Lightning video adapters feature Apple Secure Boot, run Darwin kernel | Hacker News Reader

Lightning video adapters feature Apple Secure Boot, run Darwin kernel

twitter.com

128 pointsrahuldottech14 comments

Previous discussion of XNU/Secure Boot on Lightning video adapters (2013) : https://news.ycombinator.com/item?id=5307781 .

This thread has some new information and maybe people will start playing with these more eventually.

Makes you wonder if some rogue device/adapter/charger could deliver malware (sploit/trickery) to it before or after it's booted from the host.

Unlikely since the Apple adapter contains no software. The firmware is loaded by the host device when it's plugged in.

Doesn’t that mean that a rogue device that did contain software and ignored the incoming packet might actually be trusted by the host device when it shouldn’t be? The twitter thread did say the firmware payload contained no personalization (like encryption keys, perhaps).

> The twitter thread did say the firmware payload contained no personalization (like encryption keys, perhaps).

To be clear, this just means that the firmware isn't tied to a specific device via the normal ECID/nonce pair like a normal iOS update is.

For more context, check out this comment on the Panic Blog: https://panic.com/blog/the-lightning-digital-av-adapter-surp...

Don't drink the kool-aid. An adapter that destructively compresses the signal is an unexpected and unmitigated disaster, not some cool quirk of Lightning that makes "the host not care what is hanging off it".

The real reason this abomination exists is because as recently as 2018, Apple is selling products with a connector that has the equivalent bandwidth of USB 2, not even close to transporting a single 1080p video signal.

It’s destructive, and it probably also introduces a quite decent amount of latency.

Yes - this is really quite astonishing. I imagine any video professional would be horrified to learn that their CABLES were recompressing their video. There's never an acceptable reason for that it's really amazing I haven't heard about it before.

I wonder what they use for the demos at WWDC.

Why on earth would you need a kernel in your adapter

If only a comparatively small number of adapters is produced and you already have the SoC and the kernel, it may be cheaper to develop than anything else.

There are kernels running in almost every piece of peripheral hardware you own. If you have a modern WiFi chipset, LTE baseband, or GPU it's certainly running an RTOS kernel. Once you switch to a packet-based protocol with multiple concurrent tasks (decode, packet receive, link negotiation, etc.), it just makes sense - either you're implementing timeslicing / scheduling on your own or you can let someone else do it.

The issue that causes poor video quality isn't that Apple chose to run XNU or use an SoC for their adapter, it's that they chose to use H.264 at a low bitrate due to hardware support constraints rather than another encoding scheme which could offer less lossy video.

That's some cool enginerding discovered by gindrenigne. AvE, bigclivedotcom and EEVblog would be proud!

Reply on news.ycombinator.com