Some Of The Features You Can Expect With Linux 6.7

[phoronix]

Phoronix: Some Of The Features You Can Expect With Linux 6.7

With Linux 6.6 expected to be released tomorrow as stable, the Linux 6.7 merge window in turn will be opened. Here's a preview of some of the changes expected for this next kernel cycle...

https://www.phoronix.com/news/Linux-6.7-Expected-Features

[timofonic]

I really hope Bcachefs gets merged this time!

Any LKML mails or something? Any news? I see a constant flux of patches on linux-bcachefs mailing list.

https://lore.kernel.org/linux-bcachefs/

[fitzie]

I really hope Bcachefs gets merged this time!

Any LKML mails or something? Any news? I see a constant flux of patches on linux-bcachefs mailing list.

https://lore.kernel.org/linux-bcachefs/

I've been following this a bit. There's a NAK from mingo from about two weeks ago, due to Kent not working constructively on locking. And Kent went away for a few weeks, and no one was updating the bcachefs branch into linux-next so it was 3 weeks stale. But finally someone other then Kent started to do that, and then Kent himself, so it seems like we're in good shape for a potential merge. If it goes down like 6.6, then we'll hear nothing from Kent until halfway between 6.7-rc1 -> 6.7-rc2, and then Kent will send off an email asking for submission. Honestly if he had some EQ, he'd be talking to VFS/locking maintainers and even Christoph offline to ensure they will not NAK him again. However, he didn't do that during 6.6, much to his own detriment causing his meltdown. I do think the mingo issue is really the only potential issue, but and I hope it is brought up, because it does seem like they need to have a game plan for one sufficient locking system and not two. But I think the only way it's a showstopper is if Kent throws a fit when it is discussed. Linus has been clear that working well with the guidelines (such as they are) and maintainers is a hard requirement, and Kent usually has a acerbic righteousness tone and attitude.

I'll say that there was probably a hundred fixes due to bcachefs getting into linux-next, so Kent was also totally wrongheaded there. And there's been a lot more work on cross endian/32bit systems, while mostly unimportant, was entirely overlooked and untested, but that's starting to get addressed.

I'm a long term supporter of bcachefs and donate to bcachefs/Kent on patreon, but I think delaying to this point is really the best timing all along. For example, they don't have to deal with LTS releases for another year, I think makes a lot of sense, especially as I expect they next few releases will get rid of a bunch of the Kent'isms that have surely built up due to this being a out of tree for so long.

[cl333r]

So basically Bcachefs has all the benefits of "modern" file systems (ZFS, Btrfs) but without their downsides (being slower)?

[Developer12]

So basically Bcachefs has all the benefits of "modern" file systems (ZFS, Btrfs) but without their downsides (being slower)?

It's still just as slow.

The linux kernel tried to produce it's own in-tree ZFS ("with blackjack and hookers") but ended up with the supremely buggy BTRFS. 10-15 years of development haven't meaningfully reduced the number of footguns in BTRFS, so this is a new filesystem that might, maybe, become a worthy in-tree alternative to ZFS.

That's pretty much it. Bcachefs is attempt number 2 at "let's build a modern filesystem."

[mrg666]

Actually, I am only expecting to boot Linux 6.6 tomorrow.

[Congelli501]

It's still just as slow.

The linux kernel tried to produce it's own in-tree ZFS ("with blackjack and hookers") but ended up with the supremely buggy BTRFS. 10-15 years of development haven't meaningfully reduced the number of footguns in BTRFS, so this is a new filesystem that might, maybe, become a worthy in-tree alternative to ZFS.

Btrfs is not buggy, or at least haven't been buggy for 8 years or more. It's used in production system of big companies like Facebook.
I have been using using btrfs on top of drbd for years without issues on multiple servers and hundreds of subvolumes.

[andyprough]

I don't use AMD due to the lack of coreboot support. Looks like the one item that's of interest for me is the zero-copy support for the VirtIO/Vsock. I'm always game for vm speed-ups, and this sounds very pleasant.

I don't really care about the BCA chefs, unless they know how to make good Korean short ribs.

[zexelon]

Btrfs is not buggy, or at least haven't been buggy for 8 years or more. It's used in production system of big companies like Facebook.
I have been using using btrfs on top of drbd for years without issues on multiple servers and hundreds of subvolumes.

Just to be clear Meta/FB does not use Btrfs like any normal person would. They use a very specific set of its features, so if your using it in the same way you can expect good behavior as its what they maintain... try to run any of the more common "desktop" features and its definitely more questionable.

[rmfx]

Where are the fast kernel headers ?