I'm going to be that guy, but I believe the correct term is kernel.

UCyborg wrote on 2025-04-29, 19:42:

I'm going to be that guy, but I believe the correct term is kernel.

That's correct. Unless you are Commodore.

It was always annoying to have to write up an issue with a web app's handling of the "Referer" header and then have the technical writers correct the spelling

mkarcher wrote on 2025-04-29, 18:55:

This is because the 6x86 omitted some officially optional features of the 586 architecture,

Were 586 and 686 architectures/instruction set levels an official thing? Especially 686 always seemed a colloquial term, up to compiler vendors and projects like the Linux kernel to decide what it means.

edit: Or was it an extrapolation of the CPUID family field being a 5 or 6?

jakethompson1 wrote on 2025-04-29, 20:04:

Were 586 and 686 architectures/instruction set levels an official thing? [...]

edit: Or was it an extrapolation of the CPUID family field being a 5 or 6?

That's what I mean by "586" and "686" architecture. You might as well say "x86 architecture, family 5" and "x86 architecture, family 6", which is probably formally more correct.

I guess practically it affects only a few modern industrial i486 clones, but those were designed to work in old OS environment anyway, not the other way around.

kagura1050 wrote on 2025-04-29, 07:59:

The concept of the distribution is "use and preserve 486", so somehow I feel like I've lost the battle.

You'll still have 486 support in the current LTS kernel for a number of years, and the SLTS kernel for 10+ years past that.

SquallStrife wrote on 2025-04-29, 03:57:

Sunsetting official support for a 30+ year old CPU is hardly what I'd call drastic.

Hi, okay, then here's something more drastic. ;)
Linux also drops support for PAE on all 32-Bit CPUs soon.
https://www.phoronix.com/news/Linux-6.14-x86-CPU-32-bit

So using more than 4GB of RAM is nolonger being possible on 32-Bit Linux.
This technically also affects my VIA C7D based motherboard that I used to use in early 2010s.

Jo22 wrote on 2025-04-29, 01:37:

In the future, such older 486/586 designs could still be relevant for sake of independence, maybe. As some sort of blue print fo […]
Show full quote

In the future, such older 486/586 designs could still be relevant for sake of independence, maybe.
As some sort of blue print for newer designs made in smaller fabrication process.
Which might allow manufacturing in smaller countries, too, not just among the big players in industry.

So they might be not useful for gaming PCs, but for automation and embedded use.
ARM is an modern alternative for sure but is also being heavily being protected, I suppose.
Making unlicensed clones of it could cause trouble. RISC V may or may not play a role eventually.

The RISC-V architecture is available in both 64-bit and 32-bit variants, so I doubt if any countries would be interested in the ability to continue independently manufacturing 32-bit x86 chips, especially older 486 variants. I don't think there is a lot of demand for that, even in the embedded space.

As for continued OS support for 486 and non-Pentium-compatible "586" chips, would NetBSD continue to be an option for those? I believe I read the lack of a hardware FPU potentially being a problem on that OS. Perhaps the x87 FPU emulator being removed from the Linux kernel could be spun off into some kind of independent portable module?

UPDATE: Well, isn't that funny. I searched online for FPU emulation support in NetBSD, and the first thing that came up was this new project that was announced just a few days ago:

i486SX_soft_FPU — a software FPU emulator for the classic Intel 486SX CPU, running on NetBSD 10

@kagura1050 Maybe you and your friend could base your 486 retro distro on this instead?

Jo22 wrote on 2025-04-28, 18:48:

I for one see no real value for Linux to begin with.

No OS war here, but the enterprise and cloud worldwide run on Linux, mostly, worldwide.

Possibly the idea from kernel maintainers is any nowadays distro out there stopped crafting 32 bit x86 compatible builds for a while already, moving completely to x86_64 and/or aarch64 (ARM). So there's really no need for having so old CPUs supported by the kernel.
It's an end of an era anyway as Linus himself started the kernel project with his 386.

fix_metal wrote on Yesterday, 07:19:

Jo22 wrote on 2025-04-28, 18:48:

I for one see no real value for Linux to begin with.

No OS war here, but the enterprise and cloud worldwide run on Linux, mostly, worldwide.

The problem is the cloud idea itself, I think.
Where I live, people are too much dependent on the online services from the west.
All our data goes were it doesn't belong to, where it isn't safe.
It might be worth to consider going back to previous model and store data more locally.
Or check out alternatives, use places which are neutral (Swiss?)..

fix_metal wrote on Yesterday, 07:19:

Possibly the idea from kernel maintainers is any nowadays distro out there stopped crafting 32 bit x86 compatible builds for a while already, moving completely to x86_64 and/or aarch64 (ARM). So there's really no need for having so old CPUs supported by the kernel.

I wonder why they bother to support 32-Bit PCs at all then.
It would be less of an headache to users if they knew when the last, fully functional release was available.

However, what's happening now looks more like cutting a whole tree down, one by one.
In the process, each branch with its flowers is cut step by step, until merely the stomp remains.
Eventuall this stomp will be the final, the greatest x86 Linux release there is. Or so it may seem to laymen by that time.

If you want your pretty 32-Bit x86 tree back, you have to go back year by year and check the changelogs.
Until you're confident that the tree now looks pretty enough or complete enough to you.

fix_metal wrote on Yesterday, 07:19:

It's an end of an era anyway as Linus himself started the kernel project with his 386.

Probably. I do have a boxed copy of SuSe Linux here that says 486DX - Pentium 4..
And that's not just for the kernel, but are the requirements for the whole OS distribution.

Also note that 386 support had been dropped 25 years ago already (It's expliticly excluded on the box).
By contrast, the 486/586 generation was considered current technology, still.
(The box also states that the systems without co-processor aren't being supported to begin with.)

It's both funny and confusing seeing this, considering the current change.
To me, Linux was always about being a free OS to scrapyard hardware.
That means, random surplus hardware that had been cobbled together by poor students (symbolical).

(Edit: That's how the google servers used to work originally, by the way.
They were dirty, yellowed roadside PCs with a HDD and a network card, literally!
The installation was done via network boot. Googles greatest feat was to tuck them together via velcro tapes.)

All you needed was a supported CD-ROM drive and then crossing your fingers that
on one of these 10 CDs was having a working set of drivers that matched the configuration.

Linux was good for giving outdated hardware a second life a decade later, in short. Like an old webcam. That was its main purpose, I think.
That's why I'm kind of confused by the current change.
You might think that Linux suddenly claims to be cutting-edge technology. Completely absurd to me.

PS: A product review about that SuSe generation can be found here:
https://dl.acm.org/doi/fullHtml/10.5555/364682.364705

At the end of the day, the use case for modern linux kernels on i386/i486 hardware is probably next to nonexistant. I really doubt anyone here could find a viable use case aside from FAFO experimentation.

appiah4 wrote on Yesterday, 13:32:

At the end of the day, the use case for modern linux kernels on i386/i486 hardware is probably next to nonexistant. I really doubt anyone here could find a viable use case aside from FAFO experimentation.

Probably true. On other hand, if not for sake of old hardware, then what's Linux actually good for? It's not the only *nix system out there is.
All my life, I saw the main use case of Linux to rescue/recycle old hardware.
It's cheap from a license standpoint and has good legacy support.

So what's exactly left if older CPUs are being dropped from support every once in a year now? 🤷‍♂️
Does the use of current kernal releases comes to an halt or does it slow down?
Do older Linux kernels get more attention (use, bugfixes, new features), thus?
I'm not sure if such a development would be better for Linux's future.

PS: What if developers eventually decide that all BIOS-only PCs are obsolete in general?
Or that newer Linux kernals nolonger run on PCs without TPM functionality?
At which point do they draw the line? 🤷‍♂️

Because in the end, this won't be different to Windows development.
New PCs can't run 32-Bit OSes anymore that rely on BIOS/CSM (Win 98SE, XP).
By definition, these 32-Bit platforms are obsolete, after all.
By contrast, Linux still can boot here, for the moment. It can boot via UEFI/BIOS, both in 32-Bit/64-Bit editions.

Edit: Then there's another, more philosophical question here.
The question whether or not numbers play a role (aka market share thinking).
At which point wrong becomes right? Or vice versa?

Is the leading force being dedication/principles or market share?
That's important, I think. Because market share leads to commercial thinking, rather than to altruistic or social thinking.

Imagine, a new Linux kernel was akin to a cure for a second.
Wouldn't this mean that dropping CPU support (for a minority) would be same as not giving a minority of patients their cure?

The dropping of the 486 "because of numbers" is akin to not producing a cure for dying people,
because they're "just" a minority of low number.

In real life, that's how economic thinking decides between life and death.
If not enough (wealthy) people are infected by a disease, then it's commercial not worth it to start production of a cure.

You guys may think this is overreacting, but wasn't the Open Source being lead by principles, originally?
How does economic/marketshare thinking fits in here?

My Raspberry Pi 1 is 15 years old, too.
When does Linux kernel drop support for old ARM processors, too?
How many Raspberry Pi 1 (in numbers) must exactly be in use to keep support alive?

Jo22 wrote on Yesterday, 14:16:

Probably true. On other hand, if not for sake of old hardware, then what's Linux actually good for? It's not the only *nix system out there is.
All my life, I saw the main use case of Linux to rescue/recycle old hardware.

Say hello to broken old proprietary Nvidia drivers in new Linux kernels. Yeah, there are free Nouveau drivers, but they are still shit to this day.

Jo22 wrote on Yesterday, 14:16:

if not for sake of old hardware, then what's Linux actually good for?

You keep saying this...

Meanwhile, somewhere between 80 and 96% of web servers run Linux (depending on how you measure it).

All but a handful of the supercomputers on the Top 500 list run Linux.

Linux Kernel is used by the Android operating system, powering a good chunk of smartphones, tablets, and other embedded devices.

Why?

Linux Kernel is well understood, very portable, open source, gratis, well supported, highly customisable and modular, and has a proven track record of reliability.

The *BSD derivatives are a distant second by nearly every metric. Commercial UNIX is practically dead. Even stoic, conservative, risk-averse IBM has Linux on z/Series.

Why don't you tell us... what ISN'T it good for?

SquallStrife wrote on Yesterday, 22:42:

Even stoic, conservative, risk-averse IBM has Linux on z/Series.

IBM bought Red Hat for that. They stand behind one of the oldest pillars of Linux distribution.

The Serpent Rider wrote on Yesterday, 22:48:

SquallStrife wrote on Yesterday, 22:42:

Even stoic, conservative, risk-averse IBM has Linux on z/Series.

IBM bought Red Hat for that. They stand behind one of the oldest pillars of Linux distribution.

Indeed!

Jo22 wrote on Yesterday, 14:16:

Probably true. On other hand, if not for sake of old hardware, then what's Linux actually good for? It's not the only *nix system out there is.

I don't think there are any other mainstream *nix out there, so...

Jo22 wrote on Yesterday, 14:16:

All my life, I saw the main use case of Linux to rescue/recycle old hardware.

Linux when it first started out, was kind of bleeding edge for the hardware at the time. 32-bit OS, heavy *nix like. It only got the way you are saying, because the devs have not been intentionally break things that could just work, and let them be. So it allowed longevity well beyond the expected lifespan that other vendors did not support.

But if you are in a middle of rewrite of some code, that no one wants to rewrite it, or it is getting in the way, that is what the devs are saying now. They have been pretty careful and not making drastic changes. We are talking about approximately 34 years of 486 support in linux here. So they should get credit for leaving it alone longer than anyone else.

Pretty much every other vendor, will break and remove things if it doesn't suit their business.

Jo22 wrote on Yesterday, 14:16:

My Raspberry Pi 1 is 15 years old, too.
When does Linux kernel drop support for old ARM processors, too?
How many Raspberry Pi 1 (in numbers) must exactly be in use to keep support alive?

I don't think Raspberry Pi 1 has much going on anyway. I don't think the devs have removed that ARM proc. But ARM support is a strange beast so you never know. It may be worse there on a code maintenance standpoint, and unique devices such as Pi 1 might get the ax sooner. But again, I don't think they are going to try to inflict anything on you just because they want to. Anyhow, I think it might be hard to find distros still going specifically for the Pi 1 at this point too. (Debian? I have no idea, but even they set limits) So what does it matter that the kernel has to, if the distros don't either. Most any distro don't support i486 now at this time, at least as far as what most people see.

I think I see your question being proposed a bit alarmist, as if they don't keep supporting something with updates, that we can't consider it worth our time. But I'd say that Linux, and maybe a few other like, *BSD should be commended for their long lasting support, for that it was ever done. No one else does things like this, and in some ways, this long support was likely completely free to you.

One question is that why do we want an OS that has to be constantly updated? I don't want it. Maybe that could be considered a failing of Linux, but that is still unlike Windows/Mac that is forced to be updated, and then support is removed. So if you ask me, I'm getting away from updates, because it isn't a better way. So if I run Linux on a 486, will I just use an old Linux? (or Linux kernel specifically?) Why yes, because it's still there, and it doesn't need updates to keep it running.

It's also worth remembering that Debian, a distro famously focussed on compatibility and stability, stopped supporting Pentium 1 class processors in Stretch, well ahead of the kernel itself:

https://lists.debian.org/debian-devel-announc … 5/msg00001.html

You haven't been able to run any practical distribution on a 486 in *years*, and even after this change is merged to production, there are still plenty of options for using Linux on those CPUs. It's a storm in a teacup.

Niche distros exist, (S)LTS kernels exist, almost nobody cares about 486s for non-retro-gaming purposes, and the few that do, understand that this change is a non-event in those use cases.

While messing with some recent (as in 201X) linux distros in PCem a few years ago (like slitaz), you begin to realize the Cyrix6x86 has merit. 😀

There's also deprecation for non-pae which shuts out good CPUs like Pentium M. M!!!