Untitled Linux Show 154 Transcript

Please be advised this transcript is AI-generated and may not be word for word. Time codes refer to the approximate times in the ad-supported version of the show.

00:00 - Jonathan Bennett (Host)
Hey, this week is a lot of fun. We're talking about the real-time kernel in Ubuntu 24. We're talking about the new UA link from AMD and Intel. We cover R&B and Nitrix and Rhino Linux. Then there's a wrap-up of the kernel 6.10 pre-release. Oh, and hopefully the end of the XZ backdoor story with finally, the release of a new and unadulterated version of XZTools. It's a lot of fun. You don't want to miss it, so stay tuned.

00:33 - Jeff Massie (Host)
Podcasts you love From people you trust. This is Twit.

00:42 - Jonathan Bennett (Host)
This is the Untitled Linux Show, episode 154, recorded Saturday, june 1st. The Elastic Wasteband of Linux. Hey, it is Saturday and you know what that means. It is time for the Untitled Linux Show, for some unbridled geekiness. We're going to geek out, we're going to talk Linux and open source. We've got Ken, we've got Jeff. Those other two guys are playing hooky today. They had lame excuses like graduation and work and stuff like that. That's lame. We're here to talk about Linux and we've got some fun stories to cover, and I think actually, jeff is going to kick it off with some Ubuntu news. What we don't do Ubuntu here, do we?

01:28 - Jeff Massie (Host)
Once in a while I know, and me of all people, right, you're just surprised. So Canonical just released the real-time kernel for 24.04. And before we get into the story, I think we should talk about the three versions of the kernel. There is the normal generic, there's a low latency and there's the real time, and they're all a little bit different design for different things. The generic kernel is what it sounds like. This is the normal kernel, which is suited to most general things that a person would do with their computer. You know, when you're loading on whatever distro, that's 99% chance you're probably getting the generic kernel, unless you're getting a special case distribution.

Now a listener might be thinking wait, I would love the kernel to go faster, so I must want a low latency kernel. Well, you would be correct in certain situations. While you have low latency kernel, you're getting quicker responses for certain things, but you're also losing overall throughput versus the generic kernel. The low latency kernel gets used for a lot of things, like audio recordings or things that need to stay in sync. A low latency kernel reacts very quickly. For example like suppose you needed to allocate a large chunk of memory. Like suppose you needed to allocate a large chunk of memory. But if you did it all at once and allocated all the pages at the same time it would take a while. Now when I say a while, this is computer CPU time frame a while, but a little longer than you might want. But in a low latency kernel if you allocate a few pages at a time you stay very quick to reacting to things coming into the kernel. So you can still have low latency. But it will take you longer overall to allocate that memory. So while the kernel can react quickly, the throughput of data is lower.

Now real-time kernels in general are not used for normal distributions and generally are found in distributions or loaded into distributions when very specific workloads are needed to be processed before everything else. So real-time kernels perform overall worse than generic. But when you have something critical that needs processed as soon as the workload comes in, that's where you need a real-time kernel. They're used a lot in automotive or medical applications. Data centers can also use them when a database load needs to be processed above all things. But real-time kernels work best when they're also matched with a distribution which has been tuned for the real-time kernel. So any advances you might get with the kernel could be lost to other parts if the system is not set up correctly. And the real-time kernels are also very deterministic in the way they handle specific events. If missing a deadline would result in a catastrophic failure, that's when you'd use a real-time kernel. So when I say very deterministic, you know how it's going to handle inputs from certain things. It's going to be the same every time.

Now I've oversimplified it and probably made a few technical errors for those kernel hackers in the audience, but the overall idea is correct. So we're just, you know, skimming the higher level here to just give people a general idea. Now back to the story. In the show notes, ubuntu 24.04 has added the preempt underscore RT patches to the 6.8 kernel and if you need it, you can get it through Ubuntu Pro, which is free for personal use.

The real time kernels also available for the Raspberry Pi four and five. So if you really wanted to have a play thing, you know you could or if you had a specific need for the Raspberry Pi that had a mission-critical purpose that it was fulfilling, the articles in the show notes tell you how to install the kernel and they also mention that the real-time kernel does not support the proprietary NVIDIA graphics driver. But those of you who really need it or want it, it's available to play with. One last note if you question if you need it or you think you might, you don't need it, you know, it's kind of like it's the same way of asking you know what's this menu item cost?

Well, if you have to ask, yeah, you can't afford it, you don't want to know. So real-time won't make normal PC use cases better, it will make it worse.

05:50 - Jonathan Bennett (Host)
But for you special needs people that have those certain niche workloads, it's now available for your enjoyment your enjoyment, I I would say that the one sort of exception to that rule where you think you might need it but you're not sure if you're doing audio stuff, um, in some cases running live audio, uh, you can actually benefit from the real-time kernel, um, but that's not the only way to handle that. So you, you make a valid point and it's neat to see the RT patches are still. You know, they're out there, they're being used and I've got to say the list of RT patches it's not nearly as much as it used to be. There's been a lot of good work done over the last couple of years to get a lot of that stuff merged and hopefully the rest of it soon, and it'll just be a different compile option rather than a set of patches.

06:47 - Jeff Massie (Host)
Right, and while I said NVIDIA proprietary does not work, there's mention in other places of possibly other hardware that might not work, so it's not as hardware friendly as the generic kernel. Yeah, and let's keep that in mind.

07:02 - Jonathan Bennett (Host)
What's going on there is, drivers are not necessarily written to be interruptible in the same way that the core kernel code is, and that's something I ran into back when I tried running the RT kernel. Just certain drivers and NVIDIA was one of them just did not like it. So and what'll happen is if the rest of the patches finally land, then that will sort of force the rest of the kernel, the drivers, and maybe even NVIDIA, we'll see to sort of start to play ball with the idea of interruptible kernel code. And you know, you get a lot of those bugs suddenly you discover and people have to work them out. Very true, Yep, Very true, Yep. Ken, I have been messing around with ARM devices recently and I have been trying to figure out how to get Debian or really anything other than a specific Ubuntu release on a particular bit of hardware. And there's a Debian-ish, Debian-based solution out there just for ARM hardware, isn't there?

08:03 - Ken McDonald (Host)
Yes, and Marius Nexcher even wrote about it. The new Debian Ubuntu based Armbian 24.5 code named Havier Havier introduces support for new devices including the Orange Pi 5 Pro, the I'm sorry if I'm mispronouncing this Radza Rock 5 ITX, the all-winner T527 Aviato-A1. Then there's the Radza Zero 3E, and they also have their 03W, the friendly ELEC CM3588, the 4G Phytem Pi, the Secura Pi RK3308B model, the SK-AM68, the TQM A8 MPXL and the CoolPi CM5 EVB. Armbian 24.5 also brings support for the latest Ubuntu 24.04 long-term support operating system series, support for the KDE Neon Desktop to the Armbian Jammy images, support for EMC and NVMe and USB booting to the Kodas Vim 1S or Vim 4 boards, support for the mainline Panther driver to the Linux 6.1 kernel, as well as support for the Linux kernel 6.8 on selected images. Unfortunately, as everybody probably already knows, the 6.8 kernel is now marked end of life at kernelorg. Rbn 24.5 also improved support for devices that were already supported. I'm going to recommend reading Marius' article unless you do like hearing me stumbling over all those manufacturers' names and model numbers.

10:13 - Jonathan Bennett (Host)
So it's fun. There's a lot of these devices that are out there that this supports and it's kind of neat because that brings sort of a standardized operating system for much of this ARM hardware. Unfortunately, the hardware that I've got that I'd really like to see is the Turing Pi RK1, neat little board not yet supported by this, which I think it will be soon because they're actually working pretty hard getting all of the bits upstreamed into the upstream kernel and there's also bits upstreamed into the upstream kernel and there's also for the RK1, they've also got some UEFI support, which this is a thing that I didn't realize until recently. For ARM there is a way to do UEFI, the Unified Extensible Firmware Interface, which that sort of gets you away from having to fiddle around with DTBs, the device tree binaries, and yeah, that's probably for like end user ARM. That's probably going to be the way that things move, to get to the point to where on ARM you can just grab an ISO and go, as opposed to having to get you know an image that is particularly made for your arm board, which is really it's a pain and it's a big maintenance burden for distro maintainers too. So we are getting close to sort of this kind of unified. We're moving towards, um well, uefi becoming the standard for booting on ARM as well as x86. And that'll be helpful.

I also find it interesting I looked earlier at the Armbian site and one of the things that they have set up. It's like do you want support? Do you want your board to be part of our testing matrix? Well, host it somewhere and give us an IP address. And oh, by the way, if we brick something too badly, you're going to have to reimage it. But that's fine, you know. And then it's like give us access to it and we will do nightly builds, push them on there, we'll do nightly builds, push them on there, run testing. I'm like, oh, that's actually pretty cool. I may have to contact them about this, because I have some of these lying around that's not doing anything yet, so we will see. All right.

12:38 - Jeff Massie (Host)
Anything to add about?

12:40 - Jonathan Bennett (Host)
that, jeff, I do not. Arm's not really your thing, you're more of an x86 guy.

12:44 - Jeff Massie (Host)
Yeah, arm and Pi and just not my thing. You know I think they're pretty cool, but for me, you know, even though they talk about, well, the efficiency and they're doing this and it's so much better. You know the x86 is just like a supercharged big block engine. You know it might not be the most efficient but it just the sheer brute force it has, is kind of more in my realm.

13:13 - Jonathan Bennett (Host)
Yeah, do you have a Raspberry Pi? Have you played with one yet?

13:18 - Jeff Massie (Host)
I have not.

13:25 - Jonathan Bennett (Host)
I don't own one. We need to go fund Jeff a get a Raspberry Pi fund so that he can play with one and experience the awakening that is running Linux on ARM.

13:34 - Ken McDonald (Host)
Don't it be? What about?

13:36 - Jonathan Bennett (Host)
10 copies Something like that. Yeah. Yeah, you can get a decent Raspberry Pi for 10 copiesees, yeah, although it gets a little more expensive. But if you really want an almost desktop experience, the Raspberry Pi 5 with one of the NVMe hats and NVMe drive is really pretty impressive. It really is pretty good.

14:07 - Jeff Massie (Host)
Wife of Jonathan said or just send him one of yours.

14:09 - Jonathan Bennett (Host)
No, no, all of mine that are worth sending to try to play with are actually in use for things. So I've got the original Raspberry Pi. I've got a Raspberry Pi 2. I probably could come up with a Raspberry Pi 3B+, but that's old tech now, raspberry pi. I've got a raspberry pi 2. I probably could come up with a raspberry pi 3b plus, but like that's old tech, now he needs a four or a five to really sort of get the uh, the full experience of what raspberry pi is up to these days I know what would convince him to get one oh, what's that?

14:39 - Jeff Massie (Host)
if somebody sent him a 3D printer and I got friends with them, but I don't have any of those either.

14:50 - Jonathan Bennett (Host)
Yeah, they're fun, especially once you get bit by the with 3D printers. For me at least, what really bit me was I can 3D print things to make my 3D printer better, and that was just kind of the end for me. It's like, oh, I'm hooked on this. But speaking of Raspberry Pis and 3D printers, you know, one of the first things I did is I printed a mount to be able to mount a Raspberry Pi onto the 3D printer and of course, I then started running OctoPrint and the OctoPi image on the Raspberry Pi to control the 3D printer.

15:29 - Jeff Massie (Host)
So I mean, mean it's all, it's all connected, it's all in there together. So can you get a raspberry pi 5 right now? I don't know why they were totally sold out everywhere and they they didn't they didn't have any supply.

15:35 - Jonathan Bennett (Host)
Yeah, so the raspberry pi foundation worked extremely hard on that and they are pretty much all in stock everywhere now. There's some some there. For a while the Pi 5 was difficult to get a hold of, but I think even those it's pretty much in stock most everywhere. Our Pi Locator is the place to check and I've not checked it recently, but I'm pretty sure they've got those issues pretty much worked out.

16:02 - Jeff Massie (Host)
You can just go and get one now sure they've got those, those issues pretty much worked out. You can just go and get one now. Maybe I'll have to invest and just see.

16:08 - Jonathan Bennett (Host)
See what all the uh, all the hubbub is about I mean, it's pretty cool, a little little tiny computer, little tiny computer, you could do what you needed to it's. And then you, and then you even get into like the gpio, which that's the thing about desktops you do not have. That's got some gpio in, but you don't have access to GPIO. You don't have access to SPI or I2C or any of that stuff, and it's just, it's right there on the 40 pin header on the Pi.

16:35 - Jeff Massie (Host)
And do they have a PCIe lane?

16:38 - Jonathan Bennett (Host)
That is yes, that is the. The big new thing with the Pi 5 is it's got an exposed PCIe header. Now it's not a standard PCIe, but it's a ribbon. It's a ribbon cable, but then you can buy, for example, they've got the hat. I don't think there's an. Is there an? No, they just released the official one.

17:10 - Jeff Massie (Host)
They just released an official hat to then connect an NVMe drive that hangs off of that PCI Express lane. Well, I was even thinking of a monster like U.2 drive or something that I could plug in.

17:17 - Ken McDonald (Host)
My question is how long before we'll see a kernel 6.10 running on the Raspberry Pi?

17:24 - Jonathan Bennett (Host)
Well, you could probably already boot it. I've not tried it yet, but I suspect that that would already work, because there's been a lot of work done to make the kernel happen on Raspberry Pi and lots of things work well there already. But it is time the kernel 6.10 merge window did just close and there's some fun stuff to talk about. In fact we've got 6.10. Rc1 just came out on the 26th, so a few days ago this past week. Torvalds has to say that it seems to be a regular-sized release, maybe even slightly on the smaller side, and all the stats look fairly normal. But normal obviously means much too big to post the short log, so below is, as always, just my merge log that gives an overview of who he's merged from with the barest of descriptions. There's a lot of file system work, the XFS online repair work and Bcache FS fixes sort of lead the charge there, and then of course lots and lots of driver updates. There are a couple of notable things in there. The NFS v2 client is off by default now. That's the network file system. Version 2 is old, old, old stuff. We now have version 3 and even version 4. In fact it's been a while now that you really shouldn't be running v2. You should be running at least v3. So we have NFS. V2 is off. You really shouldn't be running V2. You should be running at least V3. So we have NFS. V2 is off.

Some things for the Steam Deck has landed upstream. Amd is working on compute driver support for what they call small Ryzen APUs, that's those chips where you have both the CPU and the GPU together in an accelerated processor unit. I think is what they call that and let's see work for the Qualcomm Snapdragon X Elite, the newest Snapdragon chip. We mentioned XFS Online. Of course. Performance and performance and performance for everybody that wants that. What else do we have? The IOU ring, zero copy, has got again more performance optimization, which that's interesting. The IOU ring, that's a way to do well, so it's sort of right there in the name. It's zero copy.

That's the idea that the kernel will write out data to memory space that is shared with user space, and so then, rather than having to make a copy of it from one place to another because when you get into really high bandwidth, high throughput, cutting edge computer processing, copying from one place in RAM to another is slow it's one of the things that really gets you if you have to do that several times, and so IOU ring is one of the solutions in the kernel to avoid that.

So the kernel will just so, like things come in on the network card and the kernel just kind of assigns the network card RAM space. The network card will write directly to RAM. The kernel allows user space to access that, and so then you can run whatever you're going to run in user space. It can directly access it in RAM. It may have to make one copy as a part of processing, but then it turns right back around and again no copies have to be made after that, and then the network card can just pick it up right off of RAM and it's a ring buffer. Yes, yeah, that's where the ring part comes from.

20:53 - Jeff Massie (Host)
It's a ring buffer.

20:55 - Jonathan Bennett (Host)
Yes, yeah, that's where the ring part comes from. It's a ring buffer, so the whole thing just kind of spins on itself. It's like a snake trying to eat its own tail, but so long as you stay ahead of the tail then you can do things extremely fast. Like that's how a server can keep up with things like a 100 gigabyte network connection.

21:20 - Jeff Massie (Host)
That was Metab. Is it did that? What's that? That was meta that originally did that because they were trying to get faster performance in their large data centers.

21:30 - Jonathan Bennett (Host)
Yeah, I can, I can believe that. I know there's also been some security programs that like try to audit um, try to audit network flows real time, that take advantage of that as well. And I can't I can't remember the name of it off the top of my head. I've used it before Um. And then there's some things that got dropped, like old deck, alpha deck, DEC, the digital electronics corporation, their alpha hardware, um, we're talking old systems got dropped out, Uh.

And then let's see there was a couple of other things that I noticed that I thought were noteworthy um, work on touchscreens, and so this kind of ties back to the, the new, what Ken was talking about with. Can you, can you run it on the Raspberry Pi? If you've got like an SPI touchscreen, then there is an option here to be able to program in quirks on the kernel command line rather than having to recompile each time, and that could be extremely useful. There was some input stuff, like the Mashinike who knows how that's pronounced. Their G5 Pro controller is now going to be officially supported. And then let's see there was also some work done on the Intel P-State driver and how the Intel chip can turbo up and down and the power limiting and that's kind of been a thing that's been a thorn in Intel's side here recently.

I think last week we talked about how the Intel was having problems with some of their chips on particular motherboards that sort of tried to cut the margins a little too tight there. So hopefully that'll work a little bit better now running Linux, so all kinds of fun stuff in there. So hopefully that'll work a little bit better now running Linux, so all kinds of fun stuff in there. And I'm sure I know there are great things in 6.10 that I didn't even touch on just now. That was a very brief and very quick overview Anything in there that you guys are excited about.

23:34 - Ken McDonald (Host)
Being able to see how many populated memory slots you have at boot time.

23:39 - Jonathan Bennett (Host)
That's right. That's another one that's in there and that's pretty interesting. Yeah, that's something that's been sort of difficult. You have to dig around with some tools to figure that out If you don't want to open the side of your case and look into it.

23:50 - Jeff Massie (Host)
Of course, Well, and we had a hardware patch a while back too that also would help you assign, figure out which memory stick was in which slot, so that if there was any memory errors you knew. Oh, I got to take the memory stick out of slot x and it helped the troubleshooting in the, because when you have big servers that say have 32 or 64 modules in them, that that's a little harder.

24:18 - Jonathan Bennett (Host)
Yeah. So it's like okay, run the tool, I'll error it out. Okay, pull one of the sticks, run the tool again. Oh, it errored, I'll put that stick in. Pull the next stick. Yeah, rinse and repeat 64 times and then you know, have some better tooling.

24:40 - Jeff Massie (Host)
And one thing I thought I'd throw in there when you talked about the DEC alpha code going away. In case anybody wonders why x86 code seems to go away so much quicker, it's because there's so much development going on with x86 where things like DEC alpha or old Spark or 68,000 code, it just just sits there, nobody has to mess with it, nobody's going in and playing. So some of that code can sit in there in the kernel a lot longer just because nobody has to actively you know, patch and port and mess with it. It can just stay the way it is.

25:15 - Ken McDonald (Host)
It's not.

25:16 - Jonathan Bennett (Host)
It's not causing. What's the way to say this? You don't have to pay the tech debt on something like the Deck Alpha, whereas with the x86 support, every time somebody goes in there to change something, they've got to deal with it, and that happens all the time, and so the technical debt for Deck Alpha doesn't come due nearly as often. So yeah, makes sense. What was that, Ken?

25:43 - Ken McDonald (Host)
You don't create as many security issues as often either.

25:47 - Jonathan Bennett (Host)
Well, you don't see the security issues. I mean, is anybody out there running a Deck Alpha that really cares about how secure is this machine? I don't think they have it connected to the internet. I hope they don't have it connected to the internet.

26:00 - Jeff Massie (Host)
Deck Alpha was from 92.

26:01 - Jonathan Bennett (Host)

26:03 - Jeff Massie (Host)
I was going to say I remember when the search engine AltaVista was always bragging, they were running on Deck Alphas because that was the hot chip at the time. But I mean, you know, yeah, a lot of times too, that old hardware. If you have a DEC alpha, you're probably not trying to run the 6.9 kernel or 6.8. You're probably back on 2. Something with an old operating system that nobody's updated in 20 years and you don't care because it's still plugging away in the corner doing its one task.

26:35 - Jonathan Bennett (Host)
Yes, yes, making the machine work that's too expensive to replace.

26:41 - Ken McDonald (Host)
yep, yep, yep that is how that goes.

26:46 - Jonathan Bennett (Host)
So, speaking of making the machines work, there is, um, there, there's a thing, there's a new thing, and it's an alternative to nv link. Which nv link is that? The? Is that the technology to be able to put two graphics cards on one board? Or am I thinking of the wrong thing?

27:04 - Jeff Massie (Host)
It is, but also in servers. I mean server clusters, servers. So it's more than just a single server. It's connecting a lot of GPUs or accelerators, and I'll kind of probably switch back and forth nomenclature, but in this discussion pretty much they're pretty much the same thing. They're being used for AI type computations. So, and you know, nvidia right now is printing money with their hardware for the AI market and no one's dethroning them right now but that could change. So NVLink is getting a competitor.

Well, you know several Intel, amd, cisco, google, hpe, meta Broadcom and others have formed an alliance and come out with the UA link as a new open standard that they want to use to dethrone NV link. So you know, as some of you might think, it must be bad for, like AMD and Intel to work together, but you know this isn't the first time they've teamed up. Just last year they came up with the ultra Ethernet for high speed networking and you know then they also had help from Meta and HPE and others. But they're not strangers to joining forces. So UA-Link is kind of like the ultra Ethernet. That's standard, but it's specialized for linking together GPUs or accelerators. So there's some overhead and assumptions they can make because of the specialized arena that it's working in, of the specialized arena that it's working in, and now these can all be on the same system or in a group of systems allowing them all your GPUs and accelerators to communicate and share data at a much higher speed than current networking standards. Ultra Ethernet will still have a place, as UA-Link is specialized. So while Ultra Ethernet and UA-Link might overlap some, they're not totally interchangeable, so one is not going to dethrone the other. These aren't competing standards, they're symbiotic standards.

With large AI loads, there's a need for more processing power that can be shoved even into a single rack of servers. So this will allow up to 1,024 GPUs to be networked at an extremely high rate, and there's probably going to be more. That can even be added in later revisions of the standard, and they're hoping to have the first revision or the first specification 1.0 out in the third quarter of 2024. And in the fourth quarter they want to have the first update, which will provide even more bandwidth. The rollout and the roadmap for this specification is very aggressive. To really hit, you know they want to give NVIDIA a roundhouse. You know they want to nail it, clock them right on the chin, and UALink will even be leveraging the Infinity Fabric protocol, which, if you're an AMD fan, that's the high-speed CPU connection that AMD has.

The downside is, the estimated implementation in hardware will be around 2026, as the standard's going to be out soon, but the silicon pipeline is going to take a while to have the designs become reality. To have the designs become reality, because until the standard is pretty solidified, you can't be making all your traces in the silicon or else you might have some issues. And there's actually a real world example, that with HBM3, where Samsung designed their part before the standard solidified. Well, now their current part doesn't meet the standard and they've had to go back. Their current part doesn't meet the standard and they've had to go back and make some tweaks to meet the standard because they thought the standard was going to go one way and it didn't. So this stuff can bite you. So you have to be pretty confident before you start laying out the silicon design.

We are going to be seeing some new AI hardware. We are going to be seeing some new AI hardware from both Intel and AMD, which they're feverishly working on, if the rumors can be believed, which I'm sure they are, because they're looking and saying I want a piece of that AI market, so they are throwing a lot of resources into getting into this computation, the GPU accelerator for AI market. Now I do want to point out this isn't specific to Linux, but most places this will be used will be Linux and it's an open standard. So you know we might be bending the Linux part of the show title a little, but you know who doesn't love a nice open standard. You know, especially when companies are really starting to fight, that just means we win.

Now take a look at the article in show notes for more details and keep an eye out for more hardware coming out which, while aimed at AI and other computational tasks sorry, there it's very close to GPUs for gaming. So expect some higher power GPUs in the next couple of years because if they can make quite a bit of money making a few small tweaks to their AI hardware, the companies will you know if we can say, if they can say, oh, we'll tweak this a little bit and we serve a different market, and that there is money to be had there. You know the return on investment is good. They want the money, so we should benefit, even as standard users and game players and people using Blender and things like that, we should have an advantage coming.

If you're wondering why an open standard, it's because right now, like I said, nvidia is a juggernaut of the AI space and they know if they try to come up with their own proprietary standard and you know Intel has a standard and AMD has a standard and you know Broadcom has a standard it's just going to be that much harder to get adoption.

But if all the hardware will work with the standard and depending on how successful their hardware is, nvidia might need to support it as well, which now, ok, I'm going to kind of go off the article a little bit. So I'm saying, when the door which opens the door for mixed vendor accelerator pools now Michael Larable says he doesn't think it's going to happen, but I think it can Now it's going to take more than UA link to make them work together. That would just let them talk, but they would need to all speak the same language. But then if you look at what's happening with like RockM and other open standards like that, I think that in the future it could happen that we could have mixed vendor accelerator pools and I think the industry would really want to go to that because they you know they one vendor buy-in is not usually in your best interest. You want, you want to have options and it, I think, I think it's going to happen eventually. Ua Link won't do it alone, but it kind of gets us one step closer.

34:02 - Jonathan Bennett (Host)
Yeah. So just to make sure we understand what we're talking about, does this replace PCI Express? Are we talking about a new connector on the motherboard for plugging video cards into it? Doesn't really.

34:17 - Jeff Massie (Host)
Or is this going't really an?

34:18 - Ken McDonald (Host)
additional connector it.

34:21 - Jeff Massie (Host)
It's kind of like an additional connector, the way I understand it, because it's it's super high speed and it's only for accelerator pooling. So this is not a uh. You know, like I said, it's not like ethernet, it's not like pcie, where it's kind of a universal standard. It's very specific.

34:38 - Ken McDonald (Host)
So GPU to GPU, yeah.

34:42 - Jonathan Bennett (Host)
So I did a little, looking the way I understand it. So I've got a card here. They ignore the weird stuff on the bottom. The way it would work is you would still you'd plug your card into PCI express and then and sometimes it's not how it physically looks, but there's then basically on the top there's another connector and so you connect onto the top of the card and then so in the NVIDIA case, with NVConnect, they make these little bridges. You can put two cards right beside each other and you can bridge the two of them to take you together. It's a lot like the old SLI technology. It's essentially the same idea.

35:15 - Jeff Massie (Host)
It's essentially the same idea, it's just in this case it's to allow your models, your kernels that you're running on your cards to be able to talk to each other without having to go back through the CPU to do it and they talk about. You would have like switches and things like that that would also interface, so you can get these monster numbers of accelerators all communicating, yeah.

35:40 - Jonathan Bennett (Host)
And so I imagine on a lot of installs instead of it actually so some of NVIDIA's. It is literally a connector on the top, but from what I was reading, you also have some of these with NVLink, where it's like an extended PCI Express slot, where it's just extra pins on the end. That is an NVLink connector, and so I imagine that that is what it's going to eventually be, which we've seen this before in the computer industry, right Back in the I don't remember what the name of it was, but you had your graphics like your card, and then you're extended and some of those got really long in the old days, inside of computer chassis, the old AGP slot, yes, yes, yes, the old AGPs. So I think that is probably where this is going to go, where it's going to be essentially like an extension on the PCI Express. And will we see these? Will we see some of these make it down to consumer computers? Maybe, I don't know.

36:43 - Ken McDonald (Host)
Will we see support in the Linux kernel?

36:46 - Jonathan Bennett (Host)
Yes, absolutely yeah, yep.

36:49 - Ken McDonald (Host)
Definitely by 7.

36:53 - Jeff Massie (Host)
Yeah, because basically most of the data centers running AI, they're running Linux. Oh yeah, I mean it almost all of the data centers running AI they're running Linux. Oh yeah, almost all of them. Yeah, because really Linux won. Except for the desktop, pretty much everywhere else Linux is the dominant operating system. Yep.

37:12 - Ken McDonald (Host)
Phones servers.

37:15 - Jonathan Bennett (Host)

37:16 - Ken McDonald (Host)
AI systems.

37:17 - Jeff Massie (Host)
Mm-hmm Industrial equipment, yep, even your cars.

37:23 - Jonathan Bennett (Host)

37:25 - Ken McDonald (Host)
All the embedded hardware you've got in your home.

37:27 - Jonathan Bennett (Host)
Yep, probably your toaster.

37:30 - Jeff Massie (Host)
And I could see that they maybe have it for consumers. It would kind of be more like the NV link, where you just have the little connector between the two. You know, but I could see it, you know, someday, because if it's the standard hardware, okay, there's just that much less we have to do to change. You know, when we turn this into a gaming GPU.

37:55 - Jonathan Bennett (Host)
Yeah, I was thinking about that. So, like what do we think the scenario is going to be where you need to have two of these connected on a consumer's motherboard, like is there at some point? Are we going to have a GPU in one slot and an AI accelerator in the slot beside it and then they talk to each other over some kind of link like that? Or are we eventually going to be on the point where you have again a dedicated network card and you want them to be able to talk to each other? I don't know. There's possibilities there. I don't know that any of those are really going to catch on, except maybe the AI accelerator. I could foresee. I could understand that being a thing eventually.

38:37 - Jeff Massie (Host)
Well, I could see it like render farms or things like that.

38:40 - Jonathan Bennett (Host)
Well, true, but then we're kind of moving away from the idea of consumer hardware. Obviously, it makes sense for a render farm. I'm trying to think of why a gamer is going to need NVLink or UALink, and the only thing that comes to mind is oh, if you want really, really realistic AI behavior, then buy our AI accelerator card, and here's the clip to be able to clip it onto your GPU so that your GPU and your AI card could talk directly.

39:12 - Jeff Massie (Host)
Well, and you could have that. You got that 8K monitor at you know 120 Hertz or whatever, so you got to have your GPU. You got that 8K monitor at you know 120 hertz or whatever, so you got to have your GPU. And then, to make your opponents really really hard or lifelike, you have AI that you're playing against, so it's very Well, it's not just.

39:31 - Jonathan Bennett (Host)
it's not just the AI for like. So you're thinking kind of FPS, make your opponents smarter? I don't think that's the really interesting thing to do with AI and gaming. I think probably the most interesting thing to do with AI in gaming is, in more story-driven games, being able to do smarter dialogue that actually responds to what the player has to say.

Somebody uh smarter dialogue that actually responds to what the player has to say. Uh, somebody, somebody's already done this in a mod for you know, skyrim mod or whatever. They connected the npcs up to an llm um and of course the skyrim hated it, like the, the, the, the career, bethesda, it's like oh no, we can't believe it. But it was really interesting because you could then actually type in things to say and the NPCs would give you reasonable responses based on what you said. And it's like that is one of the coolest ideas I've ever seen in gaming to get actual, realistic responses to people's actions and words, and I I think that is an idea that could have legs.

40:43 - Jeff Massie (Host)
And I would take it one step further. It would. It would change the world you're in. Yes, so a lot of the open world, like if you're playing Witcher three or something, there's oh, I make a decision here and it changes, but you're still kind of on certain tracks. Just imagine if you had the AI that would. You could kind of have either thousands of tracks or you have almost unlimited, because everything I do my game is unique, based on everything I've done in the game.

41:12 - Jonathan Bennett (Host)
Yeah, yeah, it'd be really interesting. All right, we're not talking about Linux anymore, though, and we really should get back to it, since we are a Linux show, I do love Linux. Well, that's true, I do too, but anyway, ken is going to rescue us from the AI talk and we're going to talk about Nitrix, or is it Nitrix? I can't ever remember.

41:31 - Ken McDonald (Host)
It's one of the two, all I know is we're going to be talking about something that has licorice kernel option, which is great, because last week we covered MX Linux 23.3 offering a licorice kernel option. This week we can thank Bobby Barasov for writing about another distro with licorice kernel 6.8. As you said, Jonathan, Nitrix, kernel 6.8. As you said, Jonathan, Nitrix, this time it's version 3.5, codenamed CX, and it packs a range of updates and new features that aim to boost performance, enhance security and improve the overall user experience. Nitrix 3.5's major components have seen significant upgrades, with the main ones being guess what? The first one is, Jeff, Firefox Updated to version 126.0.1. Then we have the KDE. Plasma desktop environment is now at version 5.27.11 and KDE Frameworks to version 5.115.0. Jeff, you'll be happy to hear that the NVIDIA Linux x64 display driver is going up to version 555.42.02. Jonathan, the AMD open source driver for Vulkan is up to version 2024.q2.1.

43:11 - Jonathan Bennett (Host)
Guess when. That came out Probably pretty recently here in 2024.

43:13 - Ken McDonald (Host)
Now the Nitrous team has introduced several modifications aimed at tightening security and boosting system efficiency. They include optimized memory handling to reduce overhead and improve responsiveness. Implemented security measures such as filling freed memory pages with zeros. And enhanced randomization of kernel stack offsets to thwart potential data leaks and attacks. We know a lot about those, don't we, jonathan Mm-hmm. And last is our. The last one I'm going to mention is harding the Berkeley packet filter just in time compiler against attacks by fine-tuning the sysctl settings. Nitrix 3.5 now includes several MAUI apps as app images, such as Bonsai, buho and Pix. The NX Software Center has also received updates fixing previous app image handling and installation issues. I recommend reading Bobby's article for more details, especially about why Nitrix 3.5 is using Plasma 5.27 still.

44:34 - Jonathan Bennett (Host)
Yeah. So I'm going to spoil it, because I think it's really fascinating. Two things about Nitrix that really fascinates me. One it is a system D free distro. This is one of those distros that said we do not want system D, we are going to stick with something else. I'm not sure exactly what they're using under the hood, but it's not system D. And then they are currently shipping, as you just said, kde, kde 527. Their plan is to go to the Maui Shell, which I did a little reading about, and that is apparently a brand new solution that does indeed run over X and Wayland. That is intended to be the same experience for a phone, a desktop or a tablet. It's one of these solutions where it takes sort of the same idea and puts them on all of the different places. An OS that we all know and some of us love tried that with one of their previous versions and everybody hated it. We will see if MAUI Shell pulls it off and we don't hate it the way we did with a certain Windows release of years gone by.

45:46 - Jeff Massie (Host)
We'll see. Well, even some Linux distros have kind of went down that path, and it just now. I'll keep an open mind. I'd give it a try. But I find that a lot of times between desktop and tablet and phone they're too different of experience and how you use them is so different that it's tough to make one that fits. One size fits all. Yes, because that elastic waistband might say one size fits all but it never fits anybody really well.

46:17 - Jonathan Bennett (Host)
Yeah, yeah.

46:20 - Ken McDonald (Host)
Jonathan, have you heard of Core Control?

46:24 - Jonathan Bennett (Host)
Core Control. That does sound familiar. What is that?

46:28 - Ken McDonald (Host)
It's a free and open source GNU Linux application that allows you to control your computer hardware easily using application profiles. It aims to be flexible, comfortable and accessible to regular users, and that's one of the other things that's going to be appearing in this release.

46:48 - Jonathan Bennett (Host)
Yeah, I think we've talked about this before. Actually, that's the one where you can go in and tweak the speed that your different cores run at. You can turn cores off. Yeah, it seems like we have talked about this before. It's C-O-R-E-C-T-R-L. In fact, that may have been a command line tip from Rob, maybe a few weeks ago.

47:11 - Jeff Massie (Host)
Yeah, because we talked about he couldn't turn off core zero. Yeah, that's right, core zero is like a special case. You have to have Core 0. It's important. Yeah, it's kind of hardwired in. That's the one that's going to boot and load the initial BIOS or whatever. Yep, yep.

47:33 - Jonathan Bennett (Host)
All right, yeah, nitrix sounds like fun. So the Elastic W waistband of Linux releases Bravo, Bravo, mashed potato, all right. So I want to chat about the sort of the end of a story. Maybe, hopefully, hopefully, it's the end of this particular story. There is a new XZ release and, for those that are not keeping track, xz is well, it's a compression program and it's programmed and made by Lassie Cullen. Well, the XZ tools, that's the package that had a backdoor in it that unfortunately did get shipped in.

Just a couple of distros Not very many, thankfully, none of the really big, important ones, but a couple of distros did push it out and it backdoored SSH. And the way this happened is a volunteer, so it's an open source project. Someone started writing code for the project, volunteered and then eventually volunteered to be a co-maintainer of the project, which, for anyone that runs open source, you know, having somebody else along to help is just it's awesome, it's great when someone knows what they're doing and they offer to help, because there's a lot of work to running an open source project. And so last they agreed and said sure, here are the things you can do. Well, a programmer going by the name of Gia Tan started doing work, and some of the work that he did was adding in a backdoor, and we've talked about this sort of in detail in the past. There's been lots written about it, so like, if you want the, the technical blow by blow of how the the backdoor works. You know there are. There are other and better places than me trying to remember it off the top of my head the.

The interesting thing is that Lasse was kind of so he's the head guy of RecC. He had kind of disappeared from the internet for a week. He had just taken some time off for himself to not lose his mind Again. People that run open source projects understand the need to do that from time to time, and this got released while he was sort of out of the saddle. Now he is now back and he's been back for a while, but he is now back and there is a new release of XZ out and with this there is also some pretty interesting analysis of what went wrong.

Analysis of what went wrong. So I'm looking for the news. I had this earlier and it was pretty funny. So it's like the release notes of what's new in the latest version and the top thing was removes the back door. Ah, there it is, the news file Got it. So this is 5.6.2 and can remove the back door and then the CVE number and then a few things that were not changed and some things that did get changed, because there are legitimate changes that kind of got wrapped up in all of this too. So obviously, when, when this was all discovered, the first thing that everybody did is the five, six zero five six one. They just went and they yanked it out and they went back to the previous version. So there's actually quite a bit of work to to make this happen, this happen.

But if you're interested in the rest of the details, I have a couple of pages linked where Lasse has gone in and given sort of details on what he has found as he has gone through and looked more deeply into what all happened. And so there's the short one. But the really interesting one is it's tanakiorg xzbestore slash reviewhtml again link in the show notes. And he actually goes through and goes change by change with like every commit since Giatan joined the project and kind of gives you the background like here's why this one's okay, I understand this. So some of them he goes.

I know this looks scary, but this is actually okay, and here's why he talks a little bit about the time zones, because this is something people were looking at, sort of trying to figure out where was Gia Tan actually from? Because obviously it's a name that sounds Asian, it sounds like it could be someone from China, but when you look at the time zones it doesn't necessarily match that and so obviously the name is a pseudonym, probably for a three-letter agency from some country, and there's really no way to know at this point what country. People were looking at time zones and trying to figure it out. And he even has a note here that the mundane explanation for any weirdness in the time zone is you can put a commit in somebody else's name with permission. You can use get rebase with the dash dash reset, author date. You could possibly use committer date is author date as another get flag.

So there's a lot of questions that are still unanswered as far as that stuff goes. But if you want to know the nitty gritty details about all the changes and everything, there are links in the show notes. And, as I said, hopefully this is the end of the story. Lessa Khan is back in the saddle again and we are all going to be a little bit more cautious about who we give the keys to the kingdom to. Not just everyone should be able to make releases and add code, so hopefully that's the end of that story so you don't trust me I have met you in person, ken.

That is more than I can say about quite a few people I trust on the internet. I think ken is. Ken is on the trusted list. I've never met Ken. I don't know. I've never met Jeff in person. Jeff may not be real. I'm AI IA.

53:47 - Jeff Massie (Host)
Which would explain the hallucinations and sometimes wrong information. I'm an early version.

53:54 - Jonathan Bennett (Host)
He's the alpha of the AI. Oh goodness.

53:59 - Ken McDonald (Host)
Speaking of what version you are what version is Firefox now.

54:05 - Jeff Massie (Host)
Oh, there it is. I don't know what. We have a roadmap. So in a blog post at Mozilla, they released a roadmap of the future things they want to see in Firefox. So a couple of things that a lot of people are excited about are the grouping of tabs and vertical tab organization. I think this will be great, as I know personally I have a ton of tabs open at once and there are several of them that could be grouped together. I know personally I have a ton of tabs open at once and there are several of them that could be grouped together, and I also use an ultra-wide monitor and I'd like to put my taskbars to the side so I can use all the vertical space I can. Well, I'm not so concerned with the horizontal.

So to go along with these features is a more modernized sidebar, and they say this. More modernized sidebar will allow for a streamlined experience. There are extensions currently that a person can load to make this happen today, but the thoughts are that being native will be an improvement for security and privacy. You know there's a balance between how many extensions you want to load, how much do you trust them, versus okay, it's built into the browser, we don't have to worry about it. Another feature is a new profile management system. So basically it would allow you to better separate things you do, such as work, study and personal interests. So when you switch what you're doing, you can switch profiles. So things are much more seamless with what you're doing. So maybe you've got a lot of things set up for work and then for your private life you've got things set up differently and it'll just help, you know, make things a lot smoother.

Privacy is getting looked at and they're looking to improve. They want to make managing the settings for prevention of tracking easier and to make sure you're getting the level of privacy you want when you're surfing the web and, as we know, there's kind of a trade-off between how much privacy and convenience you have. So they're trying to better let you turn the dial to how secure you want it and not have to root through quite so many deep, dark menus. Speaking of menus, menus are getting redesigned and they're going to focus more on the most important things and make what you see cleaner and simpler. The new menu will highlight priority actions, making it easier to choose what you want. Performance is getting some force behind it as well. They want to reduce page loading time, speed up the startup and help with battery life for those on mobile devices.

They do want to make the overall experience better, and one of the ways to make it better is interoperability between different browsers. They want to make surfing in Firefox like what you would see if you're surfing in Chrome or some other browser. It'll also allow easier creation of websites which look at, which will look the same in different browsers, so there's less of the hey, this looks different in this browser versus that browser versus Firefox. Finally, it wouldn't be an update if someone didn't say AI. I think we can all. I think we all saw that one coming. I won't go into it, but it'll do AI things that you hear about and other AI things and other AI programs do AI things that you hear about and other AI things and other AI programs and AI, AI.

57:19 - Jonathan Bennett (Host)
Well, there is actually. There is something really interesting. I'm going to I'm going to interrupt you real quick, I'm going to let you finish, but oh, that meme makes me old, doesn't it? Anyway, one, one thing that they, one thing that they noted is Firefox is looking to execute all of their models locally, without relying on external services. I think that's really one fascinating. And two, also pretty, pretty exciting. Like I do not like the idea of my data going up to the cloud to be able to be, you know, AI-itized, Running it locally. I think it's a great idea, and that's one of the things that they kind of harp on here. So, yes, AI, AI, AI. We're all sick of AI, but like they're at least trying to do it right.

57:59 - Ken McDonald (Host)
And it looks like they're putting it in for the as part of the assistive technology accessibility.

58:08 - Jeff Massie (Host)
True, yeah, they're looking at having it read, pdfs and things like that for people that need accessibility. Now, I should point out, while these are getting worked on right now, there isn't a timeline when we can expect to see them, so I don't have any dates that I can give you or specific releases when these will come out. We're just going to have to keep our eyes open on the Mozilla announcements page for when the next version of Firefox has some or all of these. Personally, you know I'm excited, as I'm still a Firefox user. I don't I don't use any of the derivatives.

You know there's other ones out there based on Firefox, but you know I have in the past, but I've tried various. You know I've used Chrome at work. You know I've used Edge. I've I used them all. But you know I just kind of keep coming back to Firefox. It just seems to, I don't know, scratch an itch right or it's those old comfortable shoes, I don't know. I just keep coming back. So, you know, and I like the fact they're focusing on the basics and maybe they'll get their 3% usage rate to start climbing again. But yeah, I just I like the focus on the basics and not whiz-bang.

59:18 - Jonathan Bennett (Host)
Yeah, so there is a change coming that may get me back on Firefox, and that is Google Chrome and the change coming there to what plugins are allowed to do, and it's Google. Really, what this is is it's Google taking aim at ad blocking, and that does not sit right with me, and so if the changes that they make actually end up being as intrusive as people claiming that they will be, I may be looking for another browser. I've been on Chrome for a long time. I'm on Chrome on Linux. I may be moving back to Firefox. We'll see.

01:00:07 - Ken McDonald (Host)
I've got Chrome on my Chromebook.

01:00:10 - Jonathan Bennett (Host)
Yeah, I've used Chrome a lot, but I'm not okay with I can't help but think that there may be eventually be some legal ramifications for this but I'm not okay with google um taking aim at ad blocking in their browser because they feel like it's hurting their search business. I don't think that's right.

01:00:34 - Jeff Massie (Host)
And I'm even one of those really weird people. I have an Android phone, my browser is Firefox.

01:00:40 - Jonathan Bennett (Host)
Yeah, I've heard Firefox on Android before.

01:00:42 - Ken McDonald (Host)
Your default browser is Firefox.

01:00:44 - Jonathan Bennett (Host)

01:00:45 - Jeff Massie (Host)
Yep, and I hear people say oh, the experience is terrible, I don't really notice a difference, it works as good as chrome. But I am not a power user of, uh, mobile devices. I call, I text some basic stuff, I you know.

01:01:02 - Ken McDonald (Host)
so maybe if I was really a heavy duty user I might notice a difference, but for me it works just as good as chrome the only thing I want to add is keep an eye out for their Fire Mozilla to announce when they do their upcoming AMA on Reddit.

01:01:20 - Jeff Massie (Host)
Oh, I didn't know they had one coming. Oh yeah, that'd be interesting.

01:01:24 - Jonathan Bennett (Host)
Yeah, share a link with that. If we can, let's share that link in the Discord so that everybody can know about it, because that'd be really interesting to catch.

01:01:32 - Jeff Massie (Host)
And Keith512 says Firefox 127 is out Tuesday 11th of June.

01:01:38 - Ken McDonald (Host)
Yeah, it's at the bottom of the link I just shared.

01:01:44 - Jonathan Bennett (Host)
Oh cool, they don't give a date and time.

01:01:48 - Ken McDonald (Host)
They just say they'll be announcing it.

01:01:52 - Jonathan Bennett (Host)
Yeah, so it's not solid yet. When that solidifies, we should make sure and let folks know about it, because that'll be announcing it. Yeah, so it's not solid yet. When that solidifies, we should make sure and let folks know about it, because that'll be really interesting to catch. All right, ken, what is Rhino Linux?

01:02:07 - Ken McDonald (Host)
What is Rhino Linux?

01:02:08 - Jonathan Bennett (Host)

01:02:09 - Ken McDonald (Host)
You don't remember us talking about it. I don't Say six months ago.

01:02:15 - Jonathan Bennett (Host)
Man, I've slept since then.

01:02:17 - Ken McDonald (Host)
At least once.

01:02:19 - Jonathan Bennett (Host)
At least Thankfully, more than once.

01:02:21 - Ken McDonald (Host)
Well, I want to start off by asking you a question Do you want a distro that continuously receives system updates?

01:02:34 - Jonathan Bennett (Host)
If I really, really wanted that, I'd go run Rawhide.

01:02:39 - Ken McDonald (Host)
Well, you also might want to run Rhino Linux. Bobby Borisov wrote about the new Rhino Linux 2024.1, an Ubuntu-based rolling release providing an in-house modified XFCE desktop environment called Unicorn Desktop. It introduces several key enhancements, notably the latest release of Packstyle, version 5.0. It has seen massive improvements, including updates to the pack script format for better security and testing capabilities, and enhancements that allow for a more user-friendly experience for scripting and package management. Rhino Linux 2024.1 uses Linux kernel 6.9.1-generic for desktops, 6.9.0-ok Pine on the Pine64 devices and 6.8.0-Raspi for the Raspberry Pi. Unfortunately, Bobby reports, the release is not without its minor glitches. Users may find desktop icons not appearing on live boot ISOs, though this issue resolves post-installation. For those encountering display issues with the global menu plugin in Unicorn, a temporary fix involves adjusting the CSS settings in the system's configuration files. I recommend reading Bobby's article if you want to hear the hurdles faced by the RIDO team during this latest development cycle.

01:04:18 - Jonathan Bennett (Host)
Yeah, interesting stuff.

01:04:23 - Ken McDonald (Host)
I'm wondering when Rob's going to be distro-hopping to it for a little bit up into it for a little bit.

01:04:34 - Jeff Massie (Host)
You know a rolling release kind of based on Ubuntu. I wonder how close it tracks versus if you're running Ubuntu with the backport servers on, Because that's where you a lot of times they'll pull. For example, if you're running on 2404 and you turn backports on, you'll start pulling in things that they're going to put into 2410. So you get a lot more of that cutting edge software and and possible stability issues, because it's early, you know, beta yeah, this is interesting.

01:05:12 - Jonathan Bennett (Host)
Uh, I'm I'm kind of surprised, honestly, that uh, ubuntu doesn't have a rolling version, like that seems like an ubuntu thing to do, to to package something that's rolling, and I don't think they have one, do they?

01:05:25 - Jeff Massie (Host)
they do rhino linux I thought there was an official.

01:05:32 - Ken McDonald (Host)
No, I don't think it's an official Ubuntu flavor, but no, I thought Ubuntu had one.

01:05:40 - Jeff Massie (Host)
Maybe I'm mistaken.

01:05:43 - Jonathan Bennett (Host)
No Googling, for that is just coming up with Rhino Linux. Yeah, yeah, so that must be Okay, that's it. That's the one, unofficial. Yeah, so that's going to the official Must be Okay, that's it. That's the one, unofficial, yeah. So that's going to be.

01:05:57 - Ken McDonald (Host)
The official.

01:05:58 - Jonathan Bennett (Host)
Yeah, it's the unofficial official, it's going to be. So that idea of taking Ubuntu and making it rolling, that's going to be easier sometimes than others. Right, like the space between so right now, what 24.04 just came out, and so for right now, it's easy to have a rolling Ubuntu release. You just track 24.04. What's going to be challenging is when 24.10 comes out. How do you roll people from 24.04 over to 24.10 seamlessly, and that's so. That's going to be the thing that they're going to struggle with, because there are sometimes things that you have to do when all of those packages update and, trying to stay on top of it, like they mentioned, you run into minor issues and other problems.

01:06:45 - Ken McDonald (Host)
It's as simple as riding a wave.

01:06:50 - Jonathan Bennett (Host)
It's probably not quite that simple. It's probably not quite that simple.

01:06:54 - Jeff Massie (Host)
Yeah, I mean, and you know, as Keith512 said, you know Ubuntu packages update in ice ages. They take so long, but that's part of the stability of it too. I mean, when you're running a rolling release, you can you know it might work fine, might not.

01:07:12 - Ken McDonald (Host)
You're running a rolling release because you want to live on the edge.

01:07:16 - Jeff Massie (Host)
Yeah, and sometimes you fall off the edge or you cut yourself on the edge. So it's how stable do you want versus features or new options?

01:07:29 - Ken McDonald (Host)
How quickly do you want to get those new features?

01:07:33 - Jeff Massie (Host)

01:07:34 - Ken McDonald (Host)
Are you still driving that 1999 Ford F-150, or do you have the latest model F-150?

01:07:45 - Jeff Massie (Host)
All right 91 Sonoma.

01:07:50 - Jonathan Bennett (Host)
So there is one last story to cover and that is something interesting happening in Asahi Linux land. For those who don't know, Asahi is the Linux. It's not even a distro anymore, it's kind of a distro. It is an effort to put a good Linux desktop on the Apple Mac M1, M2, now the M3 as well hardware and for some people that is a very interesting thing to do because, as has often been said, Apple tends to make good hardware and some of us just don't really like their operating systems that much. So being able to run Linux on them is sort of interesting. And we've talked a lot about Asahi in the past.

Well, Asahi is now officially a Fedora spin. Actually, you know, they were I think it was Arch originally, and they have now kind of adopted Fedora as their base for Asahi images. And there have been some I guess you could call them growing pains along with that. Of course, the whole project has sort of been growing pains because what they're trying to do is so like it's technically challenging. So this particular story is about the process of installing Asahi. You actually start by running some binaries in mac os and now that it's fedora, there was this bit of challenge of how do we package this? Like? How do we put these two things together inside fedora when fedora has this? You know pretty hard stance that everything that's a part of fedora needs to be built on the fedora to build system, which there are reasons like security reasons, reasons. You, you, you know you're getting good software with Fedora because, like you know, the maintainers and it's all getting built there, Um, but unfortunately that doesn't work very well for trying to build these Mac OS packages. And so they uh, they added, they asked the, the FESCO, the Fedora engineering and steering committeeering Committee, for an exception to ship pre-built macOS binaries in the Asahi installer. Which the thing about this is you run it it's a Python program and you run it on macOS to do your initial install, but I believe you also run it out of Fedora to be able to do certain types of firmware updates. So it's not just something that runs for the first install and then you're done with it.

And there was a bit of consternation and discussion about this, Some people not for it or maybe not understanding exactly what was happening here. I believe the parts that are pre-built to build them you actually have to build them on Mac OS and I think you build them with the Brew build system, which, if you're running Mac and you haven't messed with it, Brew is great. By the way, Fesco talked about it in one of their virtual meetings and agreed to it. Um, and now, if you go and you look at the link, there's actually some discussion after the fact of people going.

Wait a second, what did we agree to? It's like you weren't listening very well during the meeting, Were you? Um, anyway, so there's a, there's fun stuff coming there, but, Asahi, it's Fedora now and, uh, it almost makes me want to go find a cheap Mac just to put it on to play with, because it does look pretty cool and it would be fun. It would be fun to run a Mac and be running Linux on it. Just make all of the Mac fanboys' heads explode and all the Linux fanboys happy.

Yeah, yeah, well, I mean, I mean, that's me, I'd be one. So, alright, that is our news for the week. Some fun stuff, particularly with the kernel and some distros, some hardware news. All kinds of fun stuff. Let's move into some command line tips, or they're just? Let's move into some command line tips, or are they just tips? Now, they're not necessarily command line tips.

01:11:58 - Jeff Massie (Host)
Oh, I'm back in the command line this week. Yeah, yeah, good, good.

01:12:03 - Jonathan Bennett (Host)
I think they're command line tips this week, then we're gonna let Jeff kick us off, and what are we doing on the command line? So?

01:12:10 - Jeff Massie (Host)
I have a neat little program that I think is going to help some of our professional programmers out there and I think it'd work for, you know, probably a lot of other professionals as well. Now I think you might like this one, jonathan. So BarTib. It keeps track of how much time you spend on projects or tasks. So all the data is saved in a human, in a text file in human readable format, and if you look at the link in the show notes, it'll show you different, several different ways to install the program and, if nothing else, pre-compiled binaries are available.

So the most basic commands are simply the start and stop commands. They're flags which you can set to define the project that you're adding time to. You can add a description. So while you're working on a specific project you can put in descriptions of. You know you're fixing a specific bug for this time frame. It lets you further parse what you're doing on a project. You know you can look at a list of the most recent projects and their descriptions and continue on with any one of them by choosing the number that corresponds to the item from a menu that it's generated. You know it'll.

It can keep track of your number, your time on any number of projects. It can also generate reports to find out how much time you've spent on each project and subtasks of that project. And you know I'm not going to go into everything because it's got a lot of options, but you know you can. You can customize how the time and dates are handled. You can add or change subtract from current projects. You know, maybe you you left it running and you went to lunch or whatever. You can fix that there's. You know and you can. You can make the, make it look and act like how you want. So take a look at the link in the show notes. And the article in the show notes also has a link to the GitHub page where you can get very detailed descriptions on each of the options. But for those of you that have to do project tracking, this might be a handy little thing while you're working.

01:14:06 - Jonathan Bennett (Host)
I totally agree with Harold Finch here. He says it would never work for me. I start working on something and suddenly it's 4 am I. I feel that, I feel that very much uh but it.

01:14:19 - Jeff Massie (Host)
But it would work because you just make sure you hit start and until you hit stop, it's clocking the time. So even at 4 am you, you stop it and it'll keep track of what you were doing, that's's true, I might try to use this.

01:14:36 - Jonathan Bennett (Host)
I have journaled and tracked my time before and it was kind of depressing. I just know a lot of professionals.

01:14:44 - Jeff Massie (Host)
You know a lot of the programmers I work with and even a lot of the engineers. You have to assign your time and how long did you work on specific projects? Because they you know cost accounting and you know load balancing and all sorts of stuff.

01:14:58 - Jonathan Bennett (Host)
Yeah, pretty much all of my billing is is fixed flat rate. I I do hardly any time-based billing anymore. Um, so I don't, I don't know that it makes a whole lot of sense for that. I think it's. Only I could see this being useful for me for doing something like kind of internal auditing of how much time I spend on stuff. But other than that, yeah, it's not the way business works for me. But you're right, for someone that's programming and charging by the hour, it could be great programming and charging by the hour it could be great.

01:15:32 - Jeff Massie (Host)
Well, even even if you're part of a larger group and you still get paid, you know salary, but a lot of times they want to track your project hours to find out Whoa, wait a minute. Project A took way more, is you know? Do we need to assign more resources? Do we need to move off of it? But you know, yep, whatever, it is True, all right.

01:15:54 - Jonathan Bennett (Host)
This really speaks to me from the corporate world Project C takes a little time to do. Yeah, yeah, all right, ken is up next. What do you have for us, sir?

01:16:02 - Ken McDonald (Host)
Well, this week I'm going to review some of the commands we've gone over in the past, but with a look at how to use them for checking or monitoring your user logins, shutdowns and reboots on your system. The first command I'm going to touch on is who, except this time I'm going to recommend that, instead of just typing who, that you add a dash B to see when your system last booted. Another one that we've covered in the past is last, but I'm going to put in the show notes all the flags I'm going to cover here in a minute. But I found that using the following flags dash X, which causes it to display system shutdown entries and run level changes. Dash capital F, to give you the full log in and log out times and date. That way you're not trying to guess which year it is if you've got logs going back several years.

Dash I displays the host IP address. Dash n, and in the example I'm going to give I have it followed by the numbers 1, 0, which would make it limit the output to the last 10 lines from your log. And another one that we've covered in the past is uptime. And another one that we've covered in the past is uptime, two flags in particular I recommend using with this. One is "-p", to show how long your system has been running since it last booted. The other one is "-s" and it will print the time and date it last booted up on. Now, I actually covered journal CTL before, but I'm going to recommend try adding the dash dash list dash reboots. This will give you a list of all the reboots along with a boot ID that you can then use with Journal CTL, dash B, and just copy and paste the boot ID right after space to get more detailed information about each reboot. Now you can always review these notes in our command line tip spreadsheet that is pinned under our Untitled Linux Show, or ULS, open discussion thread in Discord.

01:18:31 - Jonathan Bennett (Host)
Yeah, some interesting tips there. It seems like it would be really really neat to maybe put several of these together into a little script and run it automatically whenever you boot in Excuse me, when you SSH in, so like I'm imagining servers that I remotely administrate, it could be really useful to see the list of the last X number of logins and make sure they're all me from IP addresses that I recognize, because if you suddenly have an SSH connection from somewhere you don't recognize, you've got a problem that you need to deal with, and it's better to figure that out sooner rather than later.

01:19:09 - Ken McDonald (Host)
Or put them into another recent tip. I did the terminal bookmark. Yeah, yeah, the terminal bookmark.

01:19:16 - Jonathan Bennett (Host)
Yeah, yeah, you do that too, all right. Well, I've got one as well, a command line tip, and it's really really simple but it might be really useful for you, and that is the ping command. We've talked about ping before. I think it's been a little while. Ping lets you send a it's actually an ICMP request, so it sends a single packet off to whatever host that you specify, and if the host receives it, it just turns right back around and sends a ping response, and so that's useful for things like can I talk to this remote host? And you also get to check how long it takes, what the round trip time is, and then, by default, ping will just it'll keep running, it'll send a ping, it'll wait a second, it'll send another ping, it'll wait a second. And so you can also get an idea of. You know not only what your latency is over time, but whether you're dropping packets in the middle. You know not only what your latency is over time, but whether you're dropping packets in the middle, and so checking to see whether your ISP is having trouble pinging out to, and I usually use Google's DNS at And so that is the sort of the use case, one of the use cases of pings not the only one, but it's one of them. And so you have.

There is a way to make ping audible. And so there's one of these really old tech support stories where a guy was working in a I think it was a school and so he had the intercom button and set the intercom up by the computer and ran ping on a host that was giving him trouble and ran it audible, and so he just then walked around through the entire school fiddling with Ethernet connectors. I think this was back in the days where they were using ThinNet or ThickNet, not Twisted Pair, so it was coax and so he would just go. He walked around and fiddled with all the connectors and was listening to the ping, going over the intercom system everywhere and was using audible ping to eventually find the loose connection. But anyway, there is a set of flags that you can use with ping. You can give it a dash, four or a dash six.

Oh yeah, I suppose Daniel Schwartz on YouTube points out that control C. If you run ping and you want it to stop, control C is the break command, yes, very useful. You don't want to get stuck in ping any more than you want to get stuck in VI. So back to the tip. You can use the dash four or the dash six to force ping to use either an IPv4 address or an IPv6. And it's particularly useful if you're using host names. So if you have a system like a network like mine, where you've given out both IPv4 and IPv6 addresses to a computer, you may want to force ping to ping over one of those or the other. So it's just ping-4 and then a host name or ping-6 and a host name, and that can also be useful. If you want to check do I have IPv6 connectivity? Is there anything weird going on with my IPv4 or my IPv6 network? So really simple tip but probably useful. Useful to you, especially those of us that have actual IPv6 connectivity.

01:22:32 - Ken McDonald (Host)
And if you don't want to have it going continuous, a dash C followed by how many times you want it to count how many times out it goes.

01:22:42 - Jonathan Bennett (Host)
Yes, that's the ping count. There's a bunch of flags for ping that are useful, so you know. You can check out all of them by using man. So man space ping and it'll get you into the man page and if you're in man and you're stuck, it's the Q button to get out. By the way, control C doesn't work there. And then there's also a TLDR. Tldr space ping will give you, if TLDR is installed, of course we'll give you the too long didn't read manual for ping and of course that just sort of gives you the uh most.

Oh, my goodness, there's no tldr entry for ping. What, what? That can't be right. That can't be right. There must be something wrong with my uh, with my machine. That can't be right. Yeah, there, here it is. There is a TLDR for ping. Something wrong with my machine, my TLDR install? I'll have to figure that out. Time to update. It's time to update. Need to go to Fedora on the laptop, all right, well, it has been fun. We have hit the end of our tips, we've covered the stories for the week and I'm going to let Ken and Jeff get the last word in and we're like can go first, anything you want to plug, just, it is yours. The floor is. The floor is yours.

01:24:07 - Ken McDonald (Host)
Well, if you look over my left shoulder, you will see that I finally built my new system. I finally built my new system. It has the AMD Ryzen 7 with the integrated Radeon GPU, 16 gigabytes of memory, a 1 terabyte NVMe SST and a 4 terabyte spinning hard drive, one I had lying around. Yeah, I did install OpenSUSE Tumbleweed and noticed it has a 6.9.1-1-Default kernel and I went with the XFCE 4.18 for the desktop.

01:24:45 - Jonathan Bennett (Host)
Now, ken, you know you're running a new enough computer. Now you don't have to run XFCE.

01:24:54 - Ken McDonald (Host)
I like it.

01:25:00 - Jonathan Bennett (Host)
Okay, here's what you need to do With the new computer. You need to take a couple of days, maybe up to a week, and run all of the major desktop environments. Give GNOME a try for a week, give KDE a try for a week and just see if maybe, there is indeed greener grass on the other side of the DE fence.

01:25:20 - Ken McDonald (Host)

01:25:24 - Jonathan Bennett (Host)

01:25:24 - Ken McDonald (Host)
But yeah, I was noticing when I was installing it. It gave me the option of choosing between all three, so I may go back in and see if I can set it up to choose which environment I use when I log in.

01:25:40 - Jonathan Bennett (Host)
Usually if you have multiple environments installed, when you click on your username and you type in your password, so your whatever that thing is called a display manager, a desktop manager, whatever like SSD, sddm or GDM. Display Manager, desktop Manager, whatever like SSD, sddm or GDM usually like in the bottom left-hand corner, there'll be an option to click to change which desktop type you want to go into.

01:26:01 - Ken McDonald (Host)
Yeah, my old Lenovo Think Center Ubuntu. It's still on 20.04, had that option, went in and played with using Gnome and switching to just Wayland. Interesting what wasn't there after I logged in that I was used to having there.

01:26:28 - Jonathan Bennett (Host)
Yeah well, I mean, things are different now. It's been a while. It might be time to look again. All right, jeffff anything from you.

01:26:37 - Jeff Massie (Host)
Uh, only only to say if you want to try wayland, wait, wait till six or six one in your distribution. Don't run wayland on five, dot anything. Oh, and katie on katie. Yeah, other other than that I don't have any coffee, so there's no way. Uh, I can say 10 coffees equals a raspberry pie. I don't know, maybe I'll have to set that up or something, but I'll just leave with some poetry. This site has moved. We'll tell you where, but then we'd have to delete you. Thank you everybody. Have a great week.

01:27:14 - Jonathan Bennett (Host)
Excellent, all right, and, of course, if you want to find my work, you can find it all over on. Well, it's a Hackaday page. I need to scroll back up. I was actually reading the page so you can't tell what it's about. I think that is a miniature Blu-ray player that somebody was wrote up. Anyway, you can find my stuff over on Hackaday.

I've got the security column goes live every Friday morning and then we also do floss weekly there these days having a lot of fun with that and that records for now on Wednesdays. We've considered moving the recording date to Tuesdays. We will see if that actually happens. And then, of course, you could follow me on Twitter and YouTube. It's a JP underscore, bennett, most of the places. Yeah, would love to have you there. You can buy me a coffee even if you want to, if you really want to, but I'm doing okay.

Yeah, thank you guys for being here and thank you everyone for coming and watching. One more thing I do want to plug quickly is that you can join the club and, in fact, with the new streaming software that we're testing out, there's even a QR code that is covering up. Yeah, it's there. Join the club if you want more, if you want to be on Discord, if you want to catch the video of the show, if you want to be able to chat live, the club twits it is the place to be. And, yeah, thank you. You everyone, those that caught us live and those on the download we sure do appreciate it and we will see you next week on the untitled linux show.


All Transcripts posts