The Linux Kernel Keeps Dropping Hardware Support!
HTML-код
- Опубликовано: 26 сен 2024
- If you run mainstream hardware Linux support is great but as of late a lot of older hardware support has been dropped alongside some hardware that's actually kind of new like Itanium.
Full Episode: • #223 The Linux Distro ...
==========Support The Channel==========
► Patreon: / brodierobertson
► Paypal: www.paypal.me/...
► Amazon USA: amzn.to/3d5gykF
► Other Methods: cointr.ee/brod...
==========Guest Links==========
T2 SDE: t2sde.org/
Code Therapy: / @morerenerebe
Bits Inside: / renerebe
Blog: rene.rebe.de/
=========Audio Release=========
🎵 RSS: anchor.fm/s/14...
🎵 Apple Podcast: podcasts.apple...
🎵 Spotify: open.spotify.c...
🎵 Google Podcast: www.google.com...
=========Video Platforms==========
🎥 LBRY: odysee.com/@Te...
==========Social Media==========
🎤 Discord: brodierobertso...
🐦 Twitter: / techoverteashow
📷 Instagram: / techovertea
🌐 Mastodon: brodierobertso...
DISCLOSURE: Wherever possible I use referral links, which means if you click one of the links in this video or description and make a purchase we may receive a small commission or other compensation.
#Linux #OpenSource #FOSS #LinuxDistro #Server
I am a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and related sites.
🎵 Intro Music
Aces High by Kevin MacLeod
Link: incompetech.fi...
License: creativecommons...
The policy of removal from the kernel is simply "hey, is anyone using this on a new kernel?" - no response. Mark for deleting, if anyone really do run 6.9 kernel on Itanium and actively care for its support they would yell. Itanium machines usually just run ancient stuff, something like 4.9 or 5.4 kernel is "modern". And they have no plans to upgrade beyond that, the machines will be retired,
So no PDP-11 support then?
Not everyone follows the mailing list, there are plenty of silent users that will only realise things are broke when the next LTS kernel comea around and suddently stuff is broken.
It's more a related to maintenance than "anyone using this" thing.
@@JoaquinVacas but he says in the video itself that it's not about maintenance since it's already working.
@@xmlthegreat maybe not in all the cases but for example there are some RNDIS drivers plus really older WLAN cards that don't make sense to keep maintaining (I think this is more of a security concern than a utility one).
I run an NVIDIA 8400GS GPU in my TrueNAS Scale because it is a PCIE 1x slot so I can pass the ARC A380 to my Jellyfin container. I worry about long term driver support.
I think it's completely reasonable to drop shpport for hardware from the 90s. If we're to use the car analogy from the podcast, thats like asking Ford to keep supporting the T100. I'm sorry, but the 90s CPUs compared to the current ones are that antiquated.
If you want to use your vintage machine, like some people use vintage cars, your best bet is not to rely on the kernel for support. The car DIY guys make custom parts to repair thise vehicles so in a similar sence you'd also have to make compromises to run your vintage PC parts.
I have no idea why HP would buy Itanium parts in 2020, but uh.... that's not my or your problem. That's an HP problem.
For 99.999999% of users there's no point in having drivers for hardware from the 90s so I don't think it makes any sense to keep support. It's not something you can just keep around forever just because a handful of people might use it
Itanium is from just some years ago though, ...
I think itanium was removed more because there wasn't that much demand or desire for itanium for a while despite it technically still being "produced" on paper (and also that it is a nightmare of an architecture to work with). "just leave the drivers" isn't really great for maintenance purposes since if it is an official driver, you have an implied "guarantee" by the linux project that these drivers have high code quality and security, which cannot really be guaranteed for software that hasn't been touched for 10 years. but at the end of the day, you can always maintain the drivers separately if you need them
The idea i get is the thing was working dont see reason to remove it
Just officially call it depreciated it then? Send a message and say it's depreciated in the syslog, but if it doesn't cause any build conflicts and remains functional there's not really harm in leaving it around for the handful of people who use it. Why is there legendary rants about not breaking user-space but you can just make hardware not work?
@@iotku Yes, but how many people read syslogs before complaining? And besides, at that stage it would have the same effect as completely removing it from the tree - downstream distros will have to maintain their own patches anyway
@@AmansLab Well, it is working *now*, but what about the next update? You'd essentially have to go over every "deprecated" driver and verify that it is in fact working, and at that point it is not really "0 effort maintenance" is it? Especially with hardware configurations that are not exactly common - you'll need dedicated maintainers that have access to such outdated/rare hardware
@@switchblade6226 it is working now and was working before. if some thing break it can be handled later
The rate at which projects like Armbian are dropping embedded hardware that is very widely deployed is a major PITA. A deployed base in the embedded world can be huge, and people are not going to replace all the systems inside their ATM machines or highway signs or Taxi meters or gas pumps in order to use some new communication features.
The dropping of hardware support is sad and nothing to be celebrated about. Yes even hardware of the 90s. Some say the rate of dropping hardware support has become worse in recent years. If true, that is a worrisome trend and must be corrected.
go support the 90s hardware them, they drop when it's not used, not maintened
@@vilian9185 not true. Usually just works, or needed a one line fix ;-)
@@renerebe if it is only a line of code to fix, and no one bothered to fix it in years, so it's only a indicative that no one uses that anymore
@@vilian9185 It took my 8h to find this one line of code. How many other people are skilled to do that? Also upstream broke it in a pointless API change ¯\_(ツ)_/¯
@@renerebe fair
If you run a vintage machine in your museum, just run it on a vintage kernel then. I don't really see much of a problem here. His argument, hey it worked, why won't you carry it on. And I say, if it worked why don't you use then what worked, why does this have to be dragged along forever. In the long run the kernel sources can't grow and grow forever. There are space and compiling time considerations here.
Museum? You sound like the type of person who upgrades their phone as soon as a new model comes out. I'll answer your question with a question. Why carry hardware support on? Why throw perfectly working hardware into landfill like a discount Bill Gates? I thought one of the aims of Linux was to support old hardware? I still use old laptops from the 2000s instead of throwing them out thanks to Linux. Using old unsupported kernals is a security risk.
@@edstar83no one is using hardware this old in any way that requires a modern kernel. And if you are, reverting the commit is trivial.
They always ask if anyone is using the hardware before deleting it. Deleting dead code is an important part of maintaining a code base.
@@edstar83 René at 7:21 explicitly refers to some of his older computers as a 'living museum.' In the same way that you're not using a vintage car on the freeway for your daily commute, it's highly unlikely that you're running vintage hardware for the sake of getting real work done, and to that point, there is little reason to burden maintainers with something you keep around for hobby tinkering. It's also not like it's impossible to patch support back in if it's actually found to be necessary.
@@cavedweller333 the new kernels often performance better, have of course security fixed, and not only was Itanium sold until recently, we have customers who run it in production ;-)
I don't know what I'm talking about, but sounds like this wouldn't even be an issue if it wasn't monolithic
the audio on these rene clips are great!!
the fact that we are so dependant on upstream shipping a support is imo making linux brittle
If you run hardware from 1980 then run it with 1980 OS, as simple as that.
@@DanielRumbacher You are right, things have changed. It's because Linux is now driven not by the community but corporations like Red Hat, Canonical, Valve, NVidia. Community things are the same as 10 years ago and all the progress made to the kernel or drivers is corporate contributions. Linux is finally becoming a viable alternative to Windows for the average user, and yes I can see it becoming more commercial. I have no objections to pay for my copy of Kubuntu in the future. And I can see it coming as soon as it becomes stable and fully reliable. We now are just beta testing it for free. Microsoft was right in their leaked "Halloween files" that community driven projects can't compete with a corporation as it needs funds. There will always be niche distros for geeks and nerds tho.
@@GamingPenguinEnthusiast "It's because Linux is now driven not by the community but corporations"
that the case for about two decades now (at least when we only consider the kernel)
A lot of people came into Linux from the idea going around that you can install it on older hardware and use things again that would have had to have been thrown out because Windows stopped working on it. You want to be like Windows?
It used to be long time "Hey you can get Linux on this old machine to keep up to date and keep using your machine". Now it is sadly this mainstream Windows and Apple style OS where they keep dropping support just so they get sold latest revision.
Got me all excited, thinking you guys already did another video together! Love Rene's content, looking forward to more.
RENE❤❤
On one hand it would be nice if there was a stable driver ABI so it was easier for drivers to live outside of the kernel and not be as reliant on being part on it but on the flip side this would probably also encourage more crappy proprietary drivers
but the ABI is stable?
@@vilian9185 the driver ABI (and even the API) is famously and publicly NOT stable
you may mean the system call ABI, which very much IS stable
Rene do you have an Amiga 500?
Unfortunately not. Maybe one day, or better yet a fat'er Amiga that can boot Linux ;-)
I know the Russians were experimenting with Itanium-like architecture. I wonder, if this has anything to do with that or not...
I mean if the Russians really wanted linux for their architecture they can probably just make a proprietary fork of it, since there's really no way of enforcing GPL over there.
Too conspiratorial
No itanium in Russia. Their chip is Elbrus. Only relation to itanium is both are VLIW designs. And also got shitcanned because Russia's biggest institutions reviewed Elbrus systems and essentially called it a terrible joke.
what lmao