Linux 5.16 Features Include FUTEX2, Intel AMX, Folios, DG2/Alchemist, More Apple Silicon Support

[phoronix]

Phoronix: Linux 5.16 Features Include FUTEX2, Intel AMX, Folios, DG2/Alchemist, More Apple Silicon Support

Now that the Linux 5.16 merge window has ended with yesterday's Linux 5.16-rc1 release, here is my lengthy original overview of what I find most interesting out of this new kernel version. Linux 5.16 won't be out as stable until around the end of the calendar year or early next year, but it will sure make one nice Christmas gift with all of the shiny new features in tow.

https://www.phoronix.com/vr.php?view=30689

[skeevy420]

What a great release 5.16 is going to be

[perpetually high]

I noticed HID_NINTENDO is mainlined, but not even added as a module. So even when 5.16 gets released, gonna have to compile your own kernel, or a kernel flavor that bakes it in (like xanmo). Still waiting to see if I can verify the new FUTEX2 in 5.16-rc1 if anyone knows for certain or how to verify.

Ubuntu 5.16 default:

$ grep -ri 'hid_nintendo'
./ubuntu-5.16-defaults/config.common.ubuntu:# CONFIG_HID_NINTENDO is not set

Perpetually High's kernel:

$ grep -ri 'hid_nintendo' /boot/config-5.16-rc1-051600rc1+customidle-generic
CONFIG_HID_NINTENDO=m

[oskar-n]

$ git log --grep futex2
commit 595b28fb0c8949463d8ec1e485f36d17c870ddb2
Merge: 91e1c99e175a f98a3dccfcb0
Author: Linus Torvalds <[email protected]>
Date: Mon Nov 1 13:15:36 2021 -0700

Merge tag 'locking-core-2021-10-31' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull locking updates from Thomas Gleixner:

- Move futex code into kernel/futex/ and split up the kitchen sink into
seperate files to make integration of sys_futex_waitv() simpler.

- Add a new sys_futex_waitv() syscall which allows to wait on multiple
futexes.

The main use case is emulating Windows' WaitForMultipleObjects which
allows Wine to improve the performance of Windows Games. Also native
Linux games can benefit from this interface as this is a common wait
pattern for this kind of applications.
....

[oskar-n]

@perpetually high
It looks like it's using CONFIG_FUTEX

[Vorpal]

I'm happy that the sensors for my Asus motherboard will be supported without having to patch and build my own kernel. Not that it is hard to do so on Arch Linux, but it is annoying and time consuming.

Still need the zenpower dkms module to get all the data from my Ryzen CPU, anyone know why that isn't being upstreamed?

[F.Ultra]

I noticed HID_NINTENDO is mainlined, but not even added as a module. So even when 5.16 gets released, gonna have to compile your own kernel, or a kernel flavor that bakes it in (like xanmo). Still waiting to see if I can verify the new FUTEX2 in 5.16-rc1 if anyone knows for certain or how to verify.

Easiest way is to simply look at the actual source code https://git.kernel.org/pub/scm/linux....c?h=v5.16-rc1 and there you find:

/** * sys_futex_waitv - Wait on a list of futexes
* @waiters: List of futexes to wait on
* @nr_futexes: Length of futexv
* @flags: Flag for timeout (monotonic/realtime)
* @timeout: Optional absolute timeout.
* @clockid: Clock to be used for the timeout, realtime or monotonic.
*
* Given an array of `struct futex_waitv`, wait on each uaddr. The thread wakes
* if a futex_wake() is performed at any uaddr. The syscall returns immediately
* if any waiter has *uaddr != val. *timeout is an optional timeout value for
* the operation. Each waiter has individual flags. The `flags` argument for
* the syscall should be used solely for specifying the timeout as realtime, if
* needed. Flags for private futexes, sizes, etc. should be used on the
* individual flags of each waiter.
*
* Returns the array index of one of the woken futexes. No further information
* is provided: any number of other futexes may also have been woken by the
* same event, and if more than one futex was woken, the retrned index may
* refer to any one of them. (It is not necessaryily the futex with the
* smallest index, nor the one most recently woken, nor...)
*/
SYSCALL_DEFINE5(futex_waitv, struct futex_waitv __user *, waiters, unsigned int, nr_futexes, unsigned int, flags, struct __kernel_timespec __user *, timeout, clockid_t, clockid) {

And you can find the Nintendo Joycon driver/module here https://git.kernel.org/pub/scm/linux....c?h=v5.16-rc1

[4BvFCHGlspF4]

For me personally, the retpoline work stands out the most. Can't exactly understand it in full but I'll welcome any and all work to improve upon the software mitigations that were needed to deal with hardware problems that were entirely avoidable with better design. Not enough praise can be given, as far as I'm concerned, to the people working on optimizing those mitigations.

[pierro78]

dear linux experts
sorry for being so dumb but I have the ES8336 sound card in my Chuwi Laptop, not playing sound to internal devices ... in the pull request https://github.com/thesofproject/linux/pull/2962 the dev says a kernel fix "should land in kernel 5.16 at best (still missing the updated codec parts)" : do you think this could solve my issue ?
thanks !