[josephcsible]

If you open the hood on your Mercedes EQS, and the next time you drive it, ABS doesn't work or the speedometer goes blank, should that be considered your fault too and not evidence that the stock software is broken? (This example is not hypothetical. Mercedes actually tells you not to open the hood, and the EQS actually had a recall for those problems.)

[cududa]

It’s more like popping out a car’s headlight to get to the CAN bus to hook up some device from alibaba to unlock the door/ start the engine, then installing a third party infotainment OS on the head unit, and then wondering why ABS is broken.

[ska]

This is more like you replaced the exhaust system with something bought 3rd and reflashed your car with a 3rd party CAN bus widget, then complain when it overheats and has a flat zone around 3500 rpm.

[tester756]

By what logic?

[ska]

Jailbreaking isn't like opening the hood, it's effectively messing with the system and setup (just like the ECU example). It might not cause any problems, but it also might.

I have no problem with that by the way, but you are on your own lookout and should accept it as part of the deal. Similarly if I damage engine performance due to bad tuning it's my problem, not theirs.

I do think it's too broad of manufacturers (whether car or phone) to apply a blanket warranty denial in such cases. But in the phone case, no reason apple should accept bug reports on a jailbroken system if it can't be reproduced on a "stock" one.

[vore]

Jailbreaking is a pretty invasive procedure that exploits software faults in order for it to work. This really is not comparable.

[k9unit]

Exploits in computers and using them doesn't inherently make them more unstable. These exploits rely on existing things in code and memory to make use of. It's not like after using one, your system will be less unstable in almost all cases. It is possible that the system could somehow detect that it was tampered with, or the jailbreak itself did some things, potentially malicious, then make things like this happen. You could also say that it may not be seeable from a mac because the system is more open than the mobile version. If someone wanted to hide it, they might decide not to put it in the version that can be more easily reverse engineered.

[cududa]

Tell me you’ve never hooked an OS kernel’s undocumented features without telling me

[k9unit]

I sure have, but I have not jailbroken an iPhone. Well I hadn't imagined how a jailbreak on iOS might work either. Will it patch system files, programs, and drivers on disk persistently? Well that seems it would be the simplest way to keep the jailbreak, now that's some extra integrity checking to bypass as well. Do these jailbreakers have a number of exploits for writing to kernel memory? It's likely they would want to read it as well, but it's not always required. It seems old parts of the Darwin kernel are open source, which if it hasn't been drastically changed in every way, makes it a lot easier to understand the kernel and find problems. It would be interesting to see how these older released exploits work. How do you even do a syscall on iOS?

A bigger challenge I think would be new gen gaming consoles, this would be awesome. There does exist exploits for these I believe, but they are private and public ones get patched, then you can't play online or downgrade updates easily. Maybe it would be easier to stay on an older version and make your PlayStation think your on the latest version, maybe this is what those who have it do.

On windows, there are many problems you can run into by directly manipulating kernel structures, but that doesn't mean it cant be done safely, especially on things outside of something like win32k which is a mess. Hooking things won't get you in trouble, unless for example you're hooking integrity checked functions or data regions on Windows and get patchguarded. Which doesn't run everywhere on the windows kernel and can be maliciously disabled, not even ntoskrnl executable sections are fully protected by PG.