Linux Format July 2021

meet the team

Jonni Bidwell Do I still get a chuckle if I say Windows? Sigh. Well anyway, our friendly contact at UKFast agreed that our server needed putting out of its misery. We doff our moth-eaten caps as we remember its legacy. But we (well, I) quite like the new website that its successor is powering. Christian Cawley Keyboard-dependent text entry is in dire need of replacement. With AR and VR solutions for almost everything else, the lack of a post-QWERTY input method for commands and text, over 50 years after the Mother of all Demos is baffling. (This has nothing to do with my typing ability!) Nick Peers I’m fed up with disposable devices with built-in obsolescence, like phones that can’t easily be repaired or upgraded. Tiny manufacturer Fairphone has shown you can support a phone…

we need upgrades!

Technology never stops advancing – that’s a blessing and a curse. Look at Ubuntu 21.04 and the seemingly never-ending story of the Wayland display protocol. It’s been in development for over a decade (see our first feature on it in 2011’s LXF141 by a certain Alexander Tolstoy), yet is still only just making it into Ubuntu, after an earlier failed attempt with Ubuntu 17.10. You can point at a lot of things – Nvidia, screencasting, program incompatibilities – but ultimately it’s a user experience thing. So four years on and the planets are finally aligning for Wayland. Even that immovable mountain of Nvidia driver support appears to be falling into place. Fedora has gone full Wayland and now Canonical thinks that it’s the right time, too. This is a good thing, even…

university banned for ‘hypocrite commits’

The University of Minnesota has been banned from contributing to the Linux kernel after two researchers from the institution were discovered deliberately submitting patches with flaws. Qiushi Wu and Kangjie Lu’s paper (http://bit.ly/LXF277Paper) reveals how they submitted patches to the kernel to fix issues, but in fact introduced problems – which the researchers called ‘hypocrite commits’. They claimed this would show how malicious users could slip dangerous code into the kernel, in a similar way to how white hat hackers identify security holes in software. Of course, white hat hackers are invited to find security holes by the people behind the software, and are often given rewards for doing so. Wu and Lu neither asked nor told anyone in the community what they were doing. Their actions have led to accusations that…

google’s floc gets frosty reception

Cookies – those little digital breadcrumb trails that track and identify you on the internet – are becoming ever less popular, and while that’s good news on the whole, Google is looking to replace them with FLoC (Federated Learning of Cohorts) in its dominant Chrome web browser. The company states (http://bit.ly/LXF277FLoC) that this will enable advertisers to target their ads to people, while the “approach effectively hides individuals ‘in the crowd’ and uses on-device processing to keep a person’s web history private on the browser. “ Google claims to have run simulations that have given the company the confidence to call FLoC a “privacy-first alternative to third-party cookies” which doesn’t tie browsing activity to individuals, but to groups of ‘like-minded’ people instead. Not only that, Google also claims that “advertisers can…

linux kernel hits one million git commits

The Linux kernel has hit a major milestone, with over one million git commits and climbing – even with the removal of any that came from the University of Minnesota. You can view the number of commits on Github at https://github.com/torvalds/linux. At the time of writing, the number was 1,012,913, and it goes to show how successful the Linux kernel is – and how regularly updated it is by volunteers. The number is likely even higher considering the number of commits also included in BitKeeper, the previous repo for the kernel. With the Linux kernel turning 30 soon (look out for a big celebration in LFX280) the hard work and dedication or Linus Torvalds and the thousands of volunteers is plain to see. Who would have thought, on 5 October 1991,…

filesystem errors

“One fact of life is that file systems fail. Detecting errors over a fleet of servers is tricky since there’s no generic way to monitor different filesystems in the Linux kernel without peeking at dmesg and filtering through thousands of messages of each machine in the fleet. This approach is cumbersome and prone to missing messages in high-traffic situations. Nevertheless, I’ve been on a quest to find the right API to report file system errors in a generic way, while still providing detailed-enough information that can be used by online monitoring recovery tools. The current approach is a push interface based on fanotify, in which monitoring programs register themselves as recipients of error messages from a specific file system and are notified when an error event takes place. Each message carries a generic…