And ultimately, that's why I don't use GrapheneOS.
But it could be a great OS if it didn't insist on denying owners control of their devices, so it's a shame.
It's looking like this case wasn't even a security fix, just DRM.
Conversation
Verified boot is an important security feature primarily used to make privileged persistence much more difficult for an attacker. If they can simply write out a vulnerable SoC boot chain, it doesn't work. It's secondarily used for anti-tampering and the same thing applies to it.
2
Yes, but my security model (for my phone) assumes I always have physical custody of my phone, so verified boot is worthless to me.
I understand and agree it is important to others.
I'm not suggesting taking any of that away from them.
2
1
The primary threat model for verified boot is defending against a remote attacker trying to persist on the device, not physical security. Anti-tampering is a secondary and less important threat model for verified boot. Chromebooks don't really bother even trying to do that part.
3
1
If the owner could re-flash the entire device, it wouldn't be possible for a remote attacker to persist in the first place...?
1
You flash device by interacting with firmware on the device and there are a whole bunch of components which have firmware: SoC firmware including firmware for GPU, media encode/decode, image processing, crypto engine, TEE and much more, touchscreen, battery, USB controller, etc.
2
I guess we're now entering the realm of "what if", but it would be nice to have a ROM that can be used to clean flash it.
(And open source firmware for everything)
1
Open source firmware for everything implies making a new SoC based on open RISC-V core designs and creating a GPU and all the other components like the memory controller, USB controller, battery, touchscreen, TEE, secure element, etc. as part of that too.
2
My desktop workstation has entirely open source firmware (POWER9), so that's basically been done before.
1
2
Talos II has mostly open source firmware but the SoC itself isn't actually open yet and there are still proprietary components. OpenPOWER is open but those POWER9 CPUs are not actually OpenPOWER with open core designs, they just supposedly will be in the future at some point.
2
There are open source cores that are almost standard OpenPOWER published but they aren't competitive and are missing things that are required by the current standards since they're older. IBM is missing some follow through on publishing stuff just like Pixels and OpenTitan.

