Repository navigation
Existing Apple USB4 work and what applies to the T6034 #3
Description
Part of #1
Question
What Apple USB4 and Thunderbolt work already exists that the J516M spec can build on, and how much of it applies to the T6034 as it is?
Cover at least:
integration/aurora:drivers/thunderbolt/apple*,drivers/phy/apple(ATC),pcie-appletunnel support,t6030-usb4.dtsi,t603x-j514s-j516s-usb4.dtsi, and the T6030 dock and DP-in commits.- PR USB4 eGPU on Apple Silicon: AMD RX 7600 runs llama.cpp on Vulkan (non-posted BARs, identity tunnel windows) omacom/linux-aurora#188 (and USB4 eGPU on Apple Silicon: AMD RX 7600 runs llama.cpp on Vulkan (non-posted BARs, identity tunnel windows) omacom/linux-aurora#184 before it): eGPU on the M2 Max.
- Open upstream issues on Thunderbolt tunnels and sleep, such as j316c (MacBook Pro 16" M1 Max): after an unclean Thunderbolt disconnect, that port's PCIe-C tunnel stays down until reboot omacom/linux-aurora#324, j413 (MacBook Air M2): PCIe tunneling over Thunderbolt is never enabled on t8112, so the LG UltraFine 5K's hub (camera, mic, speakers, USB) is missing omacom/linux-aurora#143, aurora14: second left-hand Thunderbolt port never brings up a DP tunnel (DPRX negotiation failed, wedged until reboot) omacom/linux-aurora#374, 7.2.11.aurora9-1: Thunderbolt DisplayPort tunnel fails to set up (DP tunnel setup failed: -110) omacom/linux-aurora#268, s2idle: Thunderbolt display on the borrowed dcpext1 route stays dark after resume (DPTX validate times out), while HDMI holds dcpext0 omacom/linux-aurora#162, J516S s2idle: the external-display sleep veto, and USB keyboard wake omacom/linux-aurora#121, External-display sleep: extend qualified behavior without bypassing M3 link guard omacom/linux-aurora#40 and Dock DPMS: establish whether CRTC off/on still tears down the tunnel omacom/linux-aurora#38.
- aurora-silicon/linux and AsahiLinux/linux branches, and any upstream-kernel Apple USB4 series.
For each piece, the answer gives what it does, which SoCs it covers, and whether it applies to the T6034 unchanged, with changes, or not at all.
Activity
hungbenjamin402 commentedon Nov 9, 2026
Findings: research/existing-usb4.md on branch
research/existing-usb4.thunderbolt 0-1: Apple Inc. Studio Display) with the stock driver, which points the NHI-bind gap at m1n1/DROM.usb4-m3maxdoes nothing on the J516M: its root compatible isapple,t6034with noapple,t6031. T6034 (J516M): external displays over Thunderbolt, tested with a Studio Display at 5K (stacked on #405, overlaps #292) omacom/linux-aurora#406 addsapple,t6034but overlaps M3 Max (J516C): external displays over USB-C and HDMI omacom/linux-aurora#292.t6030-usb4.dtsi), and its IRQ and PMGR values differ from T6031's, so tunnel nodes need J516M ADT values. USB4 eGPU on Apple Silicon: AMD RX 7600 runs llama.cpp on Vulkan (non-posted BARs, identity tunnel windows) omacom/linux-aurora#188's amdgpu/TTM and PCI fixes apply as-is; its identity 64-bit window rule goes into the new nodes.