Linux kernel 6.11 lands with vintage TV support

io_uring is getting more capable, and PREEMPT_RT is going mainstream

Open Source Summit Europe Released remotely from Vienna, Linux kernel 6.11 is here, with improved monochrome TV support. Yes, in 2024.

Emperor penguin Linus Torvalds was attending the Open Source Summit in the Austrian capital, but that didn't stop him working-from-hotel and emitting the latest and greatest Linux kernel. Along with the usual raft of improved device drivers and hardware support, which constitutes the bulk of the code in the kernel, there are some more unusual changes in there.

Kernel 6.10 got its own version of Windows's famous blue screen of death, which is semi-graphical and called DRM Panic. You can find a decent explanation on the project page for including it in Fedora 42. (Note, this is entirely separate from the BSOD service introduced in systemd 255, but we can't help but feel that it is very much The Linux Way for two independent subsystems that do much the same thing to come along more or less at once.)

The DRM Panic handler is part of the kernel's Direct Rendering Manager, and it can display a message even if text consoles have been disabled. Now in kernel 6.11, the display can be graphical, enabling snazzy new features such as showing what went wrong in the form of QR codes.

As Phoronix describes, thanks to Raspberry Pi developers, in 6.11 DRM support extends to displaying a black-and-white TV-compatible signal. Pi hardware is often deployed in digital signage, CCTV systems, and so on, and it's entirely possible that no better display may be available. Now kernel 6.11 can output over old-fashioned low-rez monochrome video, as shown via a composite video output. We hope you never see it, but if you do, take consolation that your open source black screen of death can now be grainier and less colorful than ever.

Elsewhere in 6.11, the io_uring kernel call interface can now be used with two new types of syscall: the bind() and listen() API calls for socket-handling. io_uring appeared in kernel 5.1 back in 2019, and The Register discussed it a few years later. For us, the snag with most explanations of how it works – even the better ones – is that they're aimed at programmers, and aren't very readable if you're not.

As an analogy, io_uring does for some Linux system calls what bus-mastering DMA did for EIDE hard disks in the Windows NT era. Disk DMA let NT's multitasking kernel get on with other stuff while the sluggish hard disk loaded data into memory. This was little help with the DOS-based Windows 9x, but on NT it made EIDE disks nearly as fast as much more expensive SCSI drives. You could actually hear the difference.

io_uring allows Linux programs to stack up certain system calls in what the kernel developers call a circular buffer. (It's not really circular – it's just a linear block of memory, but it's effectively endless: when accesses reach the end, they loop back to the start.) Then the program can do something else while waiting for the kernel to return the data.

It's unique to Linux, so using it makes programs much harder to port to other Unix-like OSes such as the BSDs, but even so, it's a potential game-changer, and the more syscalls that can be handled this way, the better.

There are lots of more specialized changes in kernel 6.11 as well: improved support for AMD and Qualcomm processors, hot-pluggable Arm processors, drivers for Intel Arrow Lake and Lunar Lake chips, better virtualization on Loongson, a newer version of Rust, and so on.

This release is planned to be a short term one. The next long term kernel is expected to be 6.12, probably around November. Already merged into the work-in-progress 6.12 is the Intel-sponsored PREEMPT_RT patch. Some vendors already offer distros with the real-time patch enabled, notably Canonical with Real-time Ubuntu. Soon this will become a lot more mainstream, and Linux should start to become a lot more responsive. ®