgrade.

This was likely caused by:
Unofficial software packages not provided by Ubuntu Please use the tool 'ppa-purge' from the ppa-purge package to remove software from a Launchpad PPA and try the upgrade again.

If none of this applies, then please report this bug using the command 'ubuntu-bug ubuntu-release-upgrader-core' in a terminal. If you want to investigate this yourself the log files in '/var/log/dist-upgrade' will contain details about the upgrade. Specifically, look at 'main.log' and 'apt.log'.

The tail of /var/log/dist-upgrade/main.log contains the following error:

...
2020-10-11 18:36:30,320 WARNING Can't mark 'ubuntu-desktop' for upgrade (E:Error, pkgProblemResolver::Resolve generated breaks, this may be caused by held packages.)
2020-10-11 18:36:31,747 ERROR Dist-upgrade failed: 'Broken packages after upgrade: colord'
2020-10-11 18:36:31,747 DEBUG abort called
2020-10-11 18:36:31,755 DEBUG openCache()
2020-10-11 18:36:37,070 DEBUG Comparing 4.15.0-117 with 
2020-10-11 18:36:37,071 DEBUG Comparing 4.15.0-118 with 4.15.0-117
2020-10-11 18:36:37,824 DEBUG /openCache(), new cache size 99980

Here are the reverse package dependencies of colord:

$ apt-cache rdepends colord
colord
Reverse Depends:
  colord-sensor-argyll
  gnome-control-center
    colord:i386
  cups-filters
    colord:i386
  cups-daemon
    colord:i386
  cups
    colord:i386
  xiccd
    colord:i386
  krita
    colord:i386
  gnome-color-manager
    colord:i386
  foomatic-filters
    colord:i386
  dispcalgui
    colord:i386
  diffoscope
    colord:i386
  colord-sensor-argyll
    colord:i386
  colord-sensor-argyll
  colord-data
  colord-sensor-argyll
    colord:i386
  argyll
    colord:i386
  libcolord2
    colord:i386
  libcolord-gtk1
    colord:i386
  gnome-control-center
    colord:i386
  cups-filters
    colord:i386
  cups-daemon
    colord:i386
  cups
    colord:i386
  colord-data

I recently had a segmentation fault issue with mesa and installed a version from ppa:kisak/kisak-mesa. This might cause the upgrade problem now.

The following would happen if I remove the colord package:

$ sudo apt-get remove colord 
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following packages were automatically installed and are no longer required:
  apg colord-data gnome-control-center-faces gnome-online-accounts libcolorhug2 libnss-myhostname linux-headers-4.15.0-117 linux-headers-4.15.0-117-generic
  linux-image-4.15.0-117-generic linux-modules-4.15.0-117-generic linux-modules-extra-4.15.0-117-generic python3-macaroonbakery python3-nacl
  python3-protobuf python3-pymacaroons python3-rfc3339 ubuntu-system-service
Use 'sudo apt autoremove' to remove them.
The following packages will be REMOVED:
  colord gnome-control-center ubuntu-desktop
0 upgraded, 0 newly installed, 3 to remove and 8 not upgraded.
After this operation, 5,782 kB disk space will be freed.
Do you want to continue? [Y/n]

This does not sound like a good idea because last time I removed ubuntu-desktop the UI did no longer start-up. See linked post.

Question

How can I resolve the upgrade problem without risking the segmentation fault to reincarnate?

Related

• Ubuntu 18.04.4 startup stuck after package updates / mesa segmentation fault
• Can't seem to upgrade to 20.04 from 18.04 due to held back packages: libsensors4, libsensors-config, colord
• libsensors5 or libsensors-config upgrade problem
• Upgrade from 18.04 to 20.04 failed ...

I hope it's not making much difference that I'm posting my question here even if I'm using Pop!_OS. The problem also occurs on a Manjaro live usb.

I'm using a

• Ryzen 1500x CPU,
• 16GB of RAM,
• MSI B350m Gaming Pro motherboard with an RX 580 8GB version,
• Monitor is AOC G2260VWQ6 1920x1080x75hz.

After booting and logging in everything is fine, however, after turning the screen off and then on, going back from a screensaver, changing display settings that makes the screen "restart" or anything similar while being on a refresh rate over 60hz, which is ~75hz for my screen, the weird flicker starts happening.

Here are links with videos for different situations:

• Everything being fine after login: https://streamable.com/am3kp

• Screen starts flickering after locking (turns off the screen after a second) and then logging back in: https://streamable.com/meeiz

• Fixing the flicker by changing the resolution and reverting: https://streamable.com/5l5z2

Sometimes it still works fine, but it only happened once, restarting the screen multiple times didn't begin the flicker, until reboot.

I tested it using both DisplayPort and HDMI, with higher refresh rates on lower resolutions the problem still occurs. When switching to a different resolution and still using ~75hz the flicker doesn't go away.

It happens on both Mesa version 18 and 19, on Manjaro it still happened on even when switching to kernel version 5. Of course, the problem doesn't occur in Windows.

I'm not that advanced in Linux so I really don't know how to fix it. Please help.

EDIT: After searching a bit more I found a workaround. I had to change "auto" to "high" in file /sys/class/drm/card0/device/power_dpm_force_performance_level. Link to the thread I found the workaround on: https://bbs.archlinux.org/viewtopic.php?id=244398

I have an Nvidia 1070 system with Ubuntu 18.04, Nvidia driver 396.37, CUDA 9.0, and then TensorFlow and Keras running on it.

Having run sudo apt update, when I run sudo apt upgrade, I get the following:

$ sudo apt upgrade
Reading package lists... Done
Building dependency tree       
Reading state information... Done
You might want to run 'apt --fix-broken install' to correct these.
The following packages have unmet dependencies.
 libglx-mesa0 : Depends: libglapi-mesa (= 18.0.5-0ubuntu0~18.04.1) but 18.2.8-0ubuntu0~18.04.2 is installed
                Breaks: libglx-mesa0:i386 (!= 18.0.5-0ubuntu0~18.04.1) but 18.2.8-0ubuntu0~18.04.2 is installed
 libglx-mesa0:i386 : Breaks: libglx-mesa0 (!= 18.2.8-0ubuntu0~18.04.2) but 18.0.5-0ubuntu0~18.04.1 is installed
E: Unmet dependencies. Try 'apt --fix-broken install' with no packages (or specify a solution).

When I attempt the suggested sudo apt --fix-broken install I see the following:

$ sudo apt --fix-broken installReading package lists... Done
Building dependency tree       
Reading state information... Done
Correcting dependencies... Done
The following packages were automatically installed and are no longer required:
  libllvm6.0 libllvm6.0:i386 x11proto-dri2-dev x11proto-gl-dev
Use 'sudo apt autoremove' to remove them.
The following additional packages will be installed:
  libglx-mesa0
The following packages will be upgraded:
  libglx-mesa0
1 to upgrade, 0 to newly install, 0 to remove and 256 not to upgrade.
155 not fully installed or removed.
Need to get 0 B/135 kB of archives.
After this operation, 9216 B of additional disk space will be used.
Do you want to continue? [Y/n] y
(Reading database ... 591832 files and directories currently installed.)
Preparing to unpack .../libglx-mesa0_18.2.8-0ubuntu0~18.04.2_amd64.deb ...
Unpacking libglx-mesa0:amd64 (18.2.8-0ubuntu0~18.04.2) over (18.0.5-0ubuntu0~18.04.1) ...
dpkg: error processing archive /var/cache/apt/archives/libglx-mesa0_18.2.8-0ubuntu0~18.04.2_amd64.deb (--unpack):
 trying to overwrite '/usr/lib/x86_64-linux-gnu/libGLX_indirect.so.0', which is also in package nvidia-396 396.37-0ubuntu1
Errors were encountered while processing:
 /var/cache/apt/archives/libglx-mesa0_18.2.8-0ubuntu0~18.04.2_amd64.deb
E: Sub-process /usr/bin/dpkg returned an error code (1)

Based on my investigating this, I am prompted by this and this, to run the following command:

sudo apt-get -o Dpkg::Options::="--force-overwrite" install --fix-broken

However, I really would rather not screw up my setup (as it was a challenge getting compatible versions of CUDA, TensorFlow, Keras etc. working at all) so would value thoughts on whether this is a safe approach.

EDIT: Following this suggestion by @Jos, I made a backup of /usr/lib/x86_64-linux-gnu/libGLX_indirect.so.0 and the proceeded with the --force-overwrite approach. Following this, the infrastructure of TensorFlow, Keras and so on all seems to be working fine.

$ sudo apt-get -o Dpkg::Options::="--force-overwrite" install --fix-broken
Reading package lists... Done
Building dependency tree       
Reading state information... Done
Correcting dependencies... Done
The following packages were automatically installed and are no longer required:
  libllvm6.0 libllvm6.0:i386 x11proto-dri2-dev x11proto-gl-dev
Use 'sudo apt autoremove' to remove them.
The following additional packages will be installed:
  libglx-mesa0
The following packages will be upgraded:
  libglx-mesa0
1 to upgrade, 0 to newly install, 0 to remove and 256 not to upgrade.
155 not fully installed or removed.
Need to get 0 B/135 kB of archives.
After this operation, 9216 B of additional disk space will be used.
Do you want to continue? [Y/n] y
(Reading database ... 591832 files and directories currently installed.)
Preparing to unpack .../libglx-mesa0_18.2.8-0ubuntu0~18.04.2_amd64.deb ...
Unpacking libglx-mesa0:amd64 (18.2.8-0ubuntu0~18.04.2) over (18.0.5-0ubuntu0~18.04.1) ...
dpkg: warning: overriding problem because --force enabled:
dpkg: warning: trying to overwrite '/usr/lib/x86_64-linux-gnu/libGLX_indirect.so.0', which is also in package nvidia-396 396.37-0ubuntu1

I cannot get amdgpu to load as driver. Instead it always loads radeon.

Setup:

• Ubuntu 18.04.1 LTS

• i7-5830k

• 2x R9 390X

This question is directly based on a comment/question that can be read here.

I have a netbook with a Radeon X1200 series (RS690M) that's using the radeon driver.

Unfortunately, I'm experiencing serious graphical glitches that make my system unusable. Here is a video I took for illustrative purposes.

Blacklisting the radeon driver only results in an incredibly slow desktop which is equally unusable.

I tried doing what this question suggests to force LLVMPipe to be used (with or without the radeon driver blacklisted), but I experienced the exact same results.

How can I tell if LLVMPipe is being used in the first place?