[erikwb]

Map all four T6050 DP IN adapters, forward their HPD interrupts to the display driver, and apply the host-link, TMU and DP OUT policies needed for USB4 display tunneling. Preserve the existing M4 handshake policy.

Wait 100 ms for a T6050 request generation to settle before handing it to DCP, so transient requests do not occupy the second display processor during discovery. Block delayed IRQ and polling paths from reading registers after powered retirement.

Based on integration/aurora, which has the Apple USB4 router groundwork from #1323. Also depends on #1270 and its prerequisites for the display-tunnel API and implementation. Combine these dependencies for builds and testing; this branch contains only the Thunderbolt handoff commit. Its feature patch is unchanged by this restack.

Validation: the combined dependency tree builds Image.gz, modules and all 137 Apple DTBs, including all 11 T6050 boards. ARM64 QEMU passes 245 display KUnit tests with one existing skip, with debug mutex/work-object checks and panic_on_warn=1. All 20 affected binding documents and focused source/modeled-handoff DT checks pass; the three-port resource check and 30 reciprocal graph edges pass. Full-registry schema validation retains the documented unrelated tunable-type collision.

Hardware validation: the final combined source tree booted on J716c with clean upstream Aquamarine. The internal panel runs at 3456×2234@120 Hz, and the two external monitors were captured at 3840×2160@240 and @120 Hz. The user tested the TS5 on all three physical ports and confirmed both displays, USB input and Ethernet working. The firmware-27 swap timestamp fix that boot needed is now #1404.

The hardware image also combines the public platform PRs #936, #938, #941, #942 and #943 for storage, networking, SMC and related platform support. These are boot-test inputs, not extra ancestors of this feature branch. The combined test tree remains local and is not a PR base.

Created with GPT-6 Astra in Codex.

[DjDeveloperr]

Tested on the M5 Pro: thunderbolt and thunderbolt_apple load cleanly with this change. J714s's ATC2 router (enabled by t6050-max.dtsi) gets its mailbox but defers, because J714s has no board ports or DROM yet; it fails quietly, with no oops. Tunnels weren't exercised (nothing attached). apple-dpin-lifecycle passes in KUnit.

Setup. 14-inch M5 Pro (J714sAP, T6050, macOS 27.0 firmware). RAM boot through aura of #1273's head (the whole stack on ewb/j716c-usb4-public-base), plus our M5 Pro GPU commit, plus #1308. #1308 is needed because the base doesn't build t6050-j714s.dtb; see there. The kit DT is the J714s one compiled on this tree. Modules: nvme, display, smc, input, gpu and usb4 (mux-apple-display-crossbar, phy-apple-dptx, atc-t6050-dp, dwc3-apple, sn201202x, thunderbolt, thunderbolt_apple). No display or dock was attached. ATC0 carries the debug link.

Whole-stack result on J714s.

• Boots to a shell with no oops, WARN or BUG.
• 2ab000000.phy and 383000000.phy -> apple-t6050-dp-phy (phys .0-.5); 2aa280000.usb and 382280000.usb -> dwc3-apple. typec port0 and port1 register (device/sink, no partner). The debug port is left alone and the KIS link stayed up throughout.
• Internal panel: display-subsystem -> apple-drm, card1-eDP-1 connected and enabled at 3024x1890. GPU: card0 and renderD128.
• ATC2 router mailbox 381108000 binds and thunderbolt_apple loads. 381ac0000.acio stays deferred because J714s has no board ports/DROM yet.

KUnit on the same tree (ARM64 QEMU, panic_on_warn=1, PROVE_LOCKING, DEBUG_OBJECTS_WORK): 356 tests, 354 passed, 2 skipped by design.

[DjDeveloperr]

Leaving this open: it calls apple_dcp_tb_dp_event() and the APPLE_DP_TUNNEL_* events from #1270, which is based on #1012 (firmware-27 display), and #1012 isn't being merged in this round. Its base, #1323, is going in now; this can follow with #1270.

🤖 Generated with Claude Code

[RyanTheTide]

Needed before this can go in:

1. It does not build on its base. drivers/thunderbolt/apple.c uses apple_dcp_tb_dp_event(), APPLE_DP_TUNNEL_IRQ_HPD and APPLE_DP_TUNNEL_HPD_CHANGED, which integration/aurora does not have; they come from include/linux/soc/apple/dp-tunnel.h in #1270. Please base this branch on ewb/j716c-public-usb4-drm (or move those declarations here), so CI builds what is merged.

2. Two changes reach every Apple silicon host, not only T6050. Both are keyed on tb_port_is_apple_host_dpin(), which is true wherever host_dp_policy has TB_HOST_DP_NOTIFY (the M1, M2 and M3 DP IN profiles too):

   • tb_dp_apple_ssc_offset() now rewrites the DP OUT adapter's TMU capability word 6 on every Apple host tunnel.
   • tb_dp_activate() no longer clears ADP_DP_CS_3_NO_AUTO_LT when the tunnel goes away, so a DP OUT adapter stays without automatic link training after an Apple host tunnel leaves.
     Either limit them to T6050 (an apple_cio_hw flag, like no_asym/host_tmu), or keep them for all hosts and add results from an earlier SoC (M1/M2 class) with a Thunderbolt display: tunnel up, unplug/replug, and the dock's own DP OUT used again after the tunnel is torn down.

3. Please give the commit message the source of TB_DP_APPLE_SSC_OFFSET (1500) and the tmu_params_host_enhanced constants (how they were measured), and name word 6 of the TMU adapter capability in tb_regs.h instead of using out->cap_tmu + 6.

[RyanTheTide]

Correction to point 3 above: TMU_ADP_CS_6 already exists in tb_regs.h (0x06), with TMU_ADP_CS_6_DTS (bit 1) and TMU_ADP_CS_6_DISABLE_TMU_OBJ_CL1/CL2 (bits 2 and 3). tb_dp_apple_ssc_offset() replaces bits 15:0 of that word with 1500 (0x5dc), which sets bits 2 and 3 and clears bit 1. Please use TMU_ADP_CS_6, add a named mask for the SSC offset field, and explain why the field overlaps those bits on this adapter (or write only the bits that belong to it).