If you have a capable computer, read this:

christian.kellner.me/2019/02/2

Despite the fancy name, plugging in unknown devices is probably always going to be a huge risk & I honestly can't think of a scenario where you'd be plugging in something you found on the parking lot if you at all care about security.

Now there are of course risks such as having your charger secretly exchanged for a malicious one, but if the attacker is this determined, you probably need a whole new strategy.

Follow

Also worth noting that this affects , , etc. as well, it's a general problem with how OSes allow DMA access for performance, not something specific to .

Seeing the massive performance degradation that & fixes could cause, (up to 30% in some workloads), am not even sure their current approach is wrong in the general case. It seems to be a case of performance vs maximum security, pick one.

So it may end up being the case that we'll need to think about security a lot more explicitly/manually ourselves when performing certain actions, rather than simply relying on the OS to somehow shield us. It seems logical to me that no matter how hard we try, once an attacker has physical access, there's little one can do to stop them.

The problem here may very well be USB-C itself, which is being overloaded for so many functions, (charging, data transfer, display output/eGPU...), there are USB-C devices on eBay that fry the motherboard for one, but I do have to admit that having one cable take care of multiple devices made my own personal space much less cluttered and convenient, so it's not an easy solution, as always.

@MatejLach
I think a larger comparison would be convenience vs security. Performance is key, as it helps with convenience, proper security hinders both performance and convenience.

Researchers mainly focused on software exploits, as those were the easiest to find(convenience) , and the industry became lax on hardware security. This will affect manufacturers and consumers for a long while, until both are up to speed once again....assuming they ever were to begin with.

@TheCzar Agreed. I think part of the reason why HW security lags is also because unlike software, where developers can live off one SW for years, so it's worth paying some attention to security bugs etc. whereas with hardware, it's usually 'done' once on the market and the OEM is already working on the next one, no time to fundamentally redesign anything.

Of course the barriers to hardware verification and security testing are also much higher, as you say, so it doesn't get much attention.

@MatejLach Which is also why "USB condoms" exist - adapters that only allow for certain functions to be used.

Sign in to participate in the conversation
Matej Lach's mastodon

Hi there! I am a free software developer. I enjoy working on useful software, as well as advocating for software freedom and the use of open standards, promoting data ownership, decentralization and privacy. If this is important to you, I may be worth following. If you like Go, Rust, or Swift, it may be worth following me as well. Besides computing, I enjoy metal, a good read and occasionally some gaming, (not much time for that these days).