Linux 6.10 To Drop Support For Very Old DEC Alpha Hardware

[uid313]

Which ones, pray tell, and for what reason?

I don't know, like all of them other than x86, ARM and RISC-V. Oh well, maybe keep POWER around for a while.

[dragorth]

Just drop all DEC Alpha code from the kernel. We don't need it, I don't think anyone use it.
Drop other old architectures too.

There is a Linux Distro that is basically dedicated to running on these old systems. Its T2SDE. At least the maintainer is using it, and he has users.

He is also the guy that is trying to bring Itanium back.

[KeyboardG]

Just drop all DEC Alpha code from the kernel. We don't need it, I don't think anyone use it.
Drop other old architectures too.

If you aren't running that hardware, it doesn't even get loaded into memory. There is zero cost to you to have that code support in the Kernal.

[Dawn]

I don't know, like all of them other than x86, ARM and RISC-V. Oh well, maybe keep POWER around for a while.

There is more growth on Z than on PPC though?? Seems inconsistent. What about ARC? It gets new IP every couple of years. Loongarch? It only just got brand new hardware. Hexagon is actively developed by one of the biggest semi companies in the world. C-sky is hardly uncommon.

[uid313]

There is more growth on Z than on PPC though?? Seems inconsistent. What about ARC? It gets new IP every couple of years. Loongarch? It only just got brand new hardware. Hexagon is actively developed by one of the biggest semi companies in the world. C-sky is hardly uncommon.

Hmm, I don't know, Z has to be kept because IBM. Maybe LoongArch has to be kept if many Chinese want it, but I Think ARC, C-Sky, SuperH, etc should go.

[torsionbar28]

Hmm, I don't know, Z has to be kept because IBM. Maybe LoongArch has to be kept if many Chinese want it, but I Think ARC, C-Sky, SuperH, etc should go.

Which part of the kernel are you a maintainer for? And which code specifically are these cpu architectures causing you to do extra work?

[uid313]

Which part of the kernel are you a maintainer for? And which code specifically are these cpu architectures causing you to do extra work?

None, but the world doesn't need all these architectures, especially the ones that nobody use. It's just non-sense to keep them around. They're keeping them around for no reason.

[moltonel]

There is zero cost to you to have that code support in the Kernal.

There absolutely is :

• Supporting an architecture takes time (at best an extra location to mechanically update, at worst blocking a refactor due to being the lower common denominator), which developers won't spend on features and bugfixes that impact you.
• Because Linux is hugely influential in FOSS, this also applies to other projects (distros, languages, libs...), which are not likely to support an arch that Linux doesn't.
• This applies to your own projects, if you've ever been told by potential users "don't use $TECH, because it doesn't support $ARCH".