Linux kernel removing 486 support

[CNK]

I saw that this fix might be related to a regression introduced in TCL 16 where I think I saw you reported it would not boot on a VIA C7 CPU. Are you able to verify whether this patch fixes the booting on C7 as well?

I think I did try my stripped-down 486-patched kernel build on the VIA C7 in case one of the kernel features I disabled was causing the crash there, but it had the same problem.

As for the patch to fix the kernel for 486 CPUs, that was for code used early in the boot process before VIA C7 CPUs have problems. It's hard to see how it could be related at all.

[andyj]

Some of us remember this: https://www.networkworld.com/article/743590/opensource-subnet-how-removing-386-support-in-linux-will-destroy-the-world.html. I get the nostalgia, like classic cars, but a car you've spent a year's worth of weekends fixing up isn't a daily driver. Some of the extensions I maintain, like rsyslog and mariadb, require that the CPU be set to i586 as they will no longer compile for i486. The end is already here.

[CNK]

I get the nostalgia, like classic cars, but a car you've spent a year's worth of weekends fixing up isn't a daily driver.

Yeah well each to their own. That description fits the 36 year old car that's my daily driver pretty well.

Some of the extensions I maintain, like rsyslog and mariadb, require that the CPU be set to i586 as they will no longer compile for i486. The end is already here.

Some programs like Firefox have required SSE2 for a long time already, but the range of programs one can really run on a 486 is limited anyway due to its performance. Firefox is now in the process of dropping 32-bit Linux support entirely, which unfortunately might be the next stage for some other software projects too.

[youxiaojie]

This conversation came up a few years back, but looks like it will happen now. Ingo Molnar wants to remove 486 support from the Linux kernel:
x86: Remove support for TSC-less and CX8-less CPUs

really?kernel also showed CPU=M486?

[youxiaojie]

I am also porting alpinelinux to 486

[youxiaojie]

does the original initramfs contain xserver? if so I want to clean it and try to make it as an tcz

[CNK]

The X server is supplied as a tcz extension, but not in initramfs. I haven't tried X on a 486 in Tiny Core, but Xvesa.tcz should work if you have enough RAM.

[Paul_123]

I guess we’ll know soon if this is happening.   Another article,  we are getting quoted too.

https://arstechnica.com/gadgets/2026/04/linux-kernel-maintainers-are-following-through-on-removing-intel-486-support/

[CNK]

Once the M486 options are removed from here, I think that confirms the patch to remove 486 support is going through to the next Linux kernel release.

I'm taking it as a cue to look deeper into NetBSD, but I'm not sure how easy it is to set up a NetBSD system  that's as lightweight as Tiny Core Linux.

[CNK]

Once the M486 options are removed from here, I think that confirms the patch to remove 486 support is going through to the next Linux kernel release.

Gone now.

Next Linux kernel release is v. 7.1.

[nick65go]

https://www.phoronix.com/news/Linux-7.1-Begins-Removing-i486
Linus Torvalds Merged The Code Beginning To Remove Intel 486 CPU Support In Linux 7.1

the initial removal is getting rid of the kernel configuration (Kconfig) options for building M486 / M486SX / ELAN CPU support. With that Kconfig code now removed, building a Linux 7.1+ kernel with that Intel 486 support isn't possible and specifically affects AMD/Cyrix/IBM/Intel SL/SLC/SLC2/SLC3/SX/SX2 and UMC U5S. The M486 Kconfig option is for 486-class CPUs such as the AMD/Cyrix/IBM/Intel 486DX/DX2/DX4 and UMC U5D.

So some filesystem affected also (nfs pointers) + increase RAM on 32 bits usage because CPU cache pipeline (L1/L2) disturbed.

[nick65go]

After the "full elimination' of 486 CPU from kernel (but not from old, already existing programs) will be interesting the direction and competitiveness for small (special distro) like Tinycore, antiX, Pupy linux, etc. When base will be 586 or 686 and new programs version making more intense (ab)use of SSE# or AVX instructions.

Moreover, is not only 586+ CPU_type, we now enter the range for RAM minim/recommended competition.

And almost continuously bug discovery from AI for old programs. I do not talk about big software like LibreOffice, Firefox, but about critical ones like kernel, Openssl, Xorg / Wayland.

It will be a sharper split between (unpaid/ voluntary) developers of open-source and common (grab it for free) users/consumers. As we already see now in TC contributions