Hacker News Reader

Comment by SlowRobotAhead | Hacker News Reader

Crap... CTRL+F Nordic .... Whew! Seems like this is mainly in relation to the Ti CC26xx series. I'm using the Nordic right now and the only time it's mentioned here is the Over The Air update service.

For that that might not know, Ti the approach that some others (not Nordic) does where they give you an ARM Cortex M4 to put your code on and have another Cortex M0 that runs the ROM based BLE Stack. This drove me away from many of the Ti offerings because while you can't mess them up, the BLE Stack is typically not upgradable. With the launch of BLE 5 and it's mainly software upgrades it was important to us to have that ability. So here is an issue with the shared que between the two chips - I'm not sure if this will be patch-able all applications.

The hack to get console on the Aruba device is pretty cool.

View on HN

It's patchable, but requires each implementor to recompile the FW with an up to date stack. The patched stack is available on TI's site for months now.

Do you really think there is no such things (unpublished) in the Nordic BLE stack?

There are definitely (unpublished, and possibly unfound) vulnerabilities in the Nordic stack. The point (IIUC) is that the Nordic stack can be fixed once those vulnerabilities are known. The TI stack, on the other hand, is in ROM; if^Wwhen it's broken, it's broken forever. (And that appears to have happened just now.)

Nordic <3

They are doing great stuff

I can't speak to the CC26xx series specifically, but normally deeply embedded cores running a ROM stack like that have some additional SRAM which can be loaded at boot for firmware patches.

I'm interested in building a security-sensitive project with Nordic's stack, are there no ways to isolate their "softdevice" from the rest of the code running on the SoC?

The softdevice binary runs in supervisor mode and typically does all its work in one of two interupt contexts.

So... it gets its own ram, rom, and you can’t (not supposed to anyhow) access it. So... no, to answer your question, but depends on what you are trying to prevent I guess.

You don’t need to run the softdevice at all, can shut off it down temporarily like while doing some crypto work, but if there was an exploit it seems like it could be bad because the SD seems like it can probably do anything your code could plus some. Additionally with the ‘software interrupt’ peripheral it seems it has a built-in way to branch to any arbitrary code location.

However... there is a gambling application that uses one and the code has been through a couple audits as I was told.

Reply on news.ycombinator.com