The New Features Of The Linux 5.7 Kernel: Tiger Lake Graphics Stable, New exFAT, Zstd F2FS, Performance

[phoronix]

Phoronix: The New Features Of The Linux 5.7 Kernel: Tiger Lake Graphics Stable, New exFAT, Zstd F2FS, Performance

Linus Torvalds is expected to deliver an Easter day kernel release in the form of Linux 5.7-rc1. After a two week merge window, Linux 5.7 feature development formally ends today. Here is a look at the many exciting improvements and new features to find with Linux 5.7.

http://www.phoronix.com/vr.php?view=29069

[Vorpal]

Typo: "XFS is preparing for online repair support and other nderlying improvements."

[skeevy420]

5.9 is looking to be a nice LTS kernel. It can't get here fast enough.

I'm a bit weird in that I treat LTS releases like most people treat stable point releases and that I treat stable point releases like they're beta releases and only use stable point releases if I necessarily have to; like with the current 5.6 series so I can test out the new BTRFS checksum methods from 5.5. I'm mainly interested in xxhash vs the default/old crc32c since I don't have a need for the more hardcore methods of sha256 & blake2.

Started designing a system from the ground up when I started realizing that there are so many BTRFS setting and mount setting combinations that I actually need to run benchmarks versus going with what makes sense. Stuff like lazytime vs relatime, xxhash vs crc32c, compress and compress-force to test BTRFS vs kernel compressor algorithms, frickin 6-15 compression levels depending on the compressor used, double the aforementioned compression levels due to the aforementioned compress and compress-force settings, space cache v1 & v2....bloody hell....

[Paradigm Shifter]

5.9 is looking to be a nice LTS kernel. It can't get here fast enough.

I'm a bit weird in that I treat LTS releases like most people treat stable point releases and that I treat stable point releases like they're beta releases and only use stable point releases if I necessarily have to...

I really don't see why that is weird. LTS releases are (supposed to be) peace of mind for people who want their systems to work. I want good LTS kernels on mission critical systems, I'm happy to tinker around with non-LTS on testing boxes.

...

Anyway, what I want to see is the Intel iGPU driver stopping all of the softlock/GPU hanging shenanigans of the last n kernels. It's utterly ridiculous - on Ubuntu it's now infected kernel 4.15.0-96 with this soft lock nonsense. Every kernel newer than 4.15.0-91 does it, but the newer they are the worse it is.

To say I'm aggravated with this is putting it mildly.

[skeevy420]

I really don't see why that is weird. LTS releases are (supposed to be) peace of mind for people who want their systems to work. I want good LTS kernels on mission critical systems, I'm happy to tinker around with non-LTS on testing boxes.

...

Anyway, what I want to see is the Intel iGPU driver stopping all of the softlock/GPU hanging shenanigans of the last n kernels. It's utterly ridiculous - on Ubuntu it's now infected kernel 4.15.0-96 with this soft lock nonsense. Every kernel newer than 4.15.0-91 does it, but the newer they are the worse it is.

To say I'm aggravated with this is putting it mildly.

It's weird on Phoronix where it's required to run everything so bleeding edge your system can be mistaken for a slasher film...but for regular people I suppose it's closer to normal.

[tildearrow]

Typo: "XFS is preparing for online repair support and other nderlying improvements."

Really ,-,

[betam4x]

I really don't see why that is weird. LTS releases are (supposed to be) peace of mind for people who want their systems to work. I want good LTS kernels on mission critical systems, I'm happy to tinker around with non-LTS on testing boxes.

...

Anyway, what I want to see is the Intel iGPU driver stopping all of the softlock/GPU hanging shenanigans of the last n kernels. It's utterly ridiculous - on Ubuntu it's now infected kernel 4.15.0-96 with this soft lock nonsense. Every kernel newer than 4.15.0-91 does it, but the newer they are the worse it is.

To say I'm aggravated with this is putting it mildly.

I've never had an issue with kernel stability, and I run dedicated and cloud based servers, 3 different desktop environments (including a laptop), and lastly, a small embedded machine. All are on Linux 5.6. All of the machines have required the occasional reboot for updates, but none of them has had instability of any short (and that applies to the whole distro, not just the kernel).

I am of the belief that LTS releases in general are an example of observer bias.

[grigi]

Mostly I'd say that on older sytems, LTS releases break less stuff. E.g. a 12-year old notebook runs noticeably more stable on 4.14 than on newer kernels as they stop testing on such hardware. And sometimes disable something that isn't reliable for ALL of those systems (a good example is i915 on pre-haswell got worse, and those GPU's are so weak they need all the help they can get0.

Another thing is, especially with all the urgent security fixes getting rushed in, i've definitely had stability issues. So I tend to want to update less often now.

Sometimes you just need the new feature/bugfix, but right now I'm happy with 5.4LTS. Newer kernels don't bring me anything right now. Although the scheduler changes on 5.7 might be worth it...

[Paradigm Shifter]

I've never had an issue with kernel stability...

Obviously you don't have any systems with Intel iGPU, then. You're lucky.

edit: Or tried getting RAID working on a recent AMD system.

[AndyChow]

I've never had an issue with kernel stability, and I run dedicated and cloud based servers, 3 different desktop environments (including a laptop), and lastly, a small embedded machine. All are on Linux 5.6. All of the machines have required the occasional reboot for updates, but none of them has had instability of any short (and that applies to the whole distro, not just the kernel).

I am of the belief that LTS releases in general are an example of observer bias.

I've personally had to switch my kernel to LTS twice this year, when ZFS broke. Happened on 5.5 and 5.6 I believe.