Intel's latest problem...

[kladner]

Originally Posted by xilman

.....
The above does not provide any protection against WiFi attacks, unfortunately.

I keep WiFi shut off at the router. When Dan works from home with his laptop, it is hard wired.

[xilman]

Originally Posted by kladner

I keep WiFi shut off at the router. When Dan works from home with his laptop, it is hard wired.

What CPU/OS runs your router ...

[mackerel]

For the current vulnerability, I checked my potentially affected systems yesterday. Two had updates and were not vulnerable after applying the update (one Z370 mobo, one Z170 mobo). The rest were vulnerable with no current or indicated future update (two other Z170 mobos, one X299 mobo, one Skylake laptop).

To my understanding of the current vulnerability, it is extremely unlikely to be exploited on personal systems. I'm not going to worry about it, or take any extra precautions about it beyond occasionally checking for any new bios updates.

[mackerel]

Erm... I have to correct myself on the last bit. I got my reported problems mixed up. This one appears to be multiple buffer overflow attacks, whereas I was thinking of the USB attack vector. So... it may be a much bigger deal than I thought it was. At least my main system is patched already...

[kladner]

Originally Posted by xilman

What CPU/OS runs your router ...

Good question. I don't know. It is a Linksys, FWIW.

EDIT: Asus has an ME update tool, at least for my Z170 Mk1. It is the second release addressing the vulnerability.

[mackerel]

It's all over the internet now. A flaw has been found in what is believed to be most Intel CPUs from the last decade where it fails to separate kernel and userland stuff. As a hardware problem, the fix/workaround is apparently only possible at OS level, where there will be a performance impact when switching between the two modes.

Following article seems to be the most referenced one: https://www.theregister.co.uk/2018/0...u_design_flaw/

Obvious question, will this significantly affect Prime95 (and similar) performance? Some headline-grabbing big percentages have been thrown around, and they may be the case in certain conditions, but we're lacking in data right now. If the performance hit is sufficient, I may choose to not patch on dedicated crunchers.

[ET_]

Originally Posted by mackerel

It's all over the internet now. A flaw has been found in what is believed to be most Intel CPUs from the last decade where it fails to separate kernel and userland stuff. As a hardware problem, the fix/workaround is apparently only possible at OS level, where there will be a performance impact when switching between the two modes.

Following article seems to be the most referenced one: https://www.theregister.co.uk/2018/0...u_design_flaw/

Obvious question, will this significantly affect Prime95 (and similar) performance? Some headline-grabbing big percentages have been thrown around, and they may be the case in certain conditions, but we're lacking in data right now. If the performance hit is sufficient, I may choose to not patch on dedicated crunchers.

http://www.mersenneforum.org/showthread.php?p=476182

[mackerel]

Doh! Search fail for me. Thanks for link.

[ewmayer]

First the official-PR version with the anodyne-as-possible 'flaw' title:

Intel asks customers to halt patching for chip bug, citing flaw | Reuters

Next, a truthier take:

Linus Torvalds declares Intel fix for Meltdown/Spectre ‘COMPLETE AND UTTER GARBAGE’ | TechCrunch