This repository was archived by the owner on Nov 1, 2026. It is now read-only.
Repository navigation
I'm running iOS 16 on Win11 and I'm 80% complete. #312
xtrkcn-prog
started this conversation in
Show and tell
Replies: 1 comment 9 replies
|
and here we see what happens when you replace your brain with AI |
9 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Booting real iOS 16.5.1 (build 20F75) on iPhone 11 (N104AP, A13/T8030) inside Inferno (QEMU-based Apple Silicon emulator) on Windows 11.
Setup:
ARM-side Inferno QEMU emulating the iPhone, paired with a separate x86_64 companion QEMU (Ubuntu, KVM) running libimobiledevice/usbmuxd/irecovery/idevicerestore
Own legitimately-issued Apple TSS ticket, official IPSW - no third-party SHSH/keys used
Verified correct Erase RamDisk via BuildManifest for n104ap
Progress:
Real iOS 16 kernel boots: recovery APFS mounts, virtual NVMe partitions, SEP/display stack/launchd/libignition all initialize
Traced and confirmed via kernelcache disassembly: PHY role encoding is 0x20000 = device, 0x40000 = host, 0 = isolation - byte-exact match with trace
Found DeviceTree is intact and applied correctly (ruled out missing tunables as the cause)
Discovered usb-drd has three MMIO ranges, not two: 0x39010000 (DWC3), 0x39020000 (config) - both modeled in Inferno - and 0x3901c000 (size 0x1800) which is completely unmodeled (reads silently return zero, don't even appear in trace)
Root cause now localized precisely: AppleUSBXDCIARM::cableChangeOccurred never fires because it waits for a software notification from IOPortTransportState/the accessory stack - not a raw MMIO poll. The force-usb-device boot-arg selects the desired role but does not by itself generate a cable-connect event. As a result _cableConnected stays zero, powerOn() is never called, and the PHY configures briefly as host before dropping into isolation
Currently pinning down the exact fields and the least invasive way to inject that cable-connect notification without patching the kernel itself
Next step: implement the missing notification path in Inferno so the emulated PHY reports "cable connected," letting DWC3 complete the device-mode handshake (RUN_STOP) and continue past USB init toward SpringBoard.
All reactions