11 Features That Didn't Make It In 2020 For Linux 5.11

For as great as Linux 5.11 is with its new features, there is also some prominent material that has yet to be upstreamed into the mainline kernel -- some of which is likely to hit in 2021 while other changes have less likely ambitions for mainline.

What is not found in Linux 5.11 include these notable 11 features, if I missed any other pending work worth mentioning feel free to call it out in the comments via the forums:

PREEMPT_RT - Seemingly each year there is talk of landing the real-time (PREEMPT_RT) patches "soon" to the kernel. At the 2020 Linux Plumbers Conference there was talk of landing the PREEMPT_RT in mainline in the very near future. While months have passed and a few kernel releases since then, the RT patches are still being maintained out of tree. We'll see if 2021 is the lucky year for PREEMPT_RT in mainline.

Bcachefs - The Bcachefs file-system born out of the Linux block cache code has shown much progress over 2020. In December all known blockers were resolved and a pull request issued but it wasn't acted upon for Linux 5.11. We'll see if it comes in 2021.

The New NTFS Driver - One of the surprises of 2020 was Paragon Software offering up their previously commercial NTFS kernel driver for the mainline kernel and then taking the steps to improve the code and get it into good quality for mainline. This driver is read/write and much better off feature wise than the current kernel driver as well as the FUSE-based user-space NTFS-3G option. The new driver is getting into good shape and will hopefully be merged in the next cycle or two.

FUTEX2 - Collabora in cooperation with Valve have been working on the new Futex2 system call to help with the Steam Play / Proton efforts. So far though futex2 hasn't been sent in for mainline.

Intel AMX - One of the big features coming with Intel Xeon "Sapphire Rapids" is AMX as the Advanced Matrix Extensions and should be quite interesting for a number of different use-cases. Intel has spent months getting the compiler toolchain support around LLVM in order and towards the end of the year has been focusing on the Intel AMX patches for the Linux kernel. We'll likely see that mainlined within a cycle or two, still well in time before Sapphire Rapids CPUs ship.

Intel Multi-GPU Support - Intel spent a lot of time in 2020 working to bring-up their discrete graphics support and the DG1/SG1 and related discrete graphics chips. But for those with Intel Xe MAX notebooks or other Intel multi-GPU setups, that isn't yet in nice shape. The Intel multi-GPU / Xe MAX support is currently a mess on Linux needing to make use of virtualization if wanting to currently use multiple Intel GPUs. Hopefully their DRM kernel driver will get squared away soon for better dealing with multiple GPUs. Well, we have full faith they will, just too bad it's taken so long with Xe MAX laptops already in the wild.

DXGKRNL as the DirectX Kernel Driver - Microsoft as part of their WSL efforts this year published a "DirectX Kernel Driver" for Linux in catering to the Windows Subsystem for Linux use-case. But mainlining this DXGKRNL driver will be a challenge. Microsoft hasn't posted any new kernel patches lately for review on this driver and upstream developers aren't too eager to land it for various reasons. We'll see what happens in 2021.

DAMON - Amazon engineers have been working on DAMON for monitoring data accesses. They continue going through new rounds of review and likely will see it mainlined this year.

Reiser4/Reiser5 - It was nearly one year ago to the day that Reiser5 was announced. Several times through the year new features have been talked about by lead developer Edward Shishkin. But it's still not clear if there is enough interest or support to get Reiser5 mainlined where Reiser4 had failed to do so.

READFILE - Over the past year Greg KH has been working on a "readfile" system call as a single, simple system call to quickly and easily read a small file such as what is exposed via sysfs. Readfile has interest and will likely be mainlined. Greg recently revised the readfile patches in his Git branch, so maybe for Linux 5.12 we'll see it land.

BUS1 / In-Kernel IPC - After the failed KDBUS, BUS1 for in-kernel IPC was being pursued but it hasn't been talked about much in recent years. The BUS1 code is touched occasionally these days besides the Dbus-Broker effort that is for user-space and retains D-Bus compatibility while being focused on performance and reliability.