Intel: We've Found Severe Bugs in Secretive Management Engine, Affecting Millions (zdnet.com)
Liam Tung, writing for ZDNet: Thanks to an investigation by third-party researchers into Intel's hidden firmware in certain chips, Intel decided to audit its firmware and on Monday confirmed it had found 11 severe bugs that affect millions of computers and servers. The flaws affect Management Engine (ME), Trusted Execution Engine (TXE), and Server Platform Services (SPS). Intel discovered the bugs after Maxim Goryachy and Mark Ermolov from security firm Positive Technologies found a critical vulnerability in the ME firmware that Intel now says would allow an attacker with local access to execute arbitrary code. The researchers in August published details about a secret avenue that the US government can use to disable ME, which is not available to the public. Intel ME has been a source of concern for security-minded users, in part because only Intel can inspect the firmware, yet many researchers suspected the powerful subsystem had bugs that were ripe for abuse by attackers.
of how well "security by obscurity" works.
I want my C64 back. I want hardware I can understand and software I can control. Fuck this modern bloated 4 gigabyte web browser tab horseshit with thousands of people mashing their keyboards randomly and millions more observing my private data.
Going out on a limb here.... while Intel claims the problems affect the 6th, 7th, and 8th gen processors, I bet they probably didn't bother testing or auditing earlier systems. Hasn't ME been around much longer than that?
Really, this ought to be factory disabled by OEMs and only shipped enabled to large corporate customers.
...and very difficult to patch?
Actually on ME9 Intel changed the kernel. In ME6 they changed the platform layout.
* ME < 6: GMCH northbridge and southbridge. ME lived in the GMCH and had full access to RAM even in S5 (off) system state. Kernel is based on ThreadX. CPU is ARM core.
* ME 6-8, same kernel, but moved to PCH (formerly southbridge) and the CPU gined the GM part of GMCH. Northbridge removed from platforms. ME loses access to RAM in all states besides S0 (on) and has to make do with PRAM on PCH.
* ME9+: ME now runs on Minix and Quark CPU. Vulnerabilities become an issue.
* ME10: internal struggle for dominance between kernel and AMT teams (based in US and Israel respectively) leads to departures. (including mine)
* ME11 (12?): US team is disbanded.
whois gawk date unzip strip find touch finger mount join nice man top fsck grep eject more yes exit umount sleep dump
> I do not like the ME, but at least this is local acess exploit only
It's still fucked up.
The previous ME flaw involved gaining remote access illegitimately. This one involves being able to inject stuff into the super ultra privileged secret area that operating systems can't see or guard against once you have that access. And there's NO REASON to believe that this is the final bug that exists. So far it looks like chained vulns from internet down to a run level that the chip prevents the kernel from seeing.
There have been remote attacks capable of provisioning AMT in the wild. Intel conveniently does not acknowledged them in their NDA documents about security for some reason, can calls users with AMT turned off 'safe'. Take from that what you will about their priorities when it comes to customer's security.
Have other chipmakers clearly and unambiguously said their chips do not have a back door mechanism?
Yes, IBM's Power series of CPUs are fully open without any equivalent of the Management Engine.
Intel can't say their chips don't have a back door. They also haven't said their chips don't have a back door so at least they are honest.
AMD is working on greater disclosure and I am prodding them as hard as I can. Internally they seem to be doing the right things, or at least trying to.
ARM has their full code base published on Github. This doesn't prevent licensees from using something else, adding nefarious things etc, but I can almost guarantee most don't. You can always checksum the code if you want.
As an aside, AMD's PSP is based on ARM's stuff which is completely open source. I am fairly sure that the majority of AMD's code in this area is unchanged from the vanilla ARM version so you could consider AMD's partially open.
-Charlie