Originally posted by bitterseeds
View Post
Announcement
Collapse
No announcement yet.
Linux 6.6 Formally Becomes This Year's LTS Kernel
Collapse
X
-
As someone who compiles the kernel themselves to this day, these LTS kernels have been a huge convenience. I only have to deal with kernel build configuration once a year.
- Likes 2
Leave a comment:
-
Originally posted by andyprough View PostThere were quite a few systems that could not be upgraded from 4.9 to 4.14 this past year. When 4.14 and 4.19 reach EOL, expect that a much larger number of systems are going to fail to work properly with newer kernels.
- Likes 1
Leave a comment:
-
Originally posted by treba View Post
Yeah, but such devices won't necessarily get updates even if the kernels are still maintained. Such vendors are IMO just bad actors which have to be forced to get better, either by consumers or regulators. Such as the EU, which will start requiring vendors to support hardware for at least 5 years if they want to sell their products at all. Which, with shorter LTS cycles, will mean they'll be forced to update their kernels. Which IMO is a good thing.
Originally posted by stormcrow View Post
I think it's arguable Motorola's disclaimer is actually enforceable. Many US states don't allow warranties to exclude 3rd party maintenance.
There's more models than that, but y'all get the point. It's actually kind of funny in a sad way how it's possible to can tell how oppressive a country is by how they force phone companies to make country-exclusive models and what those models actually limit.
It really sucks how all that capitalistic corruption and oppression is powered by Linux. Possibly Linux 6.6 LTS.
I wish they could sneak in a GPL 2 revision making it a license violation to put GPL'd on things with locked platforms where the binary can't be replaced by the user.
- Likes 3
Leave a comment:
-
It would be interesting to see benchmarks comparing the last few LTS kernels (6.6, 6.1, 5.15, 5.10, 5.4), and see how performance has evolved over the longer term. It would need to be the most recent release of each, to get a fair picture of how the backported fixes and mitigations may have affected things.
- Likes 2
Leave a comment:
-
Originally posted by andyprough View PostThere were quite a few systems that could not be upgraded from 4.9 to 4.14 this past year. When 4.14 and 4.19 reach EOL, expect that a much larger number of systems are going to fail to work properly with newer kernels.
While it's good that newer machines are more energy efficient, the obverse is true about decisions like this increasing the rate of e-waste accumulation. Computers should be made to last much longer and to use fewer toxic and rare earth metal components. Constantly chasing Moore's Law speed increases at all costs is not without longterm harm for the environment and mankind.
I do have several "slightly dated" systems (some youngsters would scream "ancient"), and I haven't gotten around to put fresh kernels on all of them, however, most of them run fairly well. Some are VIA C3/C7, AMD Geode LX, somewhere's a GX floating around, SiS 55x and such boxes. I remember the SiS 55x having issues with the IDE driver during boot negotiating the transfer mode, which sucks, (esp. as it ends up on same lame PIO mode), and I must check something above 6.1. that caused severe instabilities on S2RAM on my modern AMD boxes. But usually the kernel is (at least for me) usually the sanest software in the whole chain from FW to userland.
But just in case there is an issue I'd like to be informed.
It may be that they removed Hercules Monochrome special driver and well, i386 support. However, i486 should still be fine (as I compiled Gentoo on old ones with i486 settings).
Leave a comment:
-
Originally posted by andyprough View PostThere were quite a few systems that could not be upgraded from 4.9 to 4.14 this past year. When 4.14 and 4.19 reach EOL, expect that a much larger number of systems are going to fail to work properly with newer kernels.
While it's good that newer machines are more energy efficient, the obverse is true about decisions like this increasing the rate of e-waste accumulation. Computers should be made to last much longer and to use fewer toxic and rare earth metal components. Constantly chasing Moore's Law speed increases at all costs is not without longterm harm for the environment and mankind.
Hardware that runs with a kernel like 4.14 but not 6.x falls into two categories:
- Special purpose system
- Unusual hardware with some external modules
- Systems which are decades old
If it is a special purpose system, then it should have a strong firewall config or not be connected to the internet at all, even with a recent system. Upgrading the kernel is not needed, the security should begin at a much earlier level. It can happily run with a 3.x kernel for another decade if you know your physical security and take if offline. Or it is a museum piece, then it is also offline.
If it is unusual hardware (e.g. embedded) and needs some special kernel modules and such, complain to the upstream hardware vendor for not working with the kernel. Take valve for an example of how to do it right… the SteamDeck will be supported even a decade after it becomes obsolete.
And if you system is 20–30 years old, yes, then newer kernels regularly break things. True. But if that is a system you regularly use and need to keep online, then the energy consumption compared to something like a rasperry pi zero (which will probably outperform it) does in fact justify getting rid of the old system, especially if you recycle it correctly.
Same argument when features are removed from the kernel... 99% of affected systems are perfectly okay to stay on an old kernel since they serve some special (museum) purpose.
On top, the sanity of kernel devs does also have a big impact of the whole ecosystem as well, leading to better supported devices, leading to less e-waste.
- Likes 2
Leave a comment:
-
of course you should be using kernel that has some level of support. people don't understand how LTS isn't a panacea. Either the patches are bot selected and you hope that the maintainer pays attention for bad backports or the subsystem is manually maintained. In some cases, like XFS, they didn't even have the bandwidth to maintain all the LTS kernels, so they didn't. (See: https://lwn.net/Articles/934941/ )
- Likes 2
Leave a comment:
-
Originally posted by treba View PostYeah, but such devices won't necessarily get updates even if the kernels are still maintained. Such vendors are IMO just bad actors which have to be forced to get better, either by consumers or regulators. Such as the EU, which will start requiring vendors to support hardware for at least 5 years if they want to sell their products at all. Which, with shorter LTS cycles, will mean they'll be forced to update their kernels. Which IMO is a good thing.
Leave a comment:
Leave a comment: