Home Linux KernelLinux Kernel 6.18: The Next LTS Release Arrives with Speed, Security, and Rust

Linux Kernel 6.18: The Next LTS Release Arrives with Speed, Security, and Rust

By sk
601 views 6 mins read

The Linux Kernel 6.18 release is here! This is a very promising update, packed with diverse improvements across every major subsystem. Linux Torvalds successfully tagged and pushed the final release on November 30, 2025. The kernel team incorporated a remarkable 11,974 non-merge commits into this version.

We anticipate Linux Kernel 6.18 will become the next long-term support (LTS) kernel.

Here is a look at the most significant highlights you should know about.

Linux Kernel 6.18 Highlights

Core Infrastructure and the Rust Revolution

The trend toward integrating the Rust language into the kernel reached a major milestone in 6.18.

First, the Rust Binder driver finally merged. Binder maintainers view this integration as a "true point of no return for Rust" in the kernel because Binder is important to Android and sees constant development.

The C implementation remains for now while the team verifies the Rust implementation matches the existing user-space interface precisely.

Furthermore, developers added several new Rust bindings:

  • Support for Linux-kernel-memory-model-compliant atomic operations.
  • New APIs for using maple trees and manipulating bitmaps.
  • Tools for creating DebugFS files.
  • APIs relevant to writing drivers.

The core kernel also saw major speed and usability improvements:

  • Introducing Sheaves: Developers merged work on "sheaves", which significantly improves the performance of the SLUB allocator. Sheaves are an opt-in per-CPU array-based caching layer. This mechanism makes allocation faster because it involves a preempt_disable() and no atomic operations in the fast path, avoiding the cost of local double compare-and-swap operations.
  • Low-Lock Allocation: A new function, kmalloc_nolock(), allows for memory allocation without taking additional locks.
  • File Handles for Namespaces: Kernel namespaces (such as network namespaces) now use file handles for reference, extending previous support for pidfds.
  • Configuration Management: A new "transitional" attribute simplifies renaming kernel configuration options. This makes sure users running make olddefconfig automatically migrate to the new name.
  • Toolchain Requirement: Building the kernel now requires LLVM 15.0.0 or newer.

Performance Boosts: Faster Networking and Memory

Networking and storage performance received some of the biggest speed gains in this release.

  • UDP Speed: You can expect a massive boost to network responsiveness; developers improved UDP receive performance by 47%. This measurement applies under high network load using 120-byte packets.
  • Larger Buffers: The kernel raised the default socket receive buffer size to 4MB.
  • AccECN Adoption: Developers merged the work for Accurate TCP Congestion Notification (AccECN). AccECN uses space within TCP headers instead of IP headers, a design feature suspected to make it less prone to interference from "middleboxes" that often break classic ECN implementations.
  • Buffer Sharing: The kernel also merged the direct internal buffer sharing (dibs) work.
  • Security: Encrypting TCP connections using PSP (Packet Security Protocol) is now possible.

Storage and Filesystem Evolution

The storage sector saw both a major removal and the addition of powerful new caching tools.

  • bcachefs Removal: Developers removed the bcachefs filesystem in its entirety.
  • Persistent Memory Caching (dm-pcache): A significant addition is the dm-pcache device-mapper target. This tool uses persistent memory (pmem) as a cache for slower block devices.
    • Dual Caching: Importantly, dm-pcache supports both read caching and write caching.
    • Speed Advantage: It uses a DAX data path, copying data directly between DRAM and the pmem mapping, thereby bypassing the block layer's overhead. This results in performance metrics far exceeding alternatives, achieving 2.1 M IOPS for 4K random writes compared to 352 K for bcache or 283 K for dm-writecache.
  • XFS Stability: The XFS filesystem now enables online fsck by default, as the team no longer considers it experimental.
  • Case-Folding: Both Overlayfs and f2fs now support case-folding. Overlayfs applies this setting on a per-layer basis.
  • umount() Speed: Calls to umount() should no longer provoke quadratic behavior when a mount contains many nested mounts.
  • Ext4 Updates: The ext4 filesystem gained 32-bit reserved user and group IDs and a new ioctl() interface to manage superblock parameters. They also fully removed some legacy ext3 configuration options.

Enhanced Security and Virtualization

Version 6.18 delivers key architectural and security upgrades, improving the safety and compatibility of virtualized environments.

  • KVM and CET: KVM now supports virtualizing Control-Flow Enforcement Technology (CET) on x86 chips. This includes shadow stacks and indirect branch tracking (IBT) on Intel, and shadow stacks on AMD. KVM enables both features if either one is requested.
  • SEV-SNP Protection: x86 hosts can enable SEV-SNP CipherText Hiding, preventing unauthorized CPU access from reading the private memory of secure nested paging (SNP) guests.
  • TDX Coexistence: Trust domain extensions (TDX) on x86 and kexec now work simultaneously. Previously, the TDX encryption hardware interfered with the new kernel's cachelines.
  • Multi-LSM Audit Support: The audit subsystem handles multiple Linux security modules (LSMs) enabled at the same time. This change promotes smoother use of advanced security setups.
  • BPF Signing: The kernel now supports signing BPF programs, although the specific security policies needed to fully leverage this are still under development.

Hardware Compatibility: Drivers Lead the Charge

Drivers accounted for about half of all the changes merged into 6.18. New hardware enablement included graphics, input, and architectural fixes:

  • New Graphics Drivers: Support arrived for numerous Mali GPUs (G710, G510, G310, Gx15, Gx20, and Gx25). Notably, a new Arm Mali CSF-based GPU driver written in Rust was merged.
  • Specialized Processors: The release added support for Rockchip NPU neural processors and T-HEAD TH1520 GPUs.
  • Input and Audio: HID drivers now handle haptic touchpads. The kernel also supports DualSense Playstation controller audio jacks.
  • RISC-V Expansion: RISC-V gained support for the IO Mapping Table (RIMT) via ACPI, which provides IOMMU information. It also supports the RPMI platform-management communication interface and the MPXY SBI firmware shared-memory communication system.
  • Arm Mitigations: Spectre mitigations for Arm Cortex-A720 CPUs were extended to also apply to Cortex-A720AE CPUs.
  • SPI for VMs: The new virtio SPI driver lets you provide SPI device access to virtual machines.

Install Linux Kernel 6.18

You probably want to install and start using these features immediately! Here is how you can get the new kernel:

1. Source Code:

You can download the 6.18 source code directly from the Kernel.org website or the Linus Torvalds's git tree. But, this option requires you to compile the source code manually.

2. Distribution Availability:

The upcoming Ubuntu 26.04 daily builds will ship the latest 6.18 Kernel in the coming weeks. So if you're testing Ubuntu 26.04, you can enjoy the benefits of 6.18.

3. Unofficial Tools:

If you use other Debian-based distributions, consider the Mainline utility. This free, open-source application simplifies the process, fetching, displaying, and installing Ubuntu Mainline Kernel builds.

Keep in mind that using unofficial installation methods might result in a kernel lacking Ubuntu-specific patches or necessary driver support. If you install unofficially, you receive no guarantees or support.

What's Next: The 6.19 Outlook

Since the 6.18 release process completed successfully, the development cycle for Linux 6.19 is now beginning. But, Linus noted that the next release will face schedule disruptions:

  • The 6.19 merge window overlaps with the yearly kernel maintainer summit and associated travel.
  • The later part of the 6.19 release cycle runs into the holiday season.

Linus Torvalds tentatively plans to drag out the 6.19 release by a week to compensate for time lost to the holidays, although this timing remains subject to change.

For more details, refer to the changelog in the Linux Kernel 6.18 official release notes, 6.18 merge window part 1 and part 2.

You May Also Like

Leave a Comment

* By using this form you agree with the storage and handling of your data by this website.

This site uses Akismet to reduce spam. Learn how your comment data is processed.

This website uses cookies to improve your experience. By using this site, we will assume that you're OK with it. Accept Read More