Linus Torvalds has officially announced the release of Linux kernel 6.19, marking another major milestone for the Linux ecosystem.
In his trademark style, Torvalds noted that the past week was relatively quiet, allowing the release to ship on schedule—just in time for a day dominated by televised commercials in the US.
Alongside the announcement, Torvalds confirmed that the next kernel version will be Linux 7.0, expected around mid-April. This change is purely a version number reset, not the start of a new development cycle. The reason is simple: the 6.x series has grown so large that continuing with higher minor numbers has become unwieldy.
Linux kernel 6.19 introduces several notable hardware and security improvements. AMD systems gain support for smart data cache injection, enabling I/O devices to write data directly into the L3 cache instead of main memory. Intel platforms benefit from Linear Address-Space Separation (LASS), which strengthens the boundary between kernel and user space to reduce the risk of speculative side-channel attacks.
On IBM’s s390 architecture, a new interface simplifies hotplug memory configuration, while support for legacy 31-bit binaries has been removed. The platform also gains stack-protector support via upcoming GCC 16 updates. For 64-bit Arm systems, support has been added for Memory System Resource Partitioning and Monitoring (MPAM).
At the core kernel level, a new listns() system call improves how user space can enumerate kernel namespaces, while updated reference counting prevents removed namespaces from being revived. Signal handling has also been enhanced, allowing pidfd-based processes to identify which signal caused a core dump. Additionally, BPF gains new capabilities, including support for indirect jumps on x86 platforms.
Filesystem and block I/O subsystems also see meaningful updates. FUSE now offers improved buffered reads using large folios, and the iomap layer can track partially modified folios to optimize read performance. The virtual filesystem adds support for recallable directory delegations, improving NFS behavior. Btrfs introduces a shutdown state for graceful operation completion, while ext4 now supports filesystems with block sizes larger than the system page size.
Hardware enablement continues to expand in kernel 6.19, with new and updated drivers for system timers, memory controllers, and networking hardware, including Realtek timers, Intel memory/IO hubs, and several Ethernet and wireless adapters.
Networking improvements focus heavily on performance. A redesign of TCP transmit locking significantly boosts throughput under heavy load. Sockets can now be excluded from system-wide memory limits, allowing memory enforcement to occur at the container level instead.
From a security standpoint, the kernel now includes SHA-3 and BLAKE2b algorithms in its cryptographic library. Security modules are notified when memfd objects are created, enabling policy-based decisions—functionality now supported by SELinux.
Other enhancements include better handling of transparent huge pages for device-private memory, improved zram writeback performance, and the inclusion of a live update orchestrator, allowing kernel updates without system downtime.
Virtualization and container workloads also benefit from improvements. The guest_memfd() interface now supports NUMA policies, while confidential computing gains support for PCIe link encryption and device authentication. Secure Hyper-V communication is further strengthened with the introduction of the confidential VMBus mechanism.
Rounding out the release is a new Terminus 10×18 console font, designed for improved readability on mid-resolution displays.
Linux kernel 6.19 is now available for download from kernel.org. Rolling-release distributions are expected to ship the update first, with broader availability arriving in the coming weeks.
Read more about Linux Kernel 6.19-rc4
