Linux this Week: Linus Says "Fork IT", Debian drops 32-bit, FreeBSD drops GPL, RUST is Back

[00:00:00]This week in Linux and open source news, we have Lionus Torvolts telling people to fork the Linux kernel or walk away from it. We're dropping legacy support and we're getting some Cosmic Desktop updates, including some big changes to the Linux kernel. Let's first talk about Debian. As with the release of Debian 13 called Trixie, Debian has officially dropped support for the 32-bit x86, also referred to as the i386 and i686 hardware, and a significant shift towards 64-bit computing as it's been the standard for quite a while now. This is a big deal as the Linux kernel and operating systems using it have often supported 32-bit. But now we're seeing 32-bit x86 CPUs becoming obsolete for modern desktop use. While this is good for developers, it might not be good for users as this does allow developers to focus resources on modern architectures while reducing the maintenance burden of supporting aging niche hardware like 32-bit processors. It's a big deal. For many years, Debian has supported and been the go-to choice for old computers that no longer are supported by Windows or other Linux operating systems. This change signals the final phase of a massive transition and will no doubt push other operating systems away from legacy hardware. It'll be interesting to see as this is finally official. Now, this is not a total drop. The Debbian is going to still support 32-bit ARM and other specific ports, but you will have to look elsewhere very soon as we have the final version support for 32-bit x86 until June 30th, 2028. And that's if you use Debian 12 Bookworm. Anyways, if you're looking for an alternative and you're still running older hardware, other distributions like Slackware, Void, Alpine Linux, and Gen 2 still all provide 32x86 support, but may require manual configuration. a big deal in Linux. Another big deal is that Lionus, the creator and founder of Linux, had something interesting to say this week as Lionus has clearly embraced AI as a tool for code review and allowing developers to use it in the Linux kernel in which Linus Torvville shut down conversations in the Linux kernel about banning AI. He considers AI to be just another useful development tool. And a decision here is that the kernel should be based on technical merit and not ideology or fear of new technology. So he says that developers are free to use AI if it helps but not required to use it as he does acknowledge some AI flaws as it has been creating extra work for maintainers but believes that the solution is to improve how AI is used and not to prohibit it. In this message he starts out strong by saying Linux is not one of those anti-AII projects and if someone has issues with that they can do the open source thing and fork it. He also says, "We're not forcing anybody to use it, but I will very loudly ignore people who try to argue against other people from using it. And no, AI isn't perfect, but Christ, anybody who points to the problems at AI had better be looking in the mirror and pointing at themselves at the same time." A strong message from Linusvolts. And we dive into this even further as I read the whole entire thing in which I'll put a link in the description below if you want to check this out for yourself. But as you can imagine, this had mixed review among the community as there was a surprising amount of people who thought that Lionus actually misunderstood the debate and conversation going on here. Many people online are arguing the fact that Linus responded to the wrong issue. Their interpretation was that this discussion was not that should AI ever be used, but should AI be allowed to automatically generate bug reports and patches that are clearly increasing maintainer workload and something that Lionus has already acknowledged in which these people are looking for an answer to this question as most people are saying that AI is fine as a tool but humans still need to remain responsible for the code that it produces and many people believe that AI should be used in the Linux kernel like advanced autocomplete assistant in every path. patch still needs to have human review and maintainers should not ever be flooded with AI generated spam code or as we like to call it slop. Of course, there's another camp where people are just strongly pro Lionus saying that AI is just another tool and that Linux has always adopted useful technology. Of course, you have people who are strongly opposed to this as well, questioning how copyright and licensing will be affected and how long-term maintainability, including AI generated code quality, will impact the Linux kernel. Either way, very interesting. If you want more on this, I do have another video. We go much more in depth. Let's keep going. As System 76's Cosmic Desktop environment made significant progress in the last 7 months since its initial release, as a team at System 76 reviews everything that has been happening behind the scenes, including new things that have been added. I go in much more depth about this whole deal in another video, but I'm going to highlight the major things that the team talks about, including frosted glass, which can help adjust the transparency of your windows and terminals inside of the cosmic desktop. This is something new that they recently added. It works pretty well. I have a demo of it. You can see this in here in see-through mode or nearly opaque. Anyways, they've also updated the environment with consistent corners.

[00:04:55]This was an issue as many applications were having inconsistent styles based on what was set inside of the desktop settings. We got a new workspaces overview search and a new cosmic system monitor which I check out as well. But the biggest highlights here that they've made include better gaming support, major upgrades to Cosmic Files, improved terminal tab, drag and drop, configurable shortcuts, IME support, and then you have settings improvements including better searching, cap lock, remapping, and VPN two-factor authentication options. We also have Cosmic Player improvements and thousands of little small polish fixes across the desktop, including smoother animations, better KD and Gnome compatibility, improved appllet screenshots, notifications, and the software store has been updated. This is showing how fast System 76's Cosmic Desktop is actually maturing. One thing that I absolutely love and a lot of people mention is just how much the team at System 76 and the developers communicate with the community and let them know what exactly is happening. I posted on Reddit under the Papa OS subreddit about what people actually love about Cosmic Desktop and people really like it because it's modern but not overly complicated. It's clean but not restrictive. It's also customizable but not overwhelming. And the window manager itself is powerful including having tiling support which you can easily go between. Also, a lot of people praise it for being fast and responsive and people really are attracted to that fast development process. For me, it's really about the excellent tiling support. It's probably one of the biggest selling points, at least for me personally.

[00:06:27]Cosmic lets you toggle tiling on and off instantly. Drag windows if you want while still being tiled and has a great blend or mix of floating and tiled windows. This makes it great for beginners to use shortcuts efficiently whether you're using the keyboard or mouse. Anyways, check out System 76's Cosmic Desktop if you haven't already.

[00:06:46]Let's move on. as FreeBSD just retired and completely removed the last remaining GPL license code from its operating system codebase. The final GPL component was a program called dialogue which has been replaced with BSD license alternatives. So what this means is FreeBSD is now entirely under a permissive BSD license. Why is this important? Well, now this simplifies the license. Everything in the base system now uses the BSD compatible licenses which are much more permissive than the GPL. Also, there's more freedom for companies to use it. Businesses can now use, modify, and redistribute FreeBSD's base system with fewer legal obligations and not having to release source code changes. This is a huge milestone for FreeBSD as it's been a long-term goal of the project, and they've finally reached this goal. Of course, the day-to-day experience using FreeBSD will not change. It's primarily a legal milestone rather than a technical one. But nonetheless, this is a big deal as this commit signifies the license change.

[00:07:46]Here's something interesting. Here's something interesting. The open source MVK Vulcan driver used with the Neuvo graphics driver is getting much faster.

[00:07:53]When comparing the performance data with Mesa 26.1 versus the 26.2 branch point, overall, there's good steady progress on games that are and they're closing the gap significantly. As we can tell here, the proposal is across the board increasing the frames per second. In games such as Total War Warhammer, you can see quite a significant delta in this deal, hitting around 80 FPS in comparison to before with Mesa 26.2 hitting around, let's call it 40 or 50.

[00:08:24]That's a big deal. as we see big improvements to Hitman 3, Subnotica, Recipe for Disaster, and many, many more games. As it means better gaming and open source drivers for those of us who prefer open- source software, we can expect better Vulcan gaming performance on Nvidia GPUs. Also, we've been seeing pretty rapid progress in MVK as it's matured quickly over the past few years, showing that the open source graphics stack is becoming increasingly competitive, which is great. This means we can be less reliant on proprietary drivers and more users can use the fully open source graphics stack instead of Nvidia's closed source driver. Some future improvements are also expected with ongoing work on Nvidia's new Novo kernel driver and continued Mesa's development should also help improve performance. It's great to see these improvements to the Nvidia graphics driver stack on Linux. with this week in plasma shadows for Steam and Discord and the team added a feature to easily assign running programs to specific CPU cores directly from the system monitor.

[00:09:24]This is pretty cool as one of the most notable features is Kwin now automatically apply shadow outline and cornering rounding effect to clientside decorated windows that lack these. This was an issue with programs such as theme and Discord windows. Now we're finally seeing it applied as it's just a quality of life improvement across a desktop environment. But what's bigger here, at least in my opinion, is now you can assign processes to specific CPUs or groups of CPUs in system monitor, also known as setting the CPU affinity. This means more control over performance. You can now pin demanding applications like games or video encoding to specific CPU cores without using terminal commands.

[00:10:00]This is of course useful for power users who want to improve performance and reduce stuttering. You can also isolate workloads and multi-core processors this way. This is fantastic. I love the fact that the KD team is focused on making it easier to fine-tune how your CPU is used. Of course, most of us won't need this feature day-to-day, but it's very handy for gamers and developers and advanced users who just want more control. So, I love to see this. Those are the two biggest things this week.

[00:10:26]Definitely wanted to highlight the CPU affinity feature. Great work to the KD dev team. And speaking about the team, I want to talk about Vlad's blog here in which he talks about KD Plasma 6.8 adding serverside drop shadows in the Kwin or the window manager in Wayland compositor used by the KD plasma desktop environment. There's been updates including adding this support where applications that don't draw their own window shadows now can actually have Kwin automatically add them making windows look more polished and consistent across the desktop environment. We already talked about this in the release for 6.8, but this is the back end. This is how we were able to gain the support by changing up the Kwin compositor. Anyways, this means better-l looking applications, things like Discord and Steam and other programs that previously lack proper shadow support will now blend in better with the rest of the environment, giving us more consistent appearance. And because it's part of the compositor, no app changes are required, and it improves older X11 applications. Cool stuff. And if you're interested, I'm going to post a link in the description below about how Vlad actually accomplished this. Here's something interesting and new. frame a new x11 server written entirely in x8664 assembly language which much of the code was generated using clawed code. It has no external libraries. So it doesn't include things like MSA, libby or xlib and it talks directly to the Linux kernel, graphics hardware and input devices. Why? Well, the developer is saying that this creates many optimizations including much lighter CPU usage than XOR while idle and is stable enough to use on daily desktops. This is quite interesting as there's a mixed reaction here. It is an impressive technical experiment as writing an entire X11 server in assembly years ago would have been extremely unusual and difficult. But now with AI, it shows how we can optimize things that we previously couldn't, including writing things in assembly, which would have been a monumental task. The project here demonstrates that AI can assist and create and maintain very low-level system software that would otherwise be impractical to write alone. Regardless of this, it has over 20,000 lines of code and is primarily a proof of concept and personal project, not something that most Linux users would ever be expected to use or actually install. And it's going to be very hard to get it there because people have to maintain this thing. So, how do you maintain this if you're not used to writing assembly code? Do you think you could just slop it with AI the entire time and keep a project maintained with AI while not quite understanding the low-level assembly instructions? It'll be interesting to see how this all plays out as of course it has potential performance benefits. Like I said, significantly less CPU while idle and maybe some power consumption optimizations. As the maintainer says, but is it stable? Stable enough that I daily drive it right this post on it and only occasionally when something breaks or I want a feature, I turn to my buddy Claude and describe the itch. He never gets tired. It's not opinionated and it turns out to be a really good teacher. I know more about hardware layers, cursor painting, GPU handoffs, and event watchers than I ever planned to.

[00:13:26]Interesting. What do you think about this type of development using AI? Is it good? Should more projects approach optimizing their code with AI? Let me know in the comment section below. As there was a serious performance regression in kernel version 7.0 0 of the AMD GPU stack. This update that Ubuntu released contained a bug that made AMD GPUs 42 times slower for compute tasks like AI image generation using Rockm. The bug did not affect gaming or normal desktop use and has already been fixed upstream, but Ubuntu is now shipping the affected kernel anyway because it contains important security fixes and a performance fix will arrive at the next kernel update.

[00:14:06]This is a big impact for AI users. If you use an AMD GPU or four AI workloads, your job could take minutes instead of seconds. Most of the users will not notice this as gaming, web browsing, and general desktop use are not affected by this massive slowdown. And the weight here, security versus performance, well, Ubuntu chose to release the kernel because it included important security patches even though it temporarily contains this massive regression. A fix is already available. The Linux kernel developers have already patched this bug, but it's going to take a while as Ubuntu has to create a newer patch in order for people to actually update the kernel. For people who want to mitigate this in order to make sure that you don't get this performance regression, affected users may need to stay in their current kernel version until kernel update is released with the proper fix, meaning don't update the kernel currently. Very interesting and surprising as this made it all the way through to getting released by Ubuntu.

[00:14:59]And here's an interesting one. As Bcash FS, a modern Linux file system created by Kent Over Street, which aims to combine the best features of many file systems, including stability, performance, check sums, and snapshots, and RAID like features with compression and replication. Ken spent over a decade in development before finally merging into the Linux kernel 6.7. In 2024, Bcash FS entered the kernel. problems started appearing and not necessarily because the file system was broken but because disagreements on how Bcash FS was being maintained. The biggest issues were Kent often submitted very large patch series which were difficult to review and testing didn't seem to be taking place. Linusa became increasingly frustrated with the maintenance process and the relationship broke down as vcash FS was removed from the mainline Linux kernel and shifted to an autotree module. This means that users could still install it separately, but no longer shipping as part of the official kernel and now being used on the sidelines. Although Kent has been hard at work with the performance release, a recent blog post, we get a big announcement as Kent says that it's been a long time since he's been able to post. But to catch people up, we're no longer experimental. I took the label off the website a few months ago. I think based on the usual, the incoming bug reports are slowing down and looking a lot less serious and easy to get through than they were. Consider this the belated official announcement. We've been expecting this for quite a while now, but it's official. As Kent is saying here that they've removed the experimental label because bug reports are slim to none, serious issues are very rare and he considers the file system ready for general use. This is a massive milestone after years of development and a rocky relationship being included in the Linux kernel. The biggest things that were focused on in the last few months were performance improvements including much better throughput. But one of the biggest announcements here is that BK- FS user space code has already been converted to Rust and the work to include safe Rust interfaces for the core B tree iterator API. But what perhaps is the biggest long-term announcement here is that Kent is planning to gradually rewrite parts of Bcash FS and Rust. He argues that it'll improve memory safety, make the code base easier to maintain, attract new contributors, and eventually enable formal verification of critical file system code. This is a big deal as we don't have as many people learning C anymore. Even better, it opens up formal verification. A lot of things happening in Bcash FS. Will it ever get back into the Linux kernel? We started towards the beginning with Lionus. We're ending here with Kent. Let me know what you thought about all the news in the comments section below. Make sure to subscribe below and smash that like button on your way back up. Catch me and a great community on Discord and I'll catch you in another video. Thanks for watching.

[00:17:50]Linux can be hard to understand, but I take the most commonly used terms, commands, and subjects in Linux, and I break them down into simple to read documents, including Linux terms, flashcards, a checklist, a cheat sheet, and a mind map. And if you're ready to level up your Linux experience and knowledge, go to savvynick.com now and get access to these sheets.