Bazzite would shut down if Fedora goes ahead with removing 32-bit | Hacker News Reader

Bazzite would shut down if Fedora goes ahead with removing 32-bit

gamingonlinux.com

72 pointsspeckx58 comments

if i understand this correctly, bazzite is like SteamOS but based on fedora. it needs 32-bit support to run wine and steam.

i find it hard to believe that fedora will drop 32-bit support without ensuring that wine and steam will continue to work. why would they do that? that would not make sense at all.

but there is a lot of backslash on the change discussion: https://discussion.fedoraproject.org/t/f44-change-proposal-d... so i doubt this will happen until those issues are resolved.

but i am quite confused:

the example given, that eg cpython might drop 32-bit support, should not affect supporting 32-bit for applications that need it, like wine and steam.

i thought that an application that doesn't support 32-bit mode would only prevent it from running on 32-bit machines or running on an OS in 32-bit mode on 64-bit machines, which was dropped 5 years ago. so where is the problem here?

this article explains a bit more.

https://ostechnix.com/fedora-32-bit-i686-support-withdrawal-...

currently only kernel and installers for 32-bit are dropped, but all packages are still built for 32-bit. why? that too doesn't make much sense either. are they still supporting running on 32-bit systems with an old or custom kernel? what for?

so that seems to be the problem then. somehow fedora is still supporting running on 32-bit systems. well, yes. do change that. but that doesn't mean that multilib support needs to be dropped. is there a problem to limit the build of 32-bit packages to only those that are needed?

The proposal has been retracted, but I fully expect 32-bit to be removed in time.

https://discussion.fedoraproject.org/t/f44-change-proposal-d...

What happens to old software? How do you run 32-bit Linux apps? How do you run 32-bit Windows apps like VST plugins (audio effects)?

Use VM to run old distros or old Windows.

You don't need a VM, just a container.

Steam runs all games in containers anyway: https://gitlab.steamos.cloud/steamrt/steam-runtime-tools/-/b...

Containers will not help if the kernel will not run 32-bit binaries.

Why would the kernel not run 32-bit binaries? They're talking about removing libraries only, not removing 32-bit support from the kernel.

For Windows apps there's: https://en.wikipedia.org/wiki/WoW64

you (fedora that is) build the libraries needed by 32-bit application. you stop building everything else, but you keep multilib support. that's what this proposal got wrong. it should have been only about stopping to build the 10.000 packages that nobody needs for 32-bit support.

Seems like an extreme response… other distros still exist. Plus they can still ship the 32bit binaries themselves. Fedora isn’t preventing you from using 32bit libs- fedora is just saying “hey we aren’t going to invest energy in maintaining something just 5% of our users use.”

I think the fundamental problem is that there are many different 5% groups and that you are very likely to be in some 5% groups.

In some way I feel dropping 32bit is capitulation to bloat.

I for one use 32bit libs in x86 32bit mode quite a lot to test software I build for memory bugs such that it works on actual 32bit systems.

32bit support isn't just bad bloat due to larger pointers.

Only doing 32bit and not doing amd64/x86_64 is also leaving a lot of performance on the table:

x86 traditionally has very few registers to work with, amd64 gives you quite a few more general purpose registers, so your hand-fiddled assembler will be faster, and if you use compiled code, your compiler will have an easier time with optimizations, faster function calls due to more register arguments, less register-to-memory spills and more execution station parallelism.

Memory use is also an important point, your 32bit application will be limited to at best 4GB per process, usually more like 2GB (due to the kernel/user boundary at 2/2GB, absent weird tricks). With games currently being multiple times that size, you have to swap out assets and load new ones very frequently, which is why loadpoints are getting more common and level design has to add more corridors and doors to support those load points and segment level data into 2GB chunks.

And the newer processor features, especially SIMD stuff like AVX and SSE work better with 64bit registers, if they work at all. Many support libraries that can utilize SIMD only do so in 64bit versions. Some features aren't even available if the hardware is in 32bit mode. And compilers can do automatic SIMD conversions, but those work far better (if they work at all) in x86_64 mode because of the insufficient number of registers in 32bit mode for copying around and splitting up stuff from a SIMD register.

I think the real reason that especially game developers are opposed to amd64 is that they used to be stuck on 32bit on Windows for a long time and now have to be dragged into the future by force ;)

I don't think this is very much about modern projects. Developers want to use 64-bit. It's more that removing the libraries breaks legacy 32-bit stuff that simply can't be updated.

The steam client should be upgraded, though. I understand it has to do with their ancient web view library not doing 64-bit, so they would have to change it and possibly mess up some of the layouts and overlays. But they went ahead and changed it on macOS anyway, so you know it can be done.

Isn't Steam (or more specifically steamwebhelper.exe) using Chromium Embedded Framework for web views? It's already a 64-bit process for me on Windows.

I also do wish they'd upgrade the client completely, but WoW64 isn't going anywhere. It's probably discussions like this on Linux side that would move their hand, if anything.

There is x32 if you don’t need more than 4 GiB address space but still want to use the benefits of amd64.

Gives you ~10% better performance due to better Cache efficiency

Yes, but that would be another set of libraries and another recompilation. Sticking with 32bit code is mostly to avoid that.

No, they said: we don't want to build all 32-bit packages no one is using. libs which are used by games and such will be kept. So they need to keep gcc, glibc and binutils for 32bit also. Just not multilib, i686 cross is good enough

Read the discussion. They plan to remove those libraries, too. Their current automated build system apparently requires they build i686 for every x86-64 package. They don't want to spend the effort to adjust the build system to just build a selection of i686 packages. They plan to drop the whole architecture.

It seems like these developers have tunnel vision. It's like when pulseaudio came about and the whole focus was streaming Bluetooth audio. It took over the system audio and broke everything else (games, coincidentally). They know this will break some things they don't personally use, and they don't care.

Are there stats on the number of people who are still running i686?

An interview with the Bazzite maintainer https://youtu.be/XgabGSI82M0

Bazzite maintainers have access to the source code of these libs, they can totally compile and distribute 32bits versions of them.

Title should really be: Bazzite maintainers don't want to make a slight effort to maintain stuff.

It's absolutely this. They have the ability to put these into their build pipelines and include them. Or just grab the latest 32bit versions and just use those.

It's work for them surely. But it's not the end of the world.

I think we have all solved much harder problems as software engineers.

Also their days are counted as SteamOS is now going to other handheld devices apart from the Steam Deck so... Not sure their project will still exist one year down the road.

I use Bazzite on my Steam Deck for better Btrfs, Decky Loader and Waydroid support. Support for these things on SteamOS via community projects are fragile (Decky Loader breaks often with Steam updates, steamos-btrfs runs a payload that may or may not work after updates, the widely used Waydroid installer comes with prebuilt kernel modules which I am not okay with). Bazzite has purposes other than installing SteamOS on unsupported devices.

"slight effort" seems disingenuous, given this quote from the source thread:

> If this shipped today, to accomplish this we’d need to:

> 1: Build all of the packages Steam needs in a copr or similar

> 2: Build a custom version of Proton with experimental Wow64 features turned on

> 3: Instruct all users to manually change their steam settings to use that proton version because the ones Valve provides don’t have that.

> 4: Tell anyone using OBS game capture that they’re just gonna have to give up.

Why would OBS not work?

or they can't?

Bazzite seems to be just a big Dockerfile built on top of bootc and bootc image builder right?

https://github.com/ublue-os/bazzite/blob/main/Containerfile

Nothing is stopping apps from shipping with 32 bit libraries if they need them. This isn't like what happened with ARM where processors dropped support for 32 bit code.

Obvious stoppers: man power, expertise, incompatibilities,...

Most 32 bit dependencies would not need to be upgraded. The only one I can think of off the top of my head is for the graphics driver.

Which is typically closed source, right?

Only for NVIDIA cards, which are the worst option for a Linux system anyways

That's a them problem...

Actually, there are some things.

libGL/libVK for OpenGL and Vulkan support is a big one, you need that to be fitting the exact version of your kernel Nvidia driver if you are using Nvidia cards. Of course all other vendors play nice with Open Source and Mesa libs, but especially for gaming, a lot of people unfortunately still buy Nvidia. Games cannot just ship their own 32bit libGL for reasons of license and versioning.

This wouldn't work well with Mesa, either.

On Windows, this is like telling game developers that they need to ship all the graphics drivers for their games. Let's skip the hassle for the developers' side. For the users, not only would that be a situation where you could no longer update your own drivers to fix problems, it would be a compatibility nightmare and likely introduce new problems because of interactions with the system drivers and kernel level.

That's always a good choice but

1. They need to also keep the multi arch support enabled in their kernels 2. Apps such as games and steam depend on Mesa, and you don't really want to ship Mesa

A good compromise would be for Valve to support Flatpak for Steam, but if they can't be bothered to build a 64 bit Steam they sure don't want to bother with that

1. The article is clickbait, leaving out important but complicated context of what sort of process is happening. If you want to understand what’s actually going on, watch this instead of getting alarmed at the article title: https://m.youtube.com/watch?v=XgabGSI82M0

2. These kind of change proposals are just that. Proposals only, by whoever wants to make one. Fedora is not about to suddenly drop 32bit support. Proposals are for discussion and consideration of how changes would affect things, they are not acceptance of the change or even a commitment to do so.

3. The comment on shutting down from the Bazzite founder is about what would happen if the proposal were accepted right now. Which isn’t happening. Outlining the consequences of a proposal is exactly what needs to happen. Dropping 32bit support would be catastrophic to multiple projects. But it’s not happening right now, and Bazzite is not about to shut down.

4. Things on both Fedora and Bazzite’s side are taking place such that removing 32bit support is reasonable and doesn’t break everything. This will take some time.

5. The proposal was badly worded, and is being redone.

For fucks sake.

but these points were obvious from the start. ubuntu went through that before and did it right the first time. like they actually talked to valve about it. this proposal was not thought through and should not have been made in this form in the first place. i am not involved with these decisions (but i use fedora and would be affected) but i would get annoyed if people kept making proposals that they don't think through. therefore i think the response is justified, because how otherwise do you stop bad proposals?

I’m not against the pushback on the poorly made proposal.

I’m against people (in this very thread) taking the resulting out of context comments from Bazzite as a sign that the project is fickle or doomed.

ah, sorry, i didn't read that from your comment. yes, i agree, that's a weird conclusion

So they want Fedora and/or Valve to do their work for them. Gotcha.

Reply on news.ycombinator.com