Six Features Not In The Mainline Linux 5.6 Kernel
AMD Sensor Fusion Hub Driver - Sought after by AMD Linux laptop users since 2018 and finally published in early January was the AMD Sensor Fusion Hub driver for Linux. This driver is akin to Intel's Sensor Hub driver and is used for enabling the accelerometer and gyroscopic sensors on AMD Ryzen laptops and other similar functionality. Originally AMD engineers hoped to release this driver during "summer 2019" but only arrived in January for improving the AMD Linux laptop experience. At the end of the month there was a second version of this driver but unfortunately it didn't get reviewed and vetted in time for inclusion during the Linux 5.6 merge window. Sometimes Linus Torvalds does allow new drivers to be merged after the merge window when it doesn't risk regressing any existing code/support, which is the case here, but more than likely the AMD-SFH-HID driver will end up having to wait until Linux 5.7 to see mainline.
Reiser5 - While Linux 5.6 does bring the Zonefs file-system and other storage improvements, not found there is Reiser5 nor Reiser4. There seems little chance in years that Reiser4 would reach mainline status while at the end of last year was the surprise of Reiser5 being in development. But it's certainly not ready for even mainline's staging area and since then Reiser5 announcement around Christmas, it's been rather silent... Reiser4 and Reiser5 will likely be out-of-tree for the foreseeable future unless a lot of interest happens to build around this new Reiser file-system version.
Bcachefs - Sought after by some and standing much better chances of being mainlined compared to Reiser* is Bcachefs, the file-system born out of the Linux kernel's "Bcache" block cache code. Kent Overstreet continues working on Bcachefs but isn't ready for mainline yet. Most recently he's been working towards snapshot support for the file-system, an experimental FUSE port for testing, various performance optimizations, and more. The Bcachefs code continues to be revised and perhaps later this year we'll see this goal realized.
BUS1 In-Kernel IPC - Born out of the failure of KDBUS as an in-kernel D-Bus implementation was BUS1 as a new capability-based inter-process communication implementation for Linux systems. The BUS1 kernel code hasn't been touched since last March and does appear to be stalled. But at least Dbus-Broker out of BUS1 as a D-Bus compliant implementation focused on better performance and reliability than the reference D-Bus implementation has been working out well at least until a better solution is ready.
Intel SGX Linux Enclave Driver - A lot and drawn out process on the Intel open-source side has been getting the Software Guard Extensions support squared away. The SGX foundations code for the Linux kernel and bringing up an SGX Enclave driver for dealing with Intel CPUs Memory Encryption Engine has been a long time coming. But even after over two dozen revisions, it's still not ready for mainline. Intel though is working on it and just days ago was SGX Foundations v26. Perhaps we'll see the code finally mainlined later this year.
Valve's Futex() Optimization - Last summer Valve talked up changes to the Linux kernel to extending the futex() system call needed for more optimal thread pool synchronization. Valve has been carrying the user-space patches within Proton as well as having posted some Glibc experimental patches and has made Linux kernel patches of the necessary futex change for easy testing by Linux gamers. The patch for futex to implementing the mechanism to wait on any of several futexes has yet to be mainlined. The patch was written by Collabora in collaboration with Valve. Hopefully this work and any other kernel optimizations will be on tap for the summer.
What other features are still missing from the mainline Linux kernel that you would like to see? Let us know in the forums.
41 Comments
