Root Cause of Reboot Issue Identified; Updated Guidance | Hacker News Reader

Root Cause of Reboot Issue Identified; Updated Guidance

newsroom.intel.com

105 pointsbhauer40 comments

"reboot"

Crash. My PC does not normally reboot without me having asked it to. If it does so, it's a crash (unless it's fucking windows update).

Yup, its like an automaker issuing a service bulliten for 'unintentional parking', or 'rapid decelleration'. I get there must always be some marketing involved, but that breaks down fast when it just feels dishonest. That is fast becoming a trend with Intel it seems. And intel desprately needs trust from an industury that depends on them. I really hope they can refocus, and not die a slow death of 'everything's fine'.

Out of all of these types of terms, I like the aviation industry's term for a crash the most: uncontrolled flight into terrain.

Or rocketry: rapid unscheduled disassembly.

Unequested fission surplus.

Rocketry's synonym to that is lithobraking.

To be fair, controlled flight into terrain unfortunately also has precedent.

.. causing rapid plastic deformation.

The phrase I have heard is controlled descent into terrain. Sounds much calmer.

Apple did it since approximately forever for kernel crashes as well. Anything to avoid any agency and shift the blame onto the user. https://support.apple.com/library/content/dam/edam/applecare...

From the same page:

> Your computer restarted because of a problem: https://support.apple.com/library/content/dam/edam/applecare...

> Your computer was restarted because of a problem: https://support.apple.com/library/content/dam/edam/applecare...

Isn't that the message you get if you forcefully power off (long-press power) a hanging mac, instead of a panic where it would reboot automatically after the core dump?

Yeah, it is. A kernel panic dumps a vague message into the middle of the screen prompting you to reboot, in a bunch of different languages.

But in recent years, it also reboots quickly enough that you'd have to be sitting there watching to catch it.

> unintentional parking

That one doesn't have to be a euphemism; I could totally see a self-driving car seeing a parking space beside you as you're driving down the road, and suddenly deciding that it would really rather slot itself in there than keep driving.

I'm not happy with the wording, but if the problem it actually causes the systems to entirely reset and go to post again automatically, spontaneously reboot probably is more accurate than crash, as it implies crash (but the implication may be missed by some). Also, depending on the role of the system, and how often the reboots happen, which is worse of a crash and a spontaneous reboot may be different. A reboot once a day is different than a reboot once every 10 minutes which is different than a continuous reboot prior to completing the boot process.

A reboot might be a crash in disguise, followed by a hardware watchdog timeout.

Worse than that: they were telling the press that “computers were rebooting _more_often_than_normal_”

Most people in the press, and their consumers, expect computers to be rebooted on a regular basis. "Have you tried turning it off and on again" isn't just a line from the IT Crowd, it's been the way of the PC world for decades.

Sure, but this is like describing a scenario where automobile brakes are randomly spontaneously being applied as “cars are being parked more often than usual.”

It highlights the lack of computer literacy in the population at large that they can get away with saying something like this (which makes it seem not as bad as it actually is)

Linus is not mincing his words on this issue:

"Patches slammed as 'complete and utter garbage' as Chipzilla U-turns on microcode" https://www.theregister.co.uk/2018/01/22/intel_spectre_fix_l...

Let's not forget, this stuff is very very hard. That needs to be appreciated.

But man, Intel be fuckin' up so bad right now.

I'd be interested to actually see a discussion of what the reboot issue was.

However, it's good news for everyone who uses affected products that Intel has produced a working fix.

I would presume that something this low level is quite commercially sensitive so unfortunately we might never find out.

As far as I can tell they have not released any technical information about how the original microcode update worked.

Security through obscurity at its finest, sadly.

It worked for them for so long. Old habits die hard but die they must.

They don't seem to be dying

Will Intel release it as a patch/firmware for those of us who have hardware by OEMs that aren't likely to offer anything?

These fixes are typically deployed as microcode updates, which can be loaded by the operating system during the boot process.

Interestingly enough, on HP's website with the security bulletin on Meltdown and Spectre, they have brief description of the vulnerabilities and a list of their many models. [1]

A few days ago, this had a schedule of microcode updates, with some models showing a date that the update had already been released, and most others showing anticipated release dates, with many (including one of my laptops) showing an anticipated release date of February 9 for the microcode. Now they all show "TBD."

And fwiw, I have heard a lot of people say bad things about HP products, but I have found their support to be much better than average. They are generally on point with fixing vulns, at least in my limited experience.

[1] https://support.hp.com/us-en/document/c05869091

Never dealt with their enterprise/server stuff, but the consumer end is a fucking nightmare. The website is shitty by itself, but the best part was that until recently, some very smart folks at HP decided to use the user-agent string to determine the OS of the visitor and then only show according drivers. So there I was with my Linux system trying to download WiFi drivers for Windows which I just couldn't find, I was close to freaking out. I only realized this when I tried again the next day, coincidentally using a Windows machine. They added an option to switch the OS by now but man, I had some nice insults ready for the genius coming up with this...

Too bad WiFi reception was just as shitty under Windows, literally no signal next room. Oh and the USB ports didn't work properly under Linux but OK, the system wasn't advertised as Linux compatible anyways.

Yeah, the wifi was what prompted me to go the site. I needed an updated driver. This is a Windows laptop, so the site was designed for it. I have never tried to run OpenBSD on an HP laptop, but I've heard they suck for that use.

Is anyone else finding the use of the first person singular in this post annoying? ("I apologize" vs "We apologize"). The previous post was the same. Since this was certainly highly edited by corp comms, it's hard to believe it's coincidental. It's as if Intel as a company is not taking ownership.

"We recommend that OEMs, cloud service providers, system manufacturers, software vendors and end users stop deployment of current versions." Intel seems to be saying to stop buying or deploying Broadwell and Haswell CPUs until further notice.

I may have missed the sarcasm here; but it seems they are saying to stop deployment of existing microcode updates until the fixed microcode update comes out

Intel's announcement never mentions the word "microcode". There's no reference to specific identified patches. Given that, one has to assume that the recommended remedy is not to use certain Intel hardware. PR, like contracts, has to be given the interpretation least favorable to the author.

I get that it’s fun to go down all the possible ways Intel’s statement could be misinterpreted given how legal and marketing appear to have reduced it to a mostly meaningless pile of words, but the “recommendations” section of the page that is linked right after your quote contains a longer version: “We recommend that OEMs, Cloud service providers, system manufacturers, software vendors and end users stop deployment of current versions on the below platforms, as they may introduce higher than expected reboots and other unpredictable system behavior.” The “on the below platforms” bit seems to contradict your reading...

Generally in the case of something that says "updated guidance" or is intended as a follow-up to something earlier, it pays to understand what the earlier communication was. The earlier guidance was to apply a firmware upgrade. The updated guidance is to not apply the firmware upgrade.

I saw some mention earlier of known-bad microcode updates. I assume that's what's going on.

Reply on news.ycombinator.com