FUTEX2's sys_futex_waitv() Sent In For Linux 5.16 To Help Linux Gaming

[damentz]

I like how you throw Liquorix into the mix (see what I did there?) just for advertisement's sake!

Anyhow, now that MuQSS is history, will Liquorix come crawling back to CFS?

Yes, that was a shameless plug, but I also think it's unfair to suggest a kernel as if it's the only solution to someone who doesn't know any better.

But regarding MuQSS, I think it was a good thing that it was discontinued. It figuratively ripped off the band-aid that had been festering for the last 12 months for both Con and the users of his process scheduler.

For me, it began the unpleasant research of what the current state of process schedulers on Linux was. Fortunately, the timing of Con's retirement to maintaining MuQSS couldn't have come at a better time. Alfred's PDS is MuQSS' spiritual sequel or successor. It improves on BFS/MuQSS in some major ways, with single threaded performance being one of the largest.

In other words, the latest version of Liquorix you can get today is very much superior to all the versions before 5.14.

[theriddick]

I'm taking a guess here after looking at the patches for Liquorix that it doesn't include NTFS3 and I would need the dkms module for that?

[CochainComplex]

It's important to note here that XanMod is just one of many kernels that support fsync. If you're on an Arch based kernel, Zen Kernel is first party and will get you what you're looking for without much effort (Garuda Linux comes to mind). Liquorix is also available as binaries on Arch, Debian, and Ubuntu with different behavior/performance optimizations.

Yes, indeed. Its not the only one having the patchset applied. But the only one I'm using and where I can say that it is working. At least the log suggest a successfully activation of futex2 mechanisms.
Besides quoting myself:

...and your Kernel e.g. the Xanmod kernel have the latest patchset implemented...

this implies that Xanmod is not the only kernel existing with futex2 patch

I can only recommend things I have testes. And since I have never tested Liquorix, Zen extensively I can't seriously recommend them without ignoring my lack of experience.

[CochainComplex]

I like how you throw Liquorix into the mix (see what I did there?) just for advertisement's sake!

Anyhow, now that MuQSS is history, will Liquorix come crawling back to CFS?

Cacule seems to be a good candidate. I don't know the details of the different schedulers. But Cacule makes the system very responsive. Especially in gameing it works quite fine.

[oiaohm]

this implies that Xanmod is not the only kernel existing with futex2 patch

I can only recommend things I have testes. And since I have never tested Liquorix I can't give a serious recommendation.

Using the futex2 patch set before the peer review process has been complete has been a very good way to end up with a system dead lock. Getting locking code correct is not a easy process.

Futex2 mainline as taken so long due to how often the patches have been put forwards and automated tools or humans have looked at the code and gone bugger me that not right. The not right been some very bad outcomes.

Bleeding edge has that name for a very good reason.

[damentz]

Using the futex2 patch set before the peer review process has been complete has been a very good way to end up with a system dead lock. Getting locking code correct is not a easy process.

Futex2 mainline as taken so long due to how often the patches have been put forwards and automated tools or humans have looked at the code and gone bugger me that not right. The not right been some very bad outcomes.

Bleeding edge has that name for a very good reason.

Yes, I tried the latest futex2 code a few times and it was always worse than the futex-multiple-wait that came before it. I'll wait for the official backports before merging them into Zen Kernel.