Linux kernel developers shipped another substantial batch of Rust code for the upcoming 7.2 release. Miguel Ojeda mailed in the updates days ago. The additions top forty thousand lines. They focus on safety improvements and new driver capabilities.
At the center sits the zerocopy library. It cuts down on “unsafe” Rust blocks that handle byte-to-structure conversions. Those blocks once invited memory errors. Now the kernel can manage them with verified abstractions. The change matters. Kernel code lives at the sharp edge where one wrong pointer can crash a server or expose data.
Zerocopy and Safety Gains Reshape Rust Integration
But safety forms only part of the story. The same pull brings AutoFDO support for better profile-guided optimization in Rust components. KASAN integration arrives too. It lets developers hunt memory bugs in Rust code with the same tools long used for C. These additions didn’t appear overnight. They build on years of groundwork that began when Linus Torvalds first merged Rust support.
By Linux 7.0 the experiment label came off. Miguel Ojeda posted the patch that made it official. “The experiment is done, i.e. Rust is here to stay,” he wrote, according to Phoronix. Maintainers accepted the trade-offs. Corporate backers now face pressure to train more Rust-proficient kernel contributors. The message was clear. Rust sits alongside C as a first-class citizen.
Greg Kroah-Hartman has pushed the case harder than most. At Rust Week 2026 in Utrecht he told attendees Rust delivers “more fun for maintainers” and “more secure Linux for users.” He ended his talk with a direct call for more developers to join the effort. The Linux kernel issues around 13 CVEs per day in some periods. Many trace back to classic C memory mistakes. Rust stops entire classes of those bugs at compile time. That fact alone keeps the momentum alive.
Graphics drivers show the most visible progress. Danilo Krummrich sent the main DRM Rust subsystem updates aimed at 7.2. NVIDIA’s Nova driver accounts for the bulk. This open-source effort succeeds Nouveau as the modern path for NVIDIA hardware in the kernel. It already supports Hopper and Blackwell GPUs with identification, FSP boot paths, EMEM operations, and MCTP infrastructure. GA100 cards gain IFR header detection, firmware signature selection, and FRTS boot handling.
GSP boot and unload code saw refactoring. VBIOS parsing turned stricter with checked arithmetic and bounds-checked accesses. The driver now uses Higher-Ranked Lifetime Types, or HRT, for better lifetime handling in device abstractions. PCI BARs, sysmem flush logic, and sequencer code all adopted these patterns. Arm’s Tyr driver picked up smaller fixes. Apple Silicon support remains absent for now. The full set of changes appears in Krummrich’s pull request to the DRM list.
These aren’t isolated experiments. Android already ships Rust code in production. Devices running the 6.12 kernel use a Rust implementation of the ashmem allocator. Millions of phones rely on it daily. That real-world mileage reduces fears about stability. It also proves the bindings and abstractions hold up under load.
Yet challenges remain. Some longtime maintainers still prefer pure C for their subsystems. Torvalds has signaled he would override objections if needed. The community split shows in mailing list threads and conference talks. One side sees memory safety as non-negotiable for future kernels. The other worries about added complexity and the learning curve for veteran contributors.
Numbers tell part of the tale. At the start of 2025 the kernel held roughly 34 million lines of C against 25,000 in Rust. The gap stays large. Each merge window narrows it. Linux 7.2 alone adds tens of thousands more Rust lines. The trend points one direction. New drivers and subsystems increasingly arrive in Rust first. Existing code stays in C unless someone invests in a rewrite.
Tooling improvements help. The kernel bumped its minimum Rust version in recent cycles to match Debian stable. Preparations for Rust 1.95 landed earlier. These steps reduce friction for distributors and developers. They also let the Rust for Linux team drop workarounds and adopt newer language features.
Performance questions linger. Early Rust drivers sometimes trailed hand-tuned C equivalents. Profile-guided optimization and better abstractions close that gap. Zerocopy in particular removes unnecessary copies and unsafe conversions that once hurt speed and safety alike. Early tests shared on Phoronix forums suggest the overhead shrinks with each iteration.
So the kernel moves forward on two tracks. C veterans continue their work. Rust advocates expand into new territory. The dual-language reality demands careful coordination. Bindings must stay current. Documentation must cover both worlds. Reviewers need familiarity with two languages and their idioms.
Greg Kroah-Hartman sees the upside. He argues new code should default to Rust when practical. That stance draws criticism from those who view it as forced change. The debate won’t vanish soon. But the code keeps landing. Linux 7.2 will ship with more Rust than any prior release. NVIDIA’s Nova effort stands as the clearest proof of concept yet.
Watch the merge window. It opens mid-June after 7.1 stabilizes. Expect final tweaks and perhaps a few more Rust features before the release candidate cycle begins. The foundation grows stronger with each step. Safety improves. New hardware gains support. And the kernel inches closer to a future where memory bugs become rare exceptions rather than weekly headlines.
Recent coverage confirms the pace. Phoronix reported on the DRM Rust pull earlier this month, detailing Nova’s rapid expansion. The same outlet covered Kroah-Hartman’s Rust Week remarks in May. Those accounts match the patches now queued for 7.2. No slowdown appears on the horizon.
WebProNews is an iEntry Publication