Can Trisquel make use of old 2.6 kernel from RedHat?
Hey,
So, I noticed that RedHat (and by extension, CentOS) still support the 2.6 kernel in their 6 release. I was wondering, in light of some users wanting to run Trisquel in old machines that can't handle modern software, could a Linux-Libre be made out of the 2.6 from RedHad (which has been getting backported security updates for years)?
Or for that matter, is there a Libre version of RedHat/CentOS?
In case someone thinks this is fairly obvious and I shouldn't be asking the question, my apologies, I am not making fun, I seriously think it could be an option. If not, please let me know why.
Thanks!
I don't know that Trisquel would ever do this as an official thing but third parties might offer a repository from which such a kernel could be installed. That would still leave the issue of how to actually do the install on that machine in the first place since the installer on the Trisquel ISO would still be running the newer one. Why can't these older machines use a newer kernel?
Something else to consider: RHEL 6 and CentOS 6 are only supposed until 30 November 2020, so about a year a 5 months. What's to happen once they become end-of-life and there no more updates? I am not sure this is sustainable.
Well, it's certainly better in terms of security updates than running Trisquel 4 or 5, right? Those stopped receiving updates many years ago already.
From what I read, RedHat has the longest support in linux distros (10 years). Not even Debian or Canonical support their OS for so long (5 years)
"Those stopped receiving updates many years ago"
Sure but I think you missed or at least didn't answer what I was wondering: What's the plan once that can be said about 2.6.32 from Red Hat/CentOS? Do you even have one? Without one you'll just end up in the same boat as currently; later for sure but eventually. This just kicks the can down the road.
"I thought a Pentium 2 or 3 couldn't run a Linux kernel 4.4"
I'm not so sure about that. :)
"So RedHat 6 should be less demanding than Trisquel 8, as an example? In the same way Trisquel 4 should be less demanding than Trisquel 8?"
I imagine that depends on various factors like the software being used.
Well, you are right in that I don't have a plan for that. BUT it gives us still a year and half ahead, and even so, in 2021 it would be preferable to use a kernel a few months outdated than a few years outdated. Of course we will need a better solution then, but it's the best one I see right now.
My question is if it is possible or not, that's what I don't know.
Maybe you can clarify, is a 2.6 kernel less demanding in terms of CPU, RAM, etc, than a 4.4?
As for the software, I would maybe suggest not using a GUI at all, or something like OpenBox. But that's the "easy" part, anyone can remove LibreOffice and install Abiword for example. The Kernel and is where I feel the problem lays. Again, please correct me if I am wrong, and also let us know if a 2.6 kernel can still be used in a modern OS or not.
"My question is if it is possible or not, that's what I don't know."
Of course it's *possible*. I have checked and the normal deblob scripts for Linux-libre don't work because Red Hat/CentOS have changed things around in such a way that the script fails. This is not surprising though: The Linux-libre deblob scripts are intended for use with the kernel that comes from kernel.org. Trisquel, for example, already needs to change the deblob scripts around in order to fit with what Ubuntu does to their kernel. And so the ongoing responsibility would be that someone would need to keep the deblob script updated and maintained for it, monitor Red Hat/CentOS for new versions, host the source code somewhere, and potentially host the binaries somewhere if people want binaries.
The hard part would therefore to change the script to suit the RedHat kernel, right? Assuming one wouldn't need to share the resulting kernel, hosting and such would be left out.
Just give me your opinion on my question above please, is the 4.4 kernel less demanding than 2.6? CPU, RAM, etc. Thanks.
"The hard part would therefore to change the script to suit the RedHat kernel, right? Assuming one wouldn't need to share the resulting kernel, hosting and such would be left out."
Sure.
"Just give me your opinion on my question above please, is the 4.4 kernel less demanding than 2.6? CPU, RAM, etc. Thanks."
Look at https://invidio.us/watch?v=Hlr60CAkUQQ which is a 3.16 series kernel.
And compare that to: https://invidio.us/watch?v=8PZW0qrsJAU which is a 2.3 series kernel.
Linux got fat.
But you're also not using a 30 year old Motorola 68000 CPU so your computer might be just fine. There's only one way to know: Try it! :)
I thought a Pentium 2 or 3 couldn't run a Linux kernel 4.4, but an older one like the 2.6 would be lightweight in comparison?
I mean, the Linux Kernel 10 years ago was less demanding in terms of computer power correct?
So RedHat 6 should be less demanding than Trisquel 8, as an example? In the same way Trisquel 4 should be less demanding than Trisquel 8?
ConnochaetOS 14.2 was using a 4.4 kernel, like Trisquel 8, and here were its hardware requirements: https://connochaetos.org/wiki/docs/requirements
Yes, and there are plans to maintain 4.4 for a long time:
https://www.cip-project.org/blog/2016/10/13/civil-infrastructure-platform-announces-first-super-long-term-support-kernel-at-embedded-linux-conference-europe has that "The CIP community plans to maintain 4.4 for security and bug fixes for more than 10 years."
It's possible that Linux-libre might do that as well then although I haven't really discussed it Alexandre Oliva.
Indeed, you are correct. I was merely hoping we could lower those requirements from a Pentium2 to a Pentium3. However, from jxself's explanation it seems that there is not much sense in trying to run an older kernel just for the sake of it. Still, Linux did get fat... Maybe we don't need a Linux-Libre, but instead a Linux-Slimmed-Down-Version-Libre :P
Anyway, if the kernel number version doesn't matter in terms of system requirements, it's no use to keep maintaining an old one. This thread is finished.
Thanks.