Return to

Intel FUBAR ... again - Kernel memory leak in nearly every Intel CPU of the last decade (Spectre hits everyone, Meltdown still Intel exclusive)



You can now test your ARM devices (aarch64 only right now) :smiley:


Anyone feel like buying and testing one :smiley:





10% of all profits and the money they have in the bank that is calculated at the end of the lawsuit not before to prevent them from just setting money aside


Good point.


“Appeal, appeal, appeal”

Sooner or later it goes away or in your favour… seems to work for intel when it comes to courtrooms.


Google would probably just switch to EPYC servers instead of suing, and offload their ENTIRE Intel server racks. Get ready for a flood of second hand server parts.


God yes

…Even with the meltdown shenanigans going on

If these start shipping in oem builds then what sort of damage might we see to nvidias market share in just one single year?

I am talking about hitting them where it hurts

e.g the steam survey they keep digging out.



Then they will start operating with nothing in the bank at all, since they will be selling patents to and buying patent licences from “Intel Corp. Ireland” who actually has all the money in the bank.


I am also wondering how early IBM was told about this. Google has been working on an OpenPOWER mainboard with Rackspace for a while now, it’s strange to me that Project Zero would not have tested a chip (IBM POWER9 LaGrange) that is about to be included in a system for their own company.

However, Raptor Engineering mentions some sort of changes in POWER9 production silicon. Was IBM actually told in advance but left out of the report? Or is this just a change to the microcode in on-chip ROM?


Needs a ‘repeat offender’ rule of extra % on top

Otherwise most big companies would work that 10% into their model


10% + (4% * (offense count -1) )?


I think you might have sort of answered your own question. Why does IBM have changes ready for the silicon? Because their HW was amongst first tested and they were the first informed? Seems likely to me as well.


Either that or they actually care about security.

The S in Intel stands for security.


What about

S for security? Yes/No? :stuck_out_tongue:


Yes please!


Crit weight in. Intel bad amd save jadajadajada.


non tech colleagues of mine know more.


Still nothing from the qubesOS team. Curious what they will say about it from a security standpoint and also if it impacts performance on their software.

So far… they are looking into it.




“Prefetching may fetch otherwise inaccesible instructions in Virtual 8086 mode.”

Which is quoted from an even earlier article.


You mean… let me read it first before speculation. That explains some grumblings i have read in the past.