Qualcomm, yesterday, promising Linux support for the new Snapdragon X2 processors:

Linux on Snapdragon X2 Series is moving from early bring-up toward a more complete upstream developer experience. Core support is landing, Hexagon NPU and Adreno GPU work is progressing, and real laptops with Snapdragon X2 Series processors are already beginning to boot Linux.

↫ Qualcomm’s promises from 2026

Interesting, but I feel like I’ve heard these exact words before. Qualcomm, two years ago, promising Linux support for the then-new Snapdragon X processors:

It’s been our priority not only to support Linux on our premium-tier SoCs, but to support it pronto. In fact, within one or two days of publicly announcing each generation of Snapdragon 8, we’ve posted the initial patchset for Linux kernel support. Snapdragon X Elite was no exception: we announced on October 23 of last year and posted the patchset the next day. That was the result of a lot of pre-announcement work to get everything up and running on Linux and Debian.

↫ Qualcomm’s empty promises from 2024

These promises were not at all kept. Two years later, Linux support for Snapdragon X laptops is still spotty, broken, and limited, confined to just a few bespoke Ubuntu builds, which don’t even offer full support either. It’s a complete mess, effectively unusable, and highlights once again that big technology companies are compulsive liars. I have little faith in these new promises, but who knows – maybe this time it’ll be different.

Probably not, though.

Want to run something a little more exotic on your server? How about GNU Guix?

It has been a month since migrating my home server to GNU Guix. Managing OS state declaratively through Git has eliminated configuration drift, and Guile Scheme provides a cohesive environment that complements Emacs. While adapting to a smaller package ecosystem and managing substitute timing requires occasional adjustments, the stability, reproducibility, and container isolation make it a dependable foundation.

↫ Wai Hon

I’m definitely noticing an increase in interest in Guix lately.

Earlier this year, Natalie Vock made a splash with a set of patches to the Linux kernel that greatly increased performance on AMD GPUs with lower amounts of VRAM. With that work now accepted by upstream, Vock decided to turn their attention to another interesting problem: what if you run out of VRAM, and how can we improve performance when we do?

Regardless, what I hope this blogpost can demonstrate is that even if you end up with some memory evicted to system RAM, the slowdown can be manageable. There’s measures that drivers (particularly, the kernel driver) can take to make overcommit work as fast as possible, and even applications can do their part in coordinating with the driver stack to mitigate the effects of their memory being evicted. With everything in place, VRAM overcommit isn’t really as big of a deal as one may think it is at first sight.

↫ Natalie Vock

The work Vock has done has already been in SteamOS for a while, and they’re currently in the process of upstreaming it to the vanilla kernel as well. Since this is a complex set of patches and changes, this may take a while, and as such, they’ve prepared custom kernel and Mesa branches for adventurous users. Do note that these branches won’t be maintained much, and are entirely experimental, not as well-tested as the SteamOS kernel, and probably won’t yield the same performance improvements.

Still, this is the kind of work that has a material impact for users. Not everyone has a 16GB monster GPU, especially not today with supply chains ravaged and ruined by slopmakers, so it’s great to see the Linux world working to improve performance for everyone, not just the wealthy few.

Version 7.2 of the Linux kernel has been released.

Significant features in this release include common attributes support in the bpf()vsystem call, cache-aware load balancing for the CPU scheduler, large-folio support in the Btrfs filesystem, further swap subsystem improvements, improvements to the Landlock security module, support for block devices with inline encryption hardware via the dm-inlinecrypt device-mapper target, and much more.

↫ corbet at LWN.net

If you run Linux, you’ll get it sooner or later.

Linus Torvalds, the creator of the Linux kernel and git, is employed by the Linux Foundation. This Foundation is a non-profit organisation dedicated to, as the name obviously implies, the promotion of Linux. The primary use of the funds it collects is to “help fund the infrastructure and fellows, including Linus Torvalds, who help develop the Linux kernel”. The list of megacorporations donating most of the Foundation’s funds is long.

The Linux Foundation has twelve platinum members, which donate $500000 per year, followed by twelve gold members, who donate $100000 per year. Below these two primary tiers lie the silver peasants, who each donate $5000-$25000 per year, based on number of employees. Looking at the list of twelve platinum members, I noticed something interesting.

Of the twelve platinum companies, six are “AI” companies or companies with massive investments in “AI”: Google, Huawei, Facebook, Microsoft, Oracle, and IBM/Red Hat. Then there’s Samsung Electronics, which is raking in stupendous amounts of money thanks to the “AI” bubble. Additionally, one of the gold members is Anthropic, another major “AI” company and makers of “Claude”, the sloppiest of slopcoding tools.

Many of these companies are unimaginably deep in the red when it comes to “AI”, with very little indication they’re ever going to be able to recover any of it. The situation is particularly bad for Oracle and IBM/Red Hat. Oracle’s debt has been downgraded to one notch above junk status because of its “AI” spending, while IBM’s shares experienced the largest crash in its 115 year history only a few days ago. By the way, in the first half of 2025, “AI-related capital expenditures contributed 1.1% to [US] GDP growth, outpacing the U.S. consumer as an engine of expansion”.

Fun fact: since most of The Netherlands is effectively a swamp, most of the country’s buildings are built on massive wooden or concrete poles (piles) hammered deep into the ground until they hit something more stable than mushy clay and wet sand. Otherwise, buildings in the country would simply sink into the ground. Every Dutch person who ever lived near a construction site has heard the rhythmic kathunk, kathunk, kathunk, all day long, as the massive piledriver machines spread their gospel. I guess something reminded me of this just now.

Anyway, a large chunk of the funding the Linux Foundation, Linus Torvald’s employer, receives is coming from increasingly desperate companies frantically trying to convince a populace deeply skeptical and often downright hostile towards “AI” to spend money on “AI” before the bubble bursts.

For some reason, I thought this was interesting.