Repository navigation
Description
Summary
On a Linux host on Apple Silicon (Asahi), booting an iPad Air 2 (J81) through checkm8 → Pongo depends on the cable:
- Apple USB-C to Lightning cable, direct: never reaches Pongo. Every time the iPad drops its USB connection (the checkm8 reset, the Pongo boot), the Mac's Type-C controller reports a detach and Linux tears down the whole xHCI controller. After that, Pongo never shows up on USB, and the iPad boots iOS about 21 s after
Booting PongoOS.... - Apple Lightning to USB-A cable + a passive USB-A to USB-C adapter: works, but needs two physical replugs at the iPad end, each at the right moment.
Software port resets don't replace either replug. A note in the docs, and perhaps a different replug hint, would save other Linux users a lot of failed attempts.
Host
- MacBook Air M2 (Mac14,2), Asahi (Arch Linux ARM), kernel 7.2.11,
xhci_plat_hcd+dwc3_apple. fwupdstopped andapple_mfi_fastchargeblacklisted (both otherwise grab the DFU device).- Device: iPad Air 2 Wi-Fi (J81),
orchard launchfrom internal storage (j81: internal storage (ANS1) — install and launch Omarchy on the iPad Air 2 #188), palera1n 2.4, released Pongo v0.3.1.
USB-C to Lightning cable
With this cable, each USB drop of the iPad shows up on the host as a Type-C event, and the controller is rebuilt:
macsmc 23e400000.smc: RTKit: syslog message: aceElec.cpp:1151: Elec: Elec Cause 0x0
xhci-hcd xhci-hcd.3.auto: remove, state 4
xhci-hcd xhci-hcd.3.auto: USB bus 1 deregistered
xhci-hcd xhci-hcd.3.auto: xHCI Host Controller
xhci-hcd xhci-hcd.3.auto: new USB bus registered, assigned bus number 1
This happens about 1.5 s after the Pongo upload too, without anyone touching the cable. Replugging at the iPad end then also rebuilds the controller, and Pongo never enumerated in any of these attempts. I saw this in both USB-C ports. Over the 90 minutes I used this cable there were 26 such events; with the USB-A cable and adapter there were none.
Lightning to USB-A cable with an adapter
Here the controller stays up, and this sequence works reliably:
- Replug after
Checkmate!. The device does not return in download mode by itself. - Replug about 13–20 s after
Booting PongoOS....- Replugs at 0–10 s did nothing; Pongo did not appear.
- Without a replug, or later than about 21 s, the iPad boots iOS.
- In every successful attempt,
PongoOS USB Deviceenumerated within a second or two of a replug in that window.
The existing hint ("If the Apple logo stays for about 15 seconds, reconnect the cable at the device once") matches step 2, but step 1 is needed too on this host.
Software resets don't help
- Port reset during the Pongo phase: disabling and re-enabling the hub port (
usb1-port1/disable, whichapple/replug.pyalso uses) re-enumerates a connected iPad that is running iOS. During the Pongo phase it produces no USB activity at all: five resets in the 20 s window, nothing. - orchard's own
PortResetafter checkm8: on this host the iPad does not answer addressing for a while:device not accepting address N, error -62, twice, about 6 s each. The 5 s wait between resets is shorter than that, so a reset sometimes lands just as the device enumerates. palera1n then logsFoundDevice: Unable to fetch serial number. Download mode came back in the end, but only after a physical replug. - Unbinding
dwc3-apple: this left the host without USB until a physical replug. - Reprobing the Type-C controller (
tps6598x): this brought the host back, but the iPad still needed a replug to enumerate.
Suggestions
- Docs (host requirements / device bring-up): on Apple Silicon Linux hosts, use a Lightning to USB-A cable with an adapter, not a USB-C to Lightning cable. Replug at the iPad end, after
Checkmate!and again about 15 s afterBooting PongoOS.... - The
PortResetwait could be longer than the host's address retries (about 13 s), orPortResetcould be skipped where it does not help. I can only speak for this one host.
I can test changes on this host and J81.
🤖 Generated with Claude Code
Activity