Bonum Certa Men Certa

A Code of Conduct Can Lead to Deterioration of Quality Control in Linux (Nobody Reprimanded for Technical Issues, Instead Critics at Times of Crisis Get Reprimanded)

posted by Roy Schestowitz on Dec 11, 2023,
updated Dec 11, 2023

Richard Harvey, ODESSA TOWNSHIP, Mich.

THE pertinent details about the latest Linux issue are not fully disclosed just yet, but it has been assigned a rare level of severity. We need to discuss this as Linux grew exponentially larger and now contains multiple programming languages, so many veteran Linux coders cannot even comprehend what certain parts of Linux do. Especially parts that have not matured. Their freedom of speech (expression, opinion, criticism) is curtailed further, as the culture of Rust is imported to the core of the system (criticising Microsoft is like "hate speech").

Let's be clear upfront.

This is an ongoing and still-developing story about a critical issue impacting Linux and thus many GNU/Linux distros. Debian, which many distros are derived from, is also affected. We've thus rebooted the server (IRC downtimes are unpleasant; IRC has no redundancy, as it's not spread across nodes) twice in the past day, loading the latest kernel and checking a whole bunch of stuff.

In this short post we will write primarily about what's known (this is still discussed a lot in our chats, with the "knowns" separated from the "unknowns"). The media will soon follow up (not just some blogs), so we too can follow up in the future with further details. As a C programmer myself, I can make sense of some of the code and defects. The media won't even cover those aspects as nowadays "tech" gets covered by people who just parrot buzzwords like "AI", not proficient programmers.

Having studied the Debian repositories, mailing lists (and LWN comments about those) et cetera, it seems safe to say that the issue isn't just a real issue (unlike mere hype, one might say dramatisation) and some people are already impacted. One associate had started to see accounts of trashed systems before rushing to update ours.

How did it all happen, whose fault, and what could be done to prevent it? We'll probably have more definitive answers in days to come.

There are almost no details about the problem available, as an associate explains. Debian apparently pushed out the tainted kernel many days after the problem was known and since Linus Torvalds is not in charge of the kernel anymore (he has just said he's not a manager), we cannot see him saying a word about it. There's no word from Linus about this as of today; I checked LKML and messages from Linus going as far back as weeks ago.

When did he find out about this and what did he know? "I AM SORRY" won't be enough in this case; many critical operations around the world may have silently lost some data; some won't be able to assure system integrity or even boot the system. It impacts both servers and desktops/laptops. Devices (like portable phones)? Maybe... if they use ext4.

Just about a month ago Linus complained about bad code making it into the kernel at the last minute: "If people then stop testing new kernels because they think new kernels might break their setup, we have lost something truly important."

Now we have a severe data issue. It cannot be undone, even if a fixed kernel gets installed. People are encouraged to check their data's integrity against backups, if any are available (RAID is no panacea here).

"Perhaps he knew and could not speak for fear of stepping on some corporation's CoC," an associate hypothesises. "Even the package descriptions and info were unclear as to which actual version was being deployed."

Some years ago we saw bad kernels (seldom a stable kernel, usually an RC) breaking some hardware, e.g. putting monitors or network cards at risk, but those can be replaced. Data cannot be. It's a truly complicated issue.

Confidence in Linux is the main casualty.

"The sick part is that it appears 1) Debian backported the broken patch 2) Debian deployed said broken patch even after it was known to be broken 3) they left the package up in the repository for days before pulling it," an associate alleges.

There too many uncertainties still, but getting the Git history helps check what happened and when exactly. "The fact that it was in Bullseye points to (1), (2) would be a matter of looking in the Git repository, and [regarding] (3) we installed it from the repository."

The tracking of the package can be done via this so-called "webapp" or the tracker (also security tracker), which has changes in the Git file.

Why was the public not informed until a weekend? This isn't like a security issue that can be exploited by hostile parties if revealed "too early"... or it is? Could this issue be triggered remotely with some network-bound request, taking advantage of some particular weaknesses in ext4?

Tracing back the communications about this, (timeline in the message bodies at [1, 2, 3]), the key powwow can be traced back to 6 days ago, the 5th of December, but discussion is going back to late November. We know that linux-image-6.1.0-14-amd64 was removed during the weekend from the package repository of Debian, but where was an official statement on the matter? There's still no official communication about it.

Torvalds used to be outspoken when bad code and bad practices were adopted and spread throughout the team. Now he says absolutely nothing and there's no sign that negligent/reckless acts will bear any consequences; instead, Jim Zemlin will blame the "opinionated" community - a term that he does not even understand and represent something he is hostile towards.

Quality control demands opinionated people, even blunt opinions at times.

In his latest "talk" (chat with Dirk) Linus admitted he no longer writes real code (just pseudocode portions) and that he just does a smell test of code he allows into his tree. He also admits he cannot understand Rust (but is learning a little as he goes along), so he essentially permits random people to toss large lumps of code without a vendor-neutral (independent from the vendor of origin) audit - i.e. an additional problem.

What will be done to ensure the above does not happen again? Rewrite the file systems in Rust? That would make things worse and less predictable.

Other Recent Techrights' Posts

If Matthias Kirschner Loves Free Software, He'll Change the Name of the Microsoft-Sponsored Organisation He Governs (in Order to Avoid Confusion)
The FSF-EEE does not really like Software Freedom, it just loves money (including Microsoft's)
EPO's Local Occupational Health, Safety and Ergonomics Committee (LOHSEC) in The Hague: Workers Are Getting Sicker, Conditions in Which to Assess Patent Applications Deteriorate
"According to the Office statistics the total number of days of absence has gone up from 12.4 to 13.1 total number of sick days per Full Time Equivalent (FTE) from 2023 to 2024."
The Standard Needs to Improve Its Standards for Fact-Checking, Aaron Swartz Had Nothing to Do With Reddit and He Detested the Company That Created It
The Web is already bad enough as it is
New Paper From the EPO Highlights Large-Scale Discrimination at the Office, Where People Are Rewarded for Granting More and More Illegal Patents
Even the Kremlin is probably more competent than this
 
Gemini Links 11/02/2025: NeoVim and Deploying Other People's Code
Links for the day
BetaNews is Still Publishing LLM Slop/SPAM About "Linux"
Assuming it is indeed LLM slop, it seems clear BetaNews has no intention of improving or is simply unable/unwilling to improve
IBM Layoff Rumours, Large-Scale Implementations Weeks Ahead (in March 2025)
There are some people corroborating
Over at Tux Machines...
GNU/Linux news for the past day
IRC Proceedings: Monday, February 10, 2025
IRC logs for Monday, February 10, 2025
Scheduled Maintenance Tomorrow and on Valentines
If the site (or Gemini capsule) is offline for a bit, the maintenance windows are likely the root cause
Soylent News Lessens the Scope of Discussion Due to Persistent Trolling and Online Abuse
if they make it a lot harder for new people to participate, then they limit the "general appeal" and reach
When the Livestream of Richard Stallman is Apparently Bury-Brigaded Offline You Finally Learn to Avoid Google/YouTube for Streaming
Please, people, stop uploading to Google/YouTube
Links 10/02/2025: Ban on D.E.I. Language, Listeria Risk/Outbreak
Links for the day
Links 10/02/2025: Announcing "Stringless" and Mental Health Improvement
Links for the day
Links 10/02/2025: Facebook Mass Layoffs, "Meta" Did What Aaron Swartz Had Done But to the Tune of 81.7 Terabytes
Links for the day
Microsoft Tarnishing the Brand of Arch
Of course Arch can do whatever it wants, but being associated with Microsoft is a badge of shame
The Ultimate and Inevitable Fall of OpenAI (Even Brave is 'Bigger' Now)
"When you advertise at the Super Bowl, you’ve reached just about every consumer in America. It’s the last stop. If you’re not profitable yet, you never will be."
Adding Slop to Your Blog Only Makes One Assume All the Text is LLM Slop
Simon Coter from Oracle has turned to slop
Macao is Leaving Microsoft Behind
Windows is falling to new all-time lows
Over at Tux Machines...
GNU/Linux news for the past day
IRC Proceedings: Sunday, February 09, 2025
IRC logs for Sunday, February 09, 2025
Microsoft's WSL (LSW) Shows That It Can Never Love Linux, Only Windows
that's just how Microsoft rolls
Activism in Times of War and a Coup
'Linux' Foundation works for fascism
What the Silencing of Neatnik Tells Us About Linus Torvalds Inside a Microsoft-Dominated 'Linux' Foundation
Is Linus Torvalds free to express his mind as he wishes about every topic, even just any technical topic?
Windows Down to 11.35% in Senegal, as Measured by statCounter
Another all-time low (Windows was at 99% in 2009)
"Latest Technology News" in BetaNews is LLM Slop Promoting OOXML and Proprietary Software at the Expense of LibreOffice and OpenDocument Format (ODF)
Remember that "open-source" and Open Source aren't the same; the former is fake
Links 09/02/2025: Coffee, Toxic Productivity, and Programming
Links for the day
Debian's Human Rights violations & Swiss women Nazi symbolism
Reprinted with permission from Daniel Pocock
Links 09/02/2025: Software Patents on MP3 and Another Scam Dressed Up as "Crypto"
Links for the day
Links 09/02/2025: Russian Energy Cut Off, LLM Pushers Show Signs of Desperation
Links for the day
Richard Stallman (RMS) Does Not Have Media Companies and Lobbyists on His Side, But His Message Spreads Regardless
The message of RMS is spreading in spite of all the smears
GNU/Linux Rises to All-Time High in Chile
sharp rise for GNU/Linux in Chile
Links 09/02/2025: Hottest January on Record, Panama Blackmailed
Links for the day
Why We Still Love Gemini Protocol
Gemini Protocol may seem like something "old" (it's actually very new) and something "nobody would use", but many people use it
Gemini Links 09/02/2025: "Died as a Mineral" and Game Interface for a Non-Game
Links for the day
Over at Tux Machines...
GNU/Linux news for the past day
IRC Proceedings: Saturday, February 08, 2025
IRC logs for Saturday, February 08, 2025