Robert Steele: Details on NSA Backdoor in Intel Chips(phibetaiota.net)
phibetaiota.net
Robert Steele: Details on NSA Backdoor in Intel Chips
http://phibetaiota.net/2017/03/robert-steele-details-on-nsa-backdoor-in-intel-chips/
7 comments
> We've know about this for years though. Libreboot, Coreboot and even Google have begged Intel and AMD to provide chips without backdoors installed and they have refused.
Not to defend Intel/AMD here, but they might have refused because the ME (or AMD equivalent) is now required to initialize portions of the platform.
x86, by nature of being an x86, starts in 16 bit mode and requires a lot of configuration to get it working as x86_64, not to mention the plethora of IO attached to a modern CPU.
If you read some books on the ME [0], you'll find that it not only does DRM stuff, but it's also quite involved in the initialization of the IOMMU (or something similar to the IOMMU, my memory fails me) and other modern x86 subsystems everyone has grown accustomed to.
So it may simply be that Intel isn't willing to put a dollar amount on engineering a CPU able to operate without the ME to do this extra initialization.
I think the eventual solution will be a new CPU architecture which doesn't have so much legacy baggage attached, and which allows for x86 compatibility via an emulation/translation layer. I'm actually quite interested to see how well Microsoft gets along with x86 compatibility on their upcoming ARM port of Windows 10.
[0] http://www.apress.com/us/book/9781430265719
Not to defend Intel/AMD here, but they might have refused because the ME (or AMD equivalent) is now required to initialize portions of the platform.
x86, by nature of being an x86, starts in 16 bit mode and requires a lot of configuration to get it working as x86_64, not to mention the plethora of IO attached to a modern CPU.
If you read some books on the ME [0], you'll find that it not only does DRM stuff, but it's also quite involved in the initialization of the IOMMU (or something similar to the IOMMU, my memory fails me) and other modern x86 subsystems everyone has grown accustomed to.
So it may simply be that Intel isn't willing to put a dollar amount on engineering a CPU able to operate without the ME to do this extra initialization.
I think the eventual solution will be a new CPU architecture which doesn't have so much legacy baggage attached, and which allows for x86 compatibility via an emulation/translation layer. I'm actually quite interested to see how well Microsoft gets along with x86 compatibility on their upcoming ARM port of Windows 10.
[0] http://www.apress.com/us/book/9781430265719
Workaround -- USB wifi dongle that requires a button press to connect to a new domain name.
Can anyone comment on the credibility of this news outlet or the supposed source here?
Screenshot of an anonymous post on 4chan, so approximately zero credibility.
Intel ME is well known however: https://en.m.wikipedia.org/wiki/Intel_Active_Management_Tech...
Well in the screen capped communication the source talks about "the news that Trump was spied on" as motivation for the disclosure... that's enough for me to click the back button.
I really want to believe, but extraordinary claims require extraordinary evidence, and I think it's lacking in this respect.
But two problems mean this will never get fixed: 1. People still think the "I've got nothing to hide" argument is valid and therefore don't care (yet) 2. There are no alternatives out there. ARM might have a shot at taking over this market but unfortunately the companies building ARM chips are also hiding code in their chips.