If you install Ubuntu on a computer with secure-boot enabled, then it will probably boot. And maybe that’s all you want.
If so, you probably won’t be concerned about what I am describing here.
However, secure-boot is supposed to verify many of the steps in the boot path. And that’s where I see Ubuntu as broken.
I’m basing this on tests that I have done on Ubuntu-17.04 and Ubuntu-gnome-17.04.
First a brief summary of the problems that I am seeing:
- Ubuntu will boot without checking a signature on the kernel.
- Under some circumstances, Ubuntu will complain about a bad signature, and refuse to boot, even though secure-boot has been disabled.
I have already reviewed Ubuntu-17.04. However, the Ubuntu folk (i.e. Canonical) had already announced that, starting with 18.04, they would switch their mainline version from the Unity desktop to the Gnome desktop. So I decided to also test out the 17.04 version of Ubuntu with Gnome desktop.
I installed Ubuntu-gnome in an already existing encrypted LVM. The machine that I used actually has two hard drives, with an encrypted LVM on each drive. So this was a different LVM from the one that I used for the mainline Ubuntu (with unity). Currently, both versions of Ubuntu are installed on that machine.
Ubuntu 17.04 was announced a few days ago. I had already decided that I would install it, and do a little testing. So, once I saw the announcement, I started a download.
To download, I followed the links from the announcement to the download page. From there, I selected the torrent download. I was using the “vivaldi” browser, and it gave me several options with the torrent link. I chose the option to open the file. And that started the download with “ktorrent”.
I also downloaded “SHA256SUMS.desktop” and “SHA256SUMS.desktop.gpg”. Next, I checked the gpg signature with
gpg --verify SHA256SUMS.desktop.gpg SHA256SUMS.desktop
which showed that I had a good download of the checksum file. After the torrent download had completed, I checked its validity with
sha256sum -c SHA256SUMS.desktop
That reported that the downloaded iso file was ok. It also reported that some files did not exist. I ignored that. It was just that the checksum file had checksums for other isos that I had not downloaded.
Recently, openSUSE Tumbleweed updated Gnome to version 3.24. So I decided to give it a test. And I ran into some “problems”. This post will discuss those problems.
Gnome under Wayland
My oldest Tumbleweed system was installed in November 2014. And that’s where I first tested Gnome. At the time, I was using “sddm” for logins. Selecting “Gnome” on the login menu gave me a Gnome session running under X11. This was as expected. There was also a menu item for “Gnome-Wayland”. Selecting that gave me Gnome running under Wayland for managing the graphics.
I later switched to using “gdm” as login manager. And, with “gdm”, selecting “Gnome” gave me a Wayland session.
Next, I tried on my laptop. Tumbleweed was installed there on March 14 this year. I was already configured to use “gdm” for logins there. But, try as I did, I was unable to get a Wayland session for Gnome.
[Update: See update notes at the end of this post]
I still have Solus installed, although it is not something that I regularly use. Check HERE for my report on installing.
Today, I booted into Solus. And then I did a check for updates. It showed several. And one indicated a kernel update, but that was attached to a strange package name.
I applied the updates. And then I checked to see what kernel version had been installed. And there was no kernel there at all. The kernel had been deleted. Fortunately, I had a backup copy of the kernel, so I copied that into place. And then I rebooted. And the system froze on the login window. Perhaps all of the kernel modules had also been deleted, leaving a crippled system.
So it looks as if my Solus installation is now broken beyond repair.
I’m now wondering if this was a bad April fools joke. Or was this an intrusion into the Solus site to generate a faulty update?
At this stage, I am puzzled and looking for more information.
It looks as if I misunderstood the situation.
Apparently Solus had changed its naming convention for kernels. So I simply did not recognize that there was a kernel, since the new name was not even close to what I was looking for.