AMDGPU Working On "Secure Display" Functionality

[phoronix]

Phoronix: AMDGPU Working On "Secure Display" Functionality

The AMD Radeon "AMDGPU" open-source Linux kernel driver is tacking on another new feature: Secure Display TA...

http://www.phoronix.com/scan.php?page=news_item&px=AMDGPU-Secure-Display-Patches

[uid313]

This rogue third party that they don't want to snoop on the application framebuffer is you.
They don't want you to be able to take screenshots or do video capture.

[microcode]

Hey if it can actually be accessed by applications that aren't just DRM, then I have some use for it.

[gsedej]

Anti-user/pro-corporate features.

Edit: it's not DRM related (e.g. capturing licensed movie)

[Jabberwocky]

Does this mean it will only work for a maximum of 4 screens?

[agd5f]

This has nothing to do with DRM. It's for detecting changes in a mission critical display.

[Dovi]

This has nothing to do with DRM. It's for detecting changes in a mission critical display.

Thank you for the clarification

[andrei_me]

mission critical display.

Medical equipment/vehicle dashboards/casino machines are examples of such display?

[drlamb]

Medical equipment/vehicle dashboards/casino machines are examples of such display?

These are what came to my mind as well. Can't exactly have a modern vehicle's dash display disappear while in motion...

[polarathene]

What kind of scenario/attack does it prevent?

I mean if the attacker were able to modify the framebuffer, isn't it likely they could also manipulate the display via other means? Eg if displaying some critical info on a dashboard display for making real-time decisions for some profession like surgery or precision distance measurements, couldn't the memory be modified related to the UI display?

Although I guess such a system already has memory protection in the first place, so this is just an additional measure? All I know is with CheatEngine it was really easy to modify memory for applications that either change the actual value that logic operates on, or just the UI representation (false positive if trying to cheat or "hack" a game). I assume if is that easy for users that legitimate threats are much more capable.

In those movies where they compromise camera feeds (which I assume isn't as quick and easy as fiction depicts it) and put them on a loop, I suppose this would be a scenario this could help prevent?