Linux 7.0 Released: What's New in the Kernel?

Added: 2026-05-14

[00:00:00]Linus Torvalds has announced the latest release of the Linux kernel with Linux 7.0.

[00:00:05]This is a very big release with a lot of improvements for file systems, hardware support, networking, and more. Now, when it comes to version numbers, there are many schemes for projects, and one of the most popular is called major minor bug, and that means the first number is reserved for big changes like the seven in 7.0 or major releases. And the next number is for minor releases where something small is fixed or improved.

[00:00:31]And then finally, the last set of numbers is for bug fixes and patches. In the case of the Linux kernel, none of that matters. The number chosen is completely arbitrary and means nothing.

[00:00:42]It's more of an incremental chronological system to just keep track of the progress. For a quick history lesson, the Linux kernel used to have a version scheme that had meanings to the order of the numbers, but then the 2.x series happened, and more specifically, the 2.6.x series happened, and then it was changed. Linux 2.0 was released in June of 1996, and then it wasn't until 2003 we got 2.6. The last update to that version was in 2011 with 2.6.39.

[00:01:15]Yes, that means 2.x series lasted 15 years.

[00:01:20]This is when Linus Torvalds decided that was enough of the versioning nonsense and changed the system to be incremented every 20 releases. This meant that 3.x, 5.x, and 6.x series all had 20 releases starting with zero and ending at 19.

[00:01:35]The only exception to this was 4.x, which had 21 releases, and while not completely confirmed, I like to think it was supposed so they could do the joke about 420.

[00:01:45]All right. So, now that we've got that clear, 7.0 doesn't necessarily mean that it's a major release, but it is in fact a major release. So, let's jump in. First up in this release is there's a new API for file IO error reporting. File systems on Linux currently have no standard mechanism for reporting metadata corruption and file IO errors to user space via FS notify.

[00:02:08]Each file system deals with the problem differently and error reporting is inconsistent or absent entirely. This release introduces a generic FS error infrastructure that gives file systems a standard way to queue metadata and file IO error reports for delivery to FS notify. Linux 7.0 also adds a new health monitoring feature for XFS. This release adds new functionality to deliver live information about file system health events to user space. This is done by creating an anonymous file that can be read for events by user space programs.

[00:02:45]Events like metadata health failures, file IO errors, and major changes in file system state like unmounting and shutdowns and stuff like that. In user space, a new daemon program is created that will read the event objects and initiate repairs automatically. This daemon is managed entirely by systemd and will not block unmounting of the file system unless repairs are ongoing.

[00:03:08]This release adds support for clang language extensions, which enables statically checking that required context are active or inactive by acquiring and releasing user definable context locks. An obvious application is lock safety checking for the kernel's various synchronization primitives and checking that locking rules are not violated. This can help detect bugs at compilation time. There's also better IOU ring support for filters cuz due to its design, IOU ring doesn't play well with the system call filtering done by system calls like SECCOMP or SECCOMP.

[00:03:43]IOU ring supports filtering, but it's not really usable for these cases. This release adds support for filtering, which enables tighter control over what exactly a specific OP code may do. On top of that, support is added for per-task filters, meaning that any ring created with a task that has a per-task filter will get those filters applied when it's created. These filters are inherited across forks as well. Once a filter has been registered, any further added filters may only further restrict what operations are permitted. Linux 7.0 also extends Open Tree 2 to allow faster and simpler container setup. When creating containers, the setup usually involves using clone new NS via clone 3 or unshare, which copies the caller's uh complete mount namespace. After some tasks, the runtime will recursively umount the old mount tree, thereby uh getting rid of all the mounts. This can be wasteful and slow. This release extends Open Tree with a new Open Tree namespace flag, similar to Open Tree clone. Only the indicated mount tree is copied. Instead of returning a file descriptor referring to the mount tree, it will cause the Open Tree to return a file descriptor to a new mount namespace. In that new mount namespace, the copied mount tree has been mounted on top of a copy of the real root FS.

[00:05:03]This allows Open Tree namespace to function as a combined unshare uh well, clone new NS and a pivot root. This release also has better swapping performance with swap table as part of the long-term effort to improve the swapping performance which started in version 6.18. This release includes the code corresponding to the phase two of the swap table, which aims to clean up and speed up swapping code. The Linux kernel added experimental support for ACC ECN in 6.18 after some work done in previous releases. Now, this release of 7.0 finally enables it by default.

[00:05:38]Explicit congestion notification or ECN is a mechanism where devices can mark IP packets instead of dropping them to indicate congestion to the endpoints.

[00:05:48]But, this was originally specified in such a way that only one feedback signal can be transmitted per round trip time or RTT. ACC ECN or RFC 9768 is a scheme that provides more than one feedback signal per RTT in the TCP header, which can be used to provide better TCP congestion handling, which is a much better improvement for networking. This release of Linux 7.0 also adds support new tree in the Butter FS file system.

[00:06:17]This is a logical remapping tree, which will in the future act as a layer of indirection when doing IO. When doing relocation, rather than fixing up every tree, the file system will record the old and new addresses in the remap tree.

[00:06:31]This will make things more reliable and flexible, as well as enabling some future changes Butter FS devs would like to make, such as larger data extends and reducing write amplification by removing uh copy-on-write only metadata items.

[00:06:45]And last, but certainly not least, uh there's one thing I wanted to tell you about that's coming in the this latest version of Linux 7.0, and that is Well, for all of you Rock Band 4 guitar owners for the PS4 and PS5, those are now supported with Linux 7.0. There's also a whole lot more in this latest release of Linux 7.0. I didn't talk about all the networking stuff, there's even more, and there's there's just a ton. I'll have links in the show notes for anyone who'd like to learn more about the latest release and what is in the change log.