Live data from Hacker News

“We have obtained fully functional JTAG for Intel CSME via USB DCI”

twitter.com

321–330 of 413 posts

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#321
post #293
post #172

Earlier quoted context omitted.

We should start demanding physical shutters for laptop webcams. Does anyone make those yet?

Indeed, bags of lens stick on lens covers (that allow you to open when you need the camera) is a product idea I thought of ages ago, but too lazy to actually do.

it exists already, got one as swag at a conference from Nvidia (???). strange gift, I like the sentiment though.

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#322
post #32

At first it looks nice "oh now we can get rid of it" but it also opens up a very scary near future security-wise. We've now entered a realm where an attacker could simply plug a device on an usb port of your computer for a few seconds to have it access your cpu's ME through USB JTAG and take over it, allowing him to have full access and control over what you do/read/open/type over the network, without you ever knowin…

The 'evil maid' attack is well known, and states that once someone has physical access to your computer, all bets are off. Anything that has DMA enabled (e.g. Firewire or Thunderbolt) offers an external device direct access to the system RAM that is very difficult to defend against, or they could attach a keylogger or modify your bootloader, basically unleash all manner of havok. USB JTAG is really no different from…

An important aspect of an evil maid attack is that it requires at least two instances of physical access, once before and once after use by an authorized user.

If the attack can he pulled off with only one time access, it’s worse than an evil maid attack.

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#323
post #208

Earlier quoted context omitted.

> And yet we really don’t seem to care much. I do care, a lot. I have decided to avoid Intel (and AMD) hardware like the pest. I will not buy any Core iSpyOnYou or AMD equivalent anymore. I'm an advocate of economic and judicial sanctions from the political level against Intel (and AMD). I tell people around me about the problems and explain how it is an issue of privacy, security, national sovereignty, and market po…

Your best bet will be with NXP then. They still haven't released the i.MX8 cpus, but i.MX7 boards are available (up to dual core 1.2GHz), and, as far as I know, they are the only (decent, omap3 & 4 don't have enough power these days) with full opensource support for the entire SoC, even the GPU

Thank you for pointing these out. The i.MX7 and i.MX8 SoCs seem very interesting, but I have a hard time finding information about available SBCs, mainline Linux and official Debian support. The only SBC I found is the Nitrogen7[0] and it doesn't seem to geared towards desktop use as there is no DP nor HDMI nor VGA, though the PCIe expansions is gold. I wonder why other SBC don't offer PCIe. They might not have the highest performance, but I don't need that.

[0] https://boundarydevices.com/product/nitrogen7/

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#324
post #231
post #208

Earlier quoted context omitted.

> And yet we really don’t seem to care much. I do care, a lot. I have decided to avoid Intel (and AMD) hardware like the pest. I will not buy any Core iSpyOnYou or AMD equivalent anymore. I'm an advocate of economic and judicial sanctions from the political level against Intel (and AMD). I tell people around me about the problems and explain how it is an issue of privacy, security, national sovereignty, and market po…

I heard https://beagleboard.org/black can boot and run Linux w/o any blobs either in bootloader or kernel (provided you're OK with a sub-par screen resolution and not using the onboard GPU) https://news.ycombinator.com/item?id=12584880 and others might have details on WiFi and such

The BeagleBoard X15 seems very interesting. The onboard GPU is definitely a problem though. Why didn't they go for a useable GPU? And no, if it does not have open Mesa drivers, it is not usable.

I really wonder why the GPU situation is such a huge mess. Very few are supported by free drivers and the closed drivers are, besides being closed, often of very low quality. Is patent law holding this situation stable? Isn't patent law there to promote advance? I feel it isn't working.

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#325
post #294

Earlier quoted context omitted.

But now you have to trust the switch to really deactivate the mic. I'm a recursive paranoiac !

At least you only need to verify a switch once, it won't be compromised by software changes.

Assuming the switch does really cut the mic signal and doesn't instead act on a i/o pin which calls a function to disable the mic input. The second solution would be hackable.

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#326
post #198

Earlier quoted context omitted.

DMA/Firewire over USB makes pretty much every systen vulnerable to USB attacks (ME aside.)

DMA-based attacks are blocked by the IOMMU, which is present in all modern machines (and has been for a few years). Linux has preferential enablement of DMA such that the IOMMU is initialised first, so even plugging in a device in early boot will not be able to exploit DMA.

Does it? To use the IOMMU for VFIO, I had to explicitly enable it via a kernel parameter.

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#327

Earlier quoted context omitted.

The point of ME is that it's invisible. And until then, few people had access to it. Now I can't wait for rogue monero miners to use ME to propagate :)

It doesn't matter if it's visible or invisible. The point is, it cannot go undetected while being used: - If it were to periodically "check in" with an external server to see if it needs to do any kind of spying -- admins would notice the network traffic. - If it needed to be contacted externally to "initiate" any kind of spying at all, that would mean anyone behind a NAT would be safe, and furthermore, the the momen…

"If it were to periodically "check in" with an external server to see if it needs to do any kind of spying -- admins would notice the network traffic."

Admins would use network dedicated hardware with network chipsets driven by closed source firmware. In a crazy but technically doable scenario, that closed source firmware could contain instructions to "send packets containing magic word X to address a.b.c.d" and not reporting or counting that traffic, even to applications opening the device in promiscuous mode, with all routers in between to obey the same instructions. Not a single byte would be reported, counted or sniffed unless someone sticks a digital analyzer on the network cable.

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#328

Earlier quoted context omitted.

It doesn't matter if it's visible or invisible. The point is, it cannot go undetected while being used: - If it were to periodically "check in" with an external server to see if it needs to do any kind of spying -- admins would notice the network traffic. - If it needed to be contacted externally to "initiate" any kind of spying at all, that would mean anyone behind a NAT would be safe, and furthermore, the the momen…

"If it were to periodically "check in" with an external server to see if it needs to do any kind of spying -- admins would notice the network traffic." Admins would use network dedicated hardware with network chipsets driven by closed source firmware. In a crazy but technically doable scenario, that closed source firmware could contain instructions to "send packets containing magic word X to address a.b.c.d" and not…

> Admins would use network dedicated hardware with network chipsets driven by closed source firmware.

No, I fully expect there are enough admins out there running dedicated hardware whose design & source code they have access to. That is sufficient.

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#329
Now that we know for sure people who have worked on or are working on Intel ME actually read and post here, I'd like to take the opportunity to refer those of you to a previous post of mine about it:

https://news.ycombinator.com/item?id=15120207

If you choose to defect and leak all the information you can, you will almost certainly be greatly praised for it by many. Of course there will be negative consequences, but no one ever said that standing firm and adhering to your morals was easy. Maybe if enough employees stood up for what they think is right instead of continuing to silently comply like slaves and let --- or even assist --- companies and governments slowly take away their freedom and privacy, there would be some actual change happening.

Re: “We have obtained fully functional JTAG for Intel CSME via USB DCI”

#330
post #186

Earlier quoted context omitted.

Prior to the NSA contractor Edward Snowden's revelations in 2013, Room-641A had already been exposed by an AT&T employee-turned-whistleblower Mark Klein. The EFF sued the government in 2006 over it. Tape over laptop cameras isn't just a "parents-of-friends" thing, it's a good idea. Buy a set of stickers and support the EFF: https://supporters.eff.org/shop/laptop-camera-cover-set Anyone know somebody at Wired?

> Tape over laptop cameras isn't just a "parents-of-friends" thing, it's a good idea. Buy a set of stickers and support the EFF: https://supporters.eff.org/shop/laptop-camera-cover-set Support the EFF! But I hate the stickers. Everyone puts a sticker on their webcam and completely ignores the hot mic. But you get that false sense of security…

I unplugged the microphone on my daughter's iMac years ago. She did not use it, and was ok with that.

I did not break her phone in that way. Her phone is more dangerous.

Post reply on HN