
Flatpak Doesn't Work in Ubuntu 25.10, But a Fix is Coming (phoronix.com) 74
"It's not just you: Flatpak flat-out doesn't work in the new Ubuntu 25.10 release," writes the blog OMG Ubuntu:
While Flatpak itself can be installed using apt, trying to install Flatpaks with Flatpak from the command-line throws a "could not unmount revokefs-fuse filesystem" error, followed by "Child process exited with code 1". For those who've installed the Ubuntu 'Questing Quokka' and wanted to kit it out with their favourite software from Flathub, it's a frustrating road bump.
AppArmor, the tool that enforces Ubuntu's security policies for apps, is causing the issue. According to the bug report on Launchpad, the AppArmor profile for fusermount3 lacks the privileges it needs to work properly in Ubuntu 25.10. Fusermount3 is a tool Flatpak relies on to mount and unmount filesystems... This is a bug and it is being worked on. Although there's no timeframe for a fix, it is marked as critical, so will be prioritised.
The bug was reported in early September, but not fixed in time for this week's Ubuntu 25.10 release, reports Phoronix: Only [Friday] an updated AppArmor was pushed to the "questing-proposed" archive for testing. Since then... a number of users have reported that the updated AppArmor from the proposed archive will fix the Flatpak issues being observed. From all the reports so far it looks like that proposed update is in good shape for restoring Flatpak support on Ubuntu 25.10. The Ubuntu team is considering pushing out this update sooner than the typical seven day testing period given the severity of the issue.
More details from WebProNews: Industry insiders point out that AppArmor, Ubuntu's mandatory access control system, was tightened in this release to enhance security... This isn't the first time AppArmor has caused friction; similar issues plagued Telegram Flatpak apps in Ubuntu 24.04 LTS earlier this year, as noted in coverage from OMG Ubuntu.
AppArmor, the tool that enforces Ubuntu's security policies for apps, is causing the issue. According to the bug report on Launchpad, the AppArmor profile for fusermount3 lacks the privileges it needs to work properly in Ubuntu 25.10. Fusermount3 is a tool Flatpak relies on to mount and unmount filesystems... This is a bug and it is being worked on. Although there's no timeframe for a fix, it is marked as critical, so will be prioritised.
The bug was reported in early September, but not fixed in time for this week's Ubuntu 25.10 release, reports Phoronix: Only [Friday] an updated AppArmor was pushed to the "questing-proposed" archive for testing. Since then... a number of users have reported that the updated AppArmor from the proposed archive will fix the Flatpak issues being observed. From all the reports so far it looks like that proposed update is in good shape for restoring Flatpak support on Ubuntu 25.10. The Ubuntu team is considering pushing out this update sooner than the typical seven day testing period given the severity of the issue.
More details from WebProNews: Industry insiders point out that AppArmor, Ubuntu's mandatory access control system, was tightened in this release to enhance security... This isn't the first time AppArmor has caused friction; similar issues plagued Telegram Flatpak apps in Ubuntu 24.04 LTS earlier this year, as noted in coverage from OMG Ubuntu.
Re: (Score:1)
Re: (Score:3)
That seems like a cop-out argument.
I could just as well ask why this thing is designed so stupidly as to have a learning curve at all.
Re: (Score:3)
How would you suggest making it so it had no learning curve? With a pretty GUI? Then you'd complain that it's Microsoft-like and unsuitable for headless servers.
Re: (Score:2)
For decades AIX has used a utility called "smit" for system configuration and maintenance. What makes this program so great is how it will show you the exact command line argument it issued.
Re: (Score:2)
Re: (Score:2)
Re: At least it's not SELinux. (Score:5, Insightful)
Re: At least it's not SELinux. (Score:3)
Re: (Score:3)
That's either a badly behaved application or a misconfigured policy. If it's a badly behaved application, maybe the best solution is to just ignore the errors. I use SELinux to prevent some badly behaved applications from doing things I don't want them doing when they have no configuration option to turn the behaviour off.
Re: At least it's not SELinux. (Score:2)
Re: (Score:2)
To do the same thing in systemd you need a service file and a timer file and I don't think I'll ever remember the format of those without using a reference.
You won't remember the format of something you don't use and go out of your way not to use? I'm shocked I tell you. SHOCKED! By the way I have the same problem with cron.
But really the fundamental problem is you, that person who thinks that just because systemd is part of your system that you can't install crond. Do you even Linux bro? I mean we're not even talking about systemd and you're having a meltdown.
Re: (Score:2)
Re: (Score:3)
Re: At least it's not SELinux. (Score:1)
Re: (Score:2, Informative)
Re: (Score:2)
Re: (Score:2)
I still don't understand the point of KDE wallet. I'm logged into my machine with a password and now it wants another password when I run a web browser? I just delete the dumb fucking binary. Trying to uninstall will result in complaints and broken packages.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
It's an incredibly poorly designed and unintuitive system, has poor tooling and an archaic templating system at best.
It feels like it was designed by a single engineer in some backroom who has absolutely no concept of how you should design something that is used by a large base of users.
And the fact that "Well, if you don't know how to make templates for SELinux, then you are lazy and incompetent and you shouldn't be able to use that external USB device as it requires additional configuration to use!" is ex
SELinux causes me no issues (Score:1)
I run all my Internet-facing Linux servers with SELinux in enforcing mode, and it works fine. Even after major distro upgrades, it's usually just a matter of just installing the already-compiled policy modules for the machine and it's ready to go.
Re: (Score:2)
The problem with SELinux is that it interferes with Userspace first-party applications in a way that is completely enigmatic to the end-user...even someone who wants to try out the new advertised feature that Gnome/KDE released. Case in point : GNOME Remote Desktop [fedoraproject.org]
Re: At least it's not SELinux. (Score:1)
What is this RevokeFS filesystem ? (Score:1)
What is this RevokeFS filesystem ? Anyone has any idea, because I can't find anything on the net (except this exact error)
SquashFS, OverlayFS, ResilientFS i can understand but RevokeFS ? Maybe some new/re-branded journalling thing?
Re:What is this RevokeFS filesystem ? (Score:5, Informative)
It's a user-space filesystem driver that Flatpak uses to manage temporary storage, particularly during package installation and updates. As for what it does or how it works, who knows. All these application package systems are convoluted Rube Goldberg machines held together with duct tape.
Re: (Score:3)
It's part of flatpak: https://github.com/flatpak/fla... [github.com]
It seems the main functionality is to allow reading a particular directory tree, and that for all write activity it asks a daemon if this is OK. Then presumably the daemon checks some rules or even might ask the user if it is acceptable to proceed.
Wtf? (Score:5, Interesting)
"the Rust replacements for standard gnu utilities were busted."
I had to google this - I didn't realise people were wasting time and effort in a pointless re-write of utilities that have probably been debugged better than any other code out there!
Do these Rust zealots not understand that memory leaks are only one of many potential bugs and in a fire and forget utility that may only run for literally microseconds they're borderline irrelevant compared to logic errors? And if you rewrite complicated code in another language you're going to make those sorts of mistakes.
Re: (Score:2)
Re:Wtf? (Score:5, Insightful)
"It's not about memory leaks but memory safety"
Are there many safety issues in these utilities? I presume they've done code analysis and found loads then?
"and much better maintainability that legacy GNU C code"
BS. There's nothing special about Rust wrt maintanability plus there's a FAR greater pool of C devs than Rust devs who can check any new code.
Oh, and just because code is old, debugged and works doesn't make it "legacy". No serious devs are interested in the Shiny New Thing unless it does something incredibly special and useful. I'm not seeing that here.
Re: (Score:3)
Oh, and just because code is old, debugged and works doesn't make it "legacy"
Having personally looked at the old GNU C code, it did not appear particularly difficult to maintain.
Re:Wtf? (Score:4, Informative)
Are there many safety issues in these utilities? I presume they've done code analysis and found loads then?
Not many, but there are some discovered: https://www.cve.org/CVERecord/... [cve.org] and who knows how many undiscovered. And it's not only about the existing code but also adding a new code.
BS. There's nothing special about Rust wrt maintanability plus there's a FAR greater pool of C devs than Rust devs who can check any new code.
Well, even without it's built-in compile-time safety, Rust is a modern language with a lot of features missing in C (just naming a few like traits, generics, utf-8 strings, FP-style constructs, immutability, standard library, thread safety, async, etc.), and it is much more attractive for the new developers.
Re:Wtf? (Score:5, Insightful)
"And it's not only about the existing code but also adding a new code."
I doubt many command line utils will be updated anytime soon. Most of them have been a done deal for years.
"Rust is a modern language with a lot of features missing in C"
So is C++ and there are far more C++ devs than rust devs and they could have been rewritten in C++ years ago but sensible people realised that if it ain't broke don't fix it.
"ust naming a few like traits, generics, utf-8 strings, FP-style constructs, immutability, standard library, thread safety, async"
Wow, cutting edge! See C++.
"is much more attractive for the new developers."
So is Python, perhaps they should be rewritten in that if popularity is a reason.
Re:Wtf? (Score:4)
I doubt many command line utils will be updated anytime soon. Most of them have been a done deal for years.
Yes and not because "they are done", but also because few people are really interested in maintaining and expanding GNU code written 30 years ago in the language which isn't appropriate anymore to write anything beyond the very low-level OS kernels.
So is C++ and there are far more C++ devs than rust devs and they could have been rewritten in C++ years ago but sensible people realised that if it ain't broke don't fix it.
Or may be because C++ isn't fun to work with at all and it introduces no real advantages because it has the same fundamental issues as C.
Wow, cutting edge! See C++.
C++ has no traits, no thread safety, no built-in utf-8 strings, no proper macros, no immutability by default, it is still inherently unsafe even without using the raw pointers.
So is Python, perhaps they should be rewritten in that if popularity is a reason.
Why not, it will be an interesting project and I can only welcome it. I don't understand your point in complaining that some people are busy doing something that you don't like.
Re: Wtf? (Score:2)
So much of your post is just plain wrong that I really cant be bothered. Go learn modern C++ then get back to me. In the meantime I'll mentally file you in the ignorant idiot section.
Re: (Score:2)
"Yes and not because "they are done","
They are done. "ls" for example doesn't need to do anything new.
"Or may be because C++ isn't fun to work with "
Oh I'm sorry, I didn't realise the appropriate language choice was one that was "fun". I thought we were talking about adults here, not preschoolers.
"C++ has no traits"
Wrong.
"no thread safety"
Wrong.
" no built-in utf-8 strings"
Plenty of libraries do it.
"no proper macros"
Define proper macro. C had them, C++ has them + constexpr which you've probably never heard o
Re: (Score:2)
Are there many safety issues in these utilities? I presume they've done code analysis and found loads then?
The entire world thought that about OpenSSL held up as a shining example of oh so secure open source.
Question: Were you involved in that code analysis you're talking about? To what depth was it done? How well was it resourced? Were experts involved or was this some developer grading their own homework?
The problem with open source is that everyone *THINKS* someone is doing some detailed review, and ultimately very few people are.
Re:Wtf? (Score:5, Insightful)
It's not about memory leaks but memory safety and much better maintainability that legacy GNU C code.
Most people who complain about this have never actually looked at the legacy GNU code.
Re: (Score:3)
Re: (Score:2)
AppArmor or SELinux (Score:1)
Re: (Score:3)
I'm all for pointing out the foibles of windows, but I'm afraid security and firewall rules are standard in the corporate world and have been for decades. If you think any of the corporate *nixes like Solaris of HP-UX didn't have any of this then you really need to get yourself up to speed as to how IT works in a business.
Re: AppArmor or SELinux (Score:2)
Re: (Score:2)
Oh dear. When you leave your mums basement and go get a job in the messy real world with companies having to mitigate against idiots, lazies trying to cut corners and people with bad intentions get back to us.
Re: (Score:2)
maybe some day you'll meet an end-user... until then, have fun in your fantasy-land.
Re: (Score:3)
The trouble is, the POSIX model is broken. You need root to do all sorts of things, but once you get root, it doesn't just allow you to do the thing you need to do, it allows you to do anything. This gives bugs a lot more exploitation potential. SELinux lets you check that programs are only doing the things they're supposed to be doing, and not accessing something they shouldn't.
Re: AppArmor or SELinux (Score:2)
Re: (Score:2)
"The trouble is, the POSIX model is broken"
Its not broken, people just don't use it properly. You're supposed to use groups to give permissions for certain activities but its too much effort so , what the hell, just use root, right?
Re: (Score:2)
You need root to change your UID. This means that any service that needs to impersonate users needs a component running as root. SELinux mitigates the impact of vulnerabilities in that component that are exposed during the time between when it starts and when it changes to the target UID.
Re: AppArmor or SELinux (Score:2)
Re: AppArmor or SELinux (Score:2)
Re: (Score:2)
That minimal part is still a major target for exploitation, and checking commands is one of the areas most prone to vulnerabilities as we've seen numerous times. Having something to mitigate vulnerabilities in that minimal part is an advantage out here in the real world.
You recommend sudo, but that's a critical component where we keep finding vulnerabilities. It's also a nightmare to configure properly without inadvertently opening holes. It's mainly used for "allow these users to get root" but even that
Re: (Score:2)
You don't need to change your UID if you can do what you need using a specific GID. Looks like you're another one who doesn't understand the purpose of groups.
Flatpak: a solution in search of a problem. (Score:2)
$
$ make
$ make install
Re:Flatpak: a solution in search of a problem. (Score:5, Informative)
If an app isn't available locally just download the source tree and compile it yourself.
I've been compiling software for Linux since I first started running it 1994 and it has never been more complicated than it is today. Dependency changes five levels deep or more and you get compilation errors at multiple levels. Your dismissiveness is unwarranted and makes you look ignorant.
Re: (Score:2)
Plus, you apply one update and EVERYTHING breaks, so re-building any codebase, even without any changes, is a nightmare.
Welcome to the world of "living standards".
Re:Flatpak: a solution in search of a problem. (Score:5, Insightful)
Here is where things rapidly snowball. The developer used 50 different esoteric libraries that you don't have. Those libraries have their own dependencies. Software is such a shit show that they decided it was easier to recreate the one off developer environment instead of make something truly portable.
Re: (Score:2)
Re: Flatpak: a solution in search of a problem. (Score:2)
Re: (Score:2)
$ ./configure
$ make
$ make install
This is DLL - Hell Linux style. Virtually all completely broken Linux systems I've fixed (or attempted to fix) have been the result of straying from the curated path that distribution maintainers provide.
If this shit were so easy we wouldn't need distributions. You can very rapidly break things when you need to deal with dependencies of custom compiled and self-installed packages. If this isn't you, congrats, either you're lucky, a guru, or haven't tried hard enough.
Re: (Score:2)
Flatpak is about containerizing third party applications to add a level of security that wouldn't exist otherwise. It's not just another packaging system. The thing you're quoting works for insecure applications, but in theory someone can put stuff in the code that could have access to things it shouldn't.
That said, Flatpak (and its rival Snap) are both particularly bad solutions because it appears the starting point was "How can we containerize Firefox" rather than "How can we create a framework where appl
Re: Flatpak: a solution in search of a problem. (Score:2)
And this is why Microsoft got away with Windows 11 (Score:3, Funny)
Re: (Score:2)
Because Microsoft actually uses all the telemetry to fix problems
[citation needed]
Windows 11 is causing me grief daily, and the UI is provably worse than it was in 10.
Re: And this is why Microsoft got away with Window (Score:2)
Re: (Score:2)
If I could just transform my Windoze to run as a virtual machine under Linux, my boss probably wouldn't really mind...
I don't know how hard it would be to do with your existing install, but I have Windows 11 in a QEMU/KVM VM with a virtual TPM and it works fine...
Yup. Ubuntu problems with Every. Single. Release. (Score:4, Insightful)
Meanwhile Ubuntu 22.04 and 24.04 were just dogshit. After install I spent weeks tracking down various bugs and making basic stuff on my systems work. It felt like using Linux on the desktop circa 2001 again. To say nothing about the noticeable step down in virtualization performance from version 18 to 22.
This is why I left Ubuntu. Say what one wants about Redhat and RHEL. I don't agree with all their tactics in the past or their decisions on CentOS. But from a system usability standpoint they generally don't have these issues on initial release. That's probably because they dedicate the time and resources to actually test their software!
Re: (Score:2)
Next time I "upgrade" i will also get rid of Ubuntu on all my machines (6 of them).
Have not decided which one yet, but I think I have had it with Ubuntu.
Atomic packaging systems suck (Score:4, Insightful)
With atomic packaging systems, I feel like computing is in a weird place where it has taken on all of the bad characteristics that are both a step backwards towards the days of DLL hell and a step forwards to today where there is almost no tweaking and debloat attempted by devs because powerful hardware and memory make it unnecessary in their eyes.
I don't WANT 53 copies of the same library on my system. I don't want slow startup because applications to have to be unpacked before they can be used. I don't want applications that make it difficult or impossible to share data between applications. I don't want applications that don't theme properly. I don't want applications that weaponize themselves against the owner of the machine and try to dictate how the app is used like Windows apps do.
I simply do not want atomic packaging systems. I want traditionally installed apps that only put one copy of the library on my system and I want to be able to share data between apps easily and efficiently without having to install additional apps from the same package manager to be able to use them... for example for gaming, If you want to use Lutris and choose the version that comes from an atomic package manager, you also have to install a SECOND version of all your 3d libraries for your video card that are available to the atomic Lutris. Fuck that, fuck the entire idea of that type of system.
Re: (Score:2)
On one hand, I agree with you, I want everything done the old way and all my packages updated, and for all that software to still work.
On the other hand, that's not actually realistic, and a lot of people have to do a lot of hackery and patchwork in order to make it be like that.
I chose the version of Lutris that you get from their apt repo. If it installed extra 3d libraries, I didn't notice because disk space is cheap. I noticed it installed extra copies of runners available to Steam. I'm not thrilled abo
Re: (Score:2)
Things are updated every 5 minutes and we have "living standards" everywhere. It is simply not possible to have one version of a library shared by everything. Those days do not exist anymore.
Well snap does not really work either ... (Score:2)
that terrible package manager is still broken, struggles with home directories not on the root file system and sucks in many other ways.
Same with wayland: "modern" tech that is unable to do lots of things that are possible with X (related to remote desktops, running windows remotely etc).
The next update of my Ubuntu machines will actually a replacement with some other distro that does not shove all that enshittified limited or broken stuff down their users.