they are falling short of target
Rip also i586. Make i686 the minimum baseline for Linux.
Kernel 6.10 LTS-CIP has 10 years of support. That ought to be enogh for such old hardware. Even embedded
Patch to end i486 support hits Linux kernel merge queue
It's taken nearly a full version number to get the pieces in order, but the long-awaited end of 486 chip support in the Linux kernel appears to be nigh with Linux 7.1's release later this year. Slated for the 7.1 merge window is a patch that veteran Linux kernel contributor Ingo Molnar queued up at the end of March, but which …
-
-
-
Monday 6th April 2026 17:45 GMT bombastic bob
Re: they are falling short of target
agreed. As long as the older code remains relatively unchanged it would only require "being there" and not having the basic functionality of that code altered. My guess is that libraries and kernel code can live with this until it is clear that all 32-bit support needs to be dropped. At some point, 2038 as I recall, 32-bit time_t will no longer suffice. BSD defines it as a 64-bit value now as I recall. MANY 32-bit Linux kernels [like in embedded systems] still use 32-bit time_t, so when the next epoch limit is reached, 32-bit should be gone, even on embedded. [Major distros and glibc apparently use a compile option for 64-bit time_t at the 32-bit application level].
In any case this final hurdle must undoubtedly come at some point: No more 32-bit Linux (and *BSD) builds outside of legacy code.
-
Monday 6th April 2026 19:05 GMT toejam++
Re: they are falling short of target
I wouldn't be surprised to see NetBSD keep the i386 port around for decades, even if it falls down to tier-2 status. The Commodore Amiga 680x0 and DEC Alpha are still supported by NetBSD as tier-2 systems, and they're both fairly niche now. And NetBSD has supported a 64-bit time_t under i386 since 6.0 (2012).
-
Monday 6th April 2026 21:35 GMT williamyf
Re: they are falling short of target
Tell me you jumped straight to the comments section without reading the article without saying "I came to te comment section without reading the article"
The article quotes Ingo Monlar saying: "We have various complicated hardware emulation facilities on x86-32 to support ancient 32-bit CPUs that very very few people are using with modern kernels," Molnar explained. "This compatibility glue is sometimes even causing problems that people spend time to resolve, which time could be spent on other things."
the same applies to i586
-
Tuesday 7th April 2026 08:03 GMT Roland6
Re: they are falling short of target
There is an interesting thread discussing what is a “586”: ”How you define a 586”
Given the move to chipset architectures, it would seem future generations of CPU’s will be equally complicated to support…
However, with respect to the 586, we should not forget Intel only stopped production in 2023.
-
Thursday 9th April 2026 23:15 GMT Blazde
Re: they are falling short of target
the same applies to i586
It doesn't though? The kernel list discussion mentions that the overwhelming part of dropped code will be FPU emulation to support the 486SX. There were absolutely massive architecture changes between 386 and 486, and some more minor but still very significant ones between 386 and 486. But by '586' the instruction interface had matured and almost all changes an OS needs to care about are flagged as extensions with, in theory, no guarantee that future chips will support them. Do you want to drop support for single cores? No. Drop MMX support for some truly negligible performance gain? Drop 32-bit support? Hell no. Or drop support for CPUs without a TPM like Microsoft (kinda) did...?
-
-
-
-
Monday 6th April 2026 17:47 GMT alkasetzer
Re: they are falling short of target
Why stop at i586, that wastes developer time and can cause issues that would best be served but restricting each new kernel major to the latest spec-3. For example 7.0 should be limited in support for arm 8, x86-64v2, power 8, risc-v (some subset), by 9.0 Linux would only be geared for x86-64v4 at a minimum and arm 10, power 10, etc.
That would also force hardware makers to produce new arch versions each two years which is also great for the marketing and helps sell new hardware.
And users of older kit can just run the latest unsupported kernel version because the developers would also not be keeping around museum hardware.
/End of sarcasm
On a serious note, while I agree that baselines should be set (and make sense), how those baselines are communicated matter.
Also if Linux was stable between releases no one would care. As is, your kernel driver works with that specific kernel version, so you can not, for example, use your old hardware and change the grapichs card to something new because the drm is smack middle in the kernel and needs tweaks every kernel release.
-
-
-
Tuesday 7th April 2026 05:46 GMT Eric 9001
Re: Buggritt
The Intel Atom N270 is a i686 CPU, which was very common in other netbooks too.
Despite being 32bit, it's not a bad CPU, considering that it does have constant_tsc and ssse3 - it's only really missing AES-ni, which would be nice to have.
It has low power consumption (only 2.5 W at full load - too bad the chipset is much worse) and as it doesn't have speculative execution and stupid performance shortcuts, it is immune to all Spectre attacks and meltdown too.
The only issue I've identified with the latest GNU Linux-libre is that modesetting no longer works as Intel has broken i915 for the GPU - but in GRUB_CMDLINE_LINUX you just add nomodeset, as you're only using it for ssh anyway.
The issue you have is how most up-to-date GNU distros have removed support for 32 bit and well the only sane choices remaining are Guix or Gentoo - too bad you need to be careful to not install proprietary software on either OS, as there's proprietary software in the packages tree.
Gentoo with decent software works fine with the N270 - GCC, ocaml, perl, python, asterisk, ls-sensors, glibc and openssh etc compiles fine with 1GB of RAM and 2GB of SWAP - only garbage software like LLVM won't compile.
-
Tuesday 7th April 2026 07:01 GMT blu3b3rry
Re: Buggritt
For those who are less concerned about the "proprietary" angle I found Alpine Linux ran very well indeed on my own netbook, a Samsung NC10 with the same CPU (although upgraded with 2GB RAM and fitted with a SATA SSD).
Setting it up with an XFCE desktop took a bit of tweaking with GPU drivers but it runs very snappily for what it is. I think at idle it was consuming ~200MB RAM. Ran lightweight web browsers like Falkon and Netsurf rather happily alongside Libreoffice.
Firefox was sluggish to launch but if you stuck to simpler websites and didn't use tabs it was manageable.
Another good alternative that I hear about is Antix although not got around to giving it a try yet.
-
Tuesday 7th April 2026 16:33 GMT Anonymous Coward
Re: Buggritt
> "The Intel Atom N270 is a i686 CPU, which was very common in other netbooks too. Despite being 32bit, it's not a bad CPU"
By the standards of seventeen years ago (*), it was acceptable for the cheap, low-power netbooks it was intended for, but even *then* it was still very underpowered for anything you'd want to use as a regular laptop/PC.
Today? Nah. Don't waste your time, I can't see anything based on it being powerful enough to do anything worthwhile. Certainly not enough to justify the hassle of getting 32-bit-only software running (novelty/for-the-sake-of-it attempts excepted if you want to do so anyway!)
--
(*) I had an Aspire One based on the N270, and you couldn't complain for what it was *meant* to be, for what I paid and for what one would expect circa 2009. But even then being 32-bit only already seemed an anachronism, even if it wasn't in the same market as Intel's mainstream 64-bit CPUs. (Was the N270 the last mainstream 32-bit-only x86 CPU?)
I replaced the HDD with an SSD around a decade later, since doing so could often give a new lease of life to old laptops. In this case it was still so slow I may as well not have bothered- the bottleneck was clearly the computer itself rather than the drive. It died a couple of years later anyway.
Benchmark for the Atom N270 in my Aspire One versus the Ryzen 7 in my newbuild from last year.
-
-
Tuesday 7th April 2026 18:12 GMT Anonymous Coward
Re: Buggritt
If you're that desperate or hard-up that you *have* to consider a computer that old, then virtually anything newer than that (*)- even versions of the Aspire one based on the later Atoms- is going to cost about the same, but will support 64-bit and be a *lot* less hassle to get working for that reason alone.
And even a regular fifteen-year-old laptop based on a regular CPU, however mediocre it is by today's standards or those of its time is likely to cost very little yet still outperform a 32-bit N270 several times over... because almost *anything* else will!
There's no need or compelling reason to use an N270 over anything else.
"For [..] reading pdf/word/ppt or writing simple text, it will do"
It might, just about. That's the sort of thing they were designed for. On the other hand...
"For interacting with banks and eGovt"
Are you going to be able to run a modern browser on an old, underpowered, 32-bit only machine? And, if not, is the ancient browser you found to run on that hardware going to work with modern websites? Particularly security-conscious banks?
"In developing countries like venezuela or cuba is viable for the people whose other choice is no machine at all"
If something like my Aspire One was all that was available, and there was *absolutely* no other choice, and there was absolutely no other way of doing any of those things, then of course that's what one would have to use.
You could say the same about almost anything, though, e.g. the early 2000s Pentium III laptop I got secondhand decades ago.
This all sounds a bit condescendingly like using Cuba or Venezuela as "poor starving people" stereotypes anyway. I'm not familiar with the situation in those countries specifically, but I do know that in countries where people generally can't afford modern computers, those sorts of things tend to be done on mobile phones nowadays. Rather than relying on ancient hand-me-downs those of us in the developed world don't want because they're so out of date as to be useless.
If your country is so poor that your average citizen can't afford a digital access device, then they're likely not going to have a setup where online-only access to government is the case.
(*) The only Atom CPU I can find that's newer than the N270 *and* still 32-bit-only seems to have been intended for smartphone and mobile device use.
-
Tuesday 7th April 2026 20:47 GMT williamyf
Re: Buggritt
If you search my past comments you will discover that:
1.) I am a Venezuelan (born and raised) living in Venezuela
2.) I refurbish(ed) very ancient Laptops with linux (Even Pentium4 Type Laptops) to donate to Venezuelans and Cubans (from the medical mission here) whose only other option was no laptop at all.
Yes, I am not starving, but many people here are. I may be many things, but condescending towards Venezuelans or Cubans is not one of them.
-
-
-
Tuesday 7th April 2026 18:29 GMT Roland6
Re: Buggritt
>”By the standards of seventeen years ago (*), it was acceptable for the cheap, low-power netbooks it was intended for,”
Acceptable to Intel and Microsoft may be, but not to users.
I had an Aspire One and on discovering just how compromised it was - thank you Intel and Microsoft, did very little with it; preferring to stick with lugging a full sized laptop around until the iPad was launched…
-
Tuesday 7th April 2026 18:55 GMT Anonymous Coward
Re: Buggritt
What were you trying to use your Aspire One for, though? Note that the full version of what I said was:-
By the standards of seventeen years ago (*), it was acceptable for the cheap, low-power netbooks it was intended for, but even *then* it was still very underpowered for anything you'd want to use as a regular laptop/PC.
The whole point of the Netbook thing is that they traded off power for cost and portability- they weren't originally *meant* to be full replacements for a standard x86 laptop. (Some of the earliest examples, like the Elonex ONE weren't even x86-based, nor Windows compatible, and even my x86 Aspire One came with a lightweight Linux, not Windows).
It was in part the fact that people expected to be able to run Windows on them anyway- and the fact that companies started raising the spec- and price- in response- that ultimately killed off the netbook as a distinct segment by blurring the line between them and low-end regular laptops. (The rise of the tablet probably had something to do with it as well).
-
-
-
-
Tuesday 7th April 2026 06:29 GMT Korev
Re: Buggritt
> That means I won't be ably to keep my EeePC901 up to date
I remember how they used to use the ridiculous picture of "EeePC girl" here all the time as a joke
Grey beard icon -->
-
-
Tuesday 7th April 2026 08:59 GMT Bebu sa Ware
Re: Milestone!
"I have outlived support for the first CPU I ran Linux on! Not a given in this world, and day and age."
Depending on where you live, surviving long enough to see the latest kernel evolve from .rc1 to release very much not a given. Doubly so for major Debian releases.
I suspect most of the readers here would have outlived the minicomputers, its architecture and associated OS that flourished when we entered the industry. Not to mention much of the mostly proprietary hardware and protocols that have vanished albeit largely unregretted.
-
Friday 10th April 2026 23:27 GMT Blazde
Re: Milestone!
Going down this same nostalgia path myself, I can't help wondering if Linux now works better with the mid-90s Mitsubishi parallel port external CD drive I struggled so much to get 1.0 to install from? Probably. (As much as I remember the autodetect was way off but manually selecting the correct driver after deciphering some bizarre model number formatting, putting it on the bootstrap floppy, and then accepting 1x speed reads instead of the desired 4x speed eventually worked. Possibly resolving some IRQ conflicts too. I do not miss those).
-
-
Tuesday 7th April 2026 00:53 GMT I am the Walrus
"The change has been a long time coming, and would begin the process of removing processor architecture support from the kernel for the first time since 2012, when support for 80386 processors was removed. "
"for the first time since 2012"? Itanium and Dec Alpha processor support was removed from the Lunux Kernel sometime around 2020 from memory and there was some obscure Intel Lake processor (I can't remember exactly which one) that support for was removed for last year also.
-
-
Tuesday 7th April 2026 14:39 GMT Anonymous Coward
Re: I miss numbers
Or rather 486DX vs SX. And 50MHz could be better, if the bus was also running at 50MHz. But some were running at twice the 25MHz bus speed, so could have some workloads not running as fast as a pure 33MHz CPU. And a DX100 - 50×2 or 33×3?
And then since those CPU names were just numbers that could be used by any competitor - are we talking only about Intel 486 here? Or also AMD, Cyrix, ...?
Which, by the way, how about an AMD 386DX40 was both cheaper and faster than an I486SX25? Do we get into how the 2nd level cache was not on the CPU then, but something that came with the motherboard, so the same CPU would have vastly different performance depending on where it was socketed?
I don't remember it being that clear cut nor easy to understand by the General Public...
-
-
Tuesday 7th April 2026 14:54 GMT JohnnyS777
What about space?
IIRC, Intel had a custom version of the 486 chip that was hardened for use for satellites and spacecraft, with features and shielding to deal with the issues of radiation. I don't know if that device is still in production or not, but I expect they would need to move to a custom kernel or OS if that is the case.
Anyone know?
-
Tuesday 7th April 2026 17:14 GMT williamyf
Re: What about space?
IIRC intel also had i586 and i686 class radhard devices as well.
There are a couple of 5.x and 6.x kernels in the LTS-CIP branch that will have support for 6 to 8 more years.
But in the hacylon years of i486, i586 and i686 NASA tended to use VxWork, QNX and other stuff, and NOT linux. So, not an issue.
Linux !IN SPACE! Is a quite "recent" phenomenon in the great scheme of things
-
