Linus Torvalds has just announced the official release of the new Linux kernel 6.19.
“No big surprises anywhere last week, so 6.19 is out as expected – just as the US prepares to come to a complete standstill later today watching the latest batch of televised commercials.”
Additionally, in his release announcement, Torvalds also confirmed that the next kernel release (expected mid-April) will be 7.0. The change is just a numbering reset, not a signal of a new development phase, with Torvalds citing the growing size of the 6.x series as the reason for moving to a new major version.
“I’m getting to the point where I’m being confused by large numbers (almost running out of fingers and toes again), so the next kernel is going to be called 7.0.”
The new release adds support for AMD’s smart data cache injection, allowing I/O devices to place data directly into the L3 cache instead of going through RAM. It also supports Intel’s linear address-space separation (LASS), which creates a stronger boundary between kernel and user-space memory to help prevent speculative side-channel attacks.
For IBM’s s390 architecture, there’s now a new interface for hotplug memory configuration, and support for 31-bit binaries has been dropped. The s390 platform also gets stack-protector support thanks to updates in the upcoming GCC 16 compiler. For 64-bit Arm systems, support for Arm Memory System Resource Partitioning and Monitoring (MPAM) was added.
The core kernel now has a new listns() system call, which lets user space list kernel namespaces more efficiently. Updates to namespace reference counting stop user processes from bringing back namespaces that are being removed. Signal handling has also improved, so a process with a pidfd can now find out which signal caused another process to end with a core dump. Plus, BPF also got new features, like support for indirect jumps using a special map type on x86 systems.
There are several updates to filesystems and block I/O. The FUSE subsystem now has better buffered read support with large folios, and the iomap layer can track partially updated folios to make reads more efficient. The virtual filesystem also added support for recallable directory delegations, which helps with NFS directory delegation.
Moreover, a new file dynptr feature lets BPF programs read structured file data. The Btrfs filesystem now has a shutdown state to finish current operations while blocking new ones, and ext4 can now handle filesystems with block sizes bigger than the system page size.
Hardware support in 6.19 now covers more system timers, memory controllers, and network adapters. New drivers were added for Realtek system timers, Intel memory/IO hub controllers, and several Ethernet and wireless adapters.
Networking updates in this release focus on better performance and flexibility. A major change to TCP transmit locking has led to much higher throughput under heavy loads. Network sockets can now be marked as exempt from system-wide memory limits, with limits enforced inside containers instead.
On the security side, kernel 6.19 includes new SHA-3 and BLAKE2b hash algorithms in the kernel’s cryptographic library, plus related documentation. Security modules are now alerted when a memfd is created, so they can make policy decisions about these files. SELinux now supports this feature.
On top of that, the kernel now manages transparent huge pages for device-private memory and improves zram performance with writeback batching. The live update orchestrator is also included, so the kernel can be replaced on running systems without any downtime.
For virtualization and containers, the guest_memfd() interface now supports NUMA policies to control where memory is allocated in virtual environments. Confidential computing features have expanded, with PCIe link encryption and device authentication now supported, enabling encrypted, authenticated PCIe device communication. The Hyper-V confidential VMBus mechanism was also added for secure communication between guests and devices.
Finally, a new console font, Terminus 10×18, was added to make text easier to read on mid-resolution screens.
If you want to compile Linux kernel 6.19 yourself, you can now download it from kernel.org. As always, rolling-release distro users will get the update first, with the new kernel expected in their repositories in the coming weeks.
