Oracle is the #1 contributor to the Linux kernel in 6.1

January 19, 2023 | 3 minute read
Text Size 100%:

With the release of 6.1, Oracle is the top contributor to the Linux kernel, responsible for the most lines of code changed in the kernel.

In July 2022, we published a blog about how Oracle was the #1 contributor to the core of the kernel in 5.18. That blog had an asterisk, since the statistics for those contributions were limited to the core parts of the kernel.* With the release of 6.1, the asterisk is not needed, as Oracle is the #1 contributor for total lines of code changed across the kernel.

The major addition was the official inclusion of the novel Maple Tree Data Structure. The Maple Tree is a lockless B-Tree data structure that promises to significantly reduce contention around concurrent access for memory management on small and large systems. The Maple Tree has been under development for years (the first RFC went out in 2020) and is finally included in the Linux kernel! It's so exciting to see Liam Howlett and Matthew Wilcox's work coming to fruition. LWN Writeup on the Maple Tree.

Our last "#1" statistic was for the most changes that affect the parts of Linux that everyone uses (loosely defined as fs/mm/kernel), much of this submission came in the form of regression and validation tests for the data structure that are key to ensuring it can be a reliable data structure for critical kernel data structures (but comprise code that most users will never interact with).

The great thing is, we're nowhere near done in terms of our contributions and improvements. I'm really proud of how dedicated this team is to improving the Linux kernel and ensuring that Linux continues to solve our customer's problems.

In the next kernel releases, you'll continue to see more patches bringing memory folios to more parts of the kernel, patches to shrink struct page (maybe?), patches to support RPC-over-TLS for transparently encrypted NFS, and patches to enable more parts of the kernel to start using the Maple Tree.

I'm particularly excited to see the multiple year work for XFS Online Repair coming to fruition. This is one of the biggest and most impactful changes we've ever done for XFS, and enables users to validate for XFS corruptions without unmounting their filesystems and -- even more amazingly -- to repair corruptions online. This patchset is now more than 400 patches, and has been in development since at least 2018 -- going on five years.

The most amazing thing about the XFS Online Repair patchset that's now under review is this: even in its prototype, under review state, online repair is *more fully featured than offline repair*. That's right, the online repair code can fix more things than the offline repair code. As Darrick writes, "Fuzz and stress testing of online repairs have been running well for a year now. As of this writing, online repair can fix slightly more things than offline repair, and the fsstress+repair long soak test has passed 100 million repairs with zero problems observed." This work will take several rounds before it's ready for linux-next, and even longer before it's available and packaged in Oracle Linux, but this enhancement promises to make boot-time filesystem checks a relic of the past!

You can find pointers to all this work and more on the Linux Kernel RFCs, and in our Linux kernel blog posts at blogs.oracle.com/linuxkernel. LWN is a great resource for the Linux community. Consider subscribing to support their work!

* Oracle continued to be #1 for "Core" contributions in 5.18 and 5.19. In 6.1, Oracle is #2 for "Core" contributions, behind Google.

Greg Marsden


Previous Post

Linux SLUB Allocator Internals and Debugging - KFENCE, Part 4 of 4

Imran Khan | 9 min read

Next Post


Cloud native DevOps with Oracle Linux and VirtualBox: It's all about increasing business velocity

Gursewak Sokhi | 5 min read