News

The stories I could tell about how companies will hire a team to run tests on their digital and physical systems while also limiting access to outside nodes disconnected or screened from their core, primary, IMPORTANT systems.

Kicker is that plenty of people who work for these companies get it. Very rarely does someone in a position to do something about it actually understand. A few thousand dollars and they could have hired a hat or two to run penetration on systems and fixed the vulnerabilities, or at least shored them up so this fucking 000 bug didn't impact them so harshly.

But naaaaaaah. Gotta cut payroll, brb.

499,999,990. Could be worse.

Edit: oh I see someone else beat me to it

Make it per affected device and then we can talk.

The answer is simple: they have no idea what they are talking about. And that is true for almost every topic they are reporting about.

Honestly, with how terrible Windows 11 has been degrading in the last 8 or 9 months, it's probably good to turn up the heat on MS even if it isn't completely deserved. They're pissing away their operating system goodwill so fast.

There have been some discussions on other Lemmy threads, the tl;dr is basically:

• Microsoft has a driver certification process called WHQL.
• This would have caught the CrowdStrike glitch before it ever went production, as the process goes through an extreme set of tests and validations.
• AV companies get to circumvent this process, even though other driver vendors have to use it.
• The part of CrowdStrike that broke Windows, however, likely wouldn't have been part of the WHQL certification anyways.
• Some could argue software like this shouldn't be kernel drivers, maybe they should be treated like graphics drivers and shunted away from the kernel.
• These tech companies are all running too fast and loose with software and it really needs to stop, but they're all too blinded by the cocaine dreams of AI to care.

It's sort of 90% of one and 10% of the other. Mostly the issue is a crowdstrike problem, but Microsoft really should have it so their their operating system doesn't continuously boot loop if a driver is failing. It should be able to detect that and shut down the affected driver. Of course equally the driver shouldn't be crashing just because it doesn't understand some code it's being fed.

Also there is an argument to be made that Microsoft should have pushed back more at allowing crowdstrike to effectively bypass their kernel testing policies. Since obviously that negates the whole point of the tests.

Of course both these issues also exist in Linux so it's not as if this is a Microsoft unique problem.

It was a Crowdstrike-triggered issue that only affected Microsoft Windows machines. Crowdstrike on Linux didn't have issues and Windows without Crowdstrike didn't have issues. It's appropriate to refer to it as a Microsoft-Crowdstrike outage.