Félix

(replying to knutaf)

@knutaf I have never tried to hold an iPhone for days like this but the Windows kernel debugging experience is exceptional. Never tried holding an iPhone for days but I wouldn’t be shocked if we were a lot less graceful

3 replies →
3 replies

Saagar Jha

(replying to Félix)
@fay59 @knutaf iOS gets upset if you tie up the wrong userspace process much less the kernel. If you turn off watchdogs you’ll probably hit SSL errors pretty soon though

Gui Rambo

(replying to Saagar Jha)

@saagar Yup. I once unloaded SpringBoard for too long and got a reboot. Same goes if your hold up backboardd for too long (I eventually learned to disable the watchdog before doing that stuff)

2 replies →
2 replies

Amy

(replying to Gui Rambo)

@_inside @saagar not too surprisingly, macOS will do the same with WindowServer etc. i think it gives you 2 minutes before panicking

Saagar Jha

(replying to Amy)
@a @_inside There are a couple of these. A particularly annoying one is logd

Saagar Jha

(replying to Gui Rambo)
@_inside Yeah I remember using a particularly bad jailbreak once and I could not for the life of me get it to attach to Springboard because by the time LLDB would be ready to debug the watchdog would immediately fire

Siguza

(replying to Félix)

@fay59 @knutaf I have played around with my Bonobo and some demoted phone in ramdisk boot at some point... disabling the watchdog panics is simple, but idk how to make the OS aware of the fact that there's a debugger and be like "no, you can't put that CPU to sleep right now". Basically all I got working was live memory patching where you ignore all locking (just let the CPUs run) and just yolo it.

Siguza

(replying to Siguza)

@fay59 @knutaf iBoot and SecureROM are infinitely more forgiving though.

Pierre H.

(replying to Félix)

@fay59 @knutaf resuming ? What is resuming?

Mateusz Krzywicki

(replying to Pierre H.)

@madcoder @fay59 @knutaf in lldb *OS case its when you try to single step and you never see next instruction