[scottjones]

Summary

The 14" M3 MacBook Pro (J504) runs the same T8122 display and GPU firmware images as the M3 MacBook Airs. Its ADT names DCP image 90F849E1-… and GPU image DF697F05-…, and its iBoot is the Airs' 14.x one. This PR lets the kernel drive both by default, as on the Airs.

 drm/apple  iomfb_v14_7.c
   dcp_v14_boards
     T6030, J613, J615
+    J504: J613 image, 3024x1964 with 74 notch rows, 60/120 Hz, brightness in nits
     J516C, J514C
 drm/asahi  m3
-  validated_boards: j613, j615
+  T8122_BOARDS: j613, j615, j504   (exact list, host-tested)
 dts  t8122-j504.dts
+  &dcp panel (the J514S 14" mini-LED panel), &framebuffer0 panel link

The GPU starts at the same conservative cap as the Airs: state 2 (618 MHz), the 22 W target and the Air thermal limit. Raising it for a fanned chassis is a separate qualification. External displays and the HDMI port are not in this PR.

Paired m1n1 change: omacom/m1n1#25. Ship order: m1n1 must not ship before this kernel. A J504 with the new m1n1 and an older kernel gets the display handoff, the old kernel has no J504 record ("display not started: no board record for this machine"), and the panel stays on the boot framebuffer. Release them together, or the kernel first. The m1n1-aurora package bump will carry a floor on the linux-aurora release that has this change.

Evidence

Hardware: Jordan Goodman's J504 in the lab (jordan-m3, 10-core GPU), with stock 7.2.11-14 as the baseline.

• Before (j1105-174533-6072): m1n1 "not armed: no display handoff", DCP left disabled, simpledrm at 3024x1890, llvmpipe. OMT 28 pass / 8 fail.
• After (j1105-183940-5ec1, normal cmdline, no m1n1 config lines):
  • "J504 display started: 3024x1964@120, 74 notch rows shown".
  • GPU firmware identified as T8122 C0 14.8.3. The firmware accepted the InitData and is running at cap 2 on 10 cores; the desktop renders on it.
  • OMT 30 pass / 5 fail: gpu.driver, gpu.vulkan, display.outputs and display.controller pass.
• Brightness (j1105-191604-59b8): apple-panel-bl max 500. Set to 125, 375 and 140; each read back exactly. That readback is the driver's own state, so a visual check that the mini-LED panel actually steps is still owed. Jordan will check it at the next session.
• Host tests:
  • tools/asahi/dcp14-panel-tests.py and tools/asahi/m3-runtime-tests.rs pass.
  • Dropping the J504 record, moving it to another machine, opening it to any T8122 machine, a wrong panel width, and dropping or swapping the J504 board entry each fail a test.
  • KUnit cases are added to v14-board-test.c and iomfb_v14_7_backlight_test.c, for CI to run.
• Restore: the Mac was restored afterwards (j1105-195107-f2ca).

Merge Danger

Door: two-way

Blast Radius: J504 only

Only a root with apple,j504 gets the new record and the GPU start. The Airs, the M3 iMacs (J433/J434) and the other SoCs are unchanged, and a J504 without the new m1n1 behaves as today. No CTS, sustained-load or thermal qualification is claimed.

🤖 Generated with Claude Code

[scottjones]

Hardware results from our Mac lab, kernel built from 61cf178.

J504 (14" M3 MacBook Pro), j1107-134052-6e88:

• stage2 9b662fca. eDP-1 at 3024x1964. G15G at board defaults, firmware ready.
• Backlight set to 125, 375 and 140, each read back correctly.
• GUI smoke passed. Vulkan 782 dispatches, 0 bad of 24064. 0 DART faults.
• Test suite: 0 regressed vs stock, 5 fixed (display.backlight, display.controller, display.notch-strip, gpu.driver, gpu.vulkan).

No-regression check on a 13" M3 Air (J613), j1107-132933-b912:

• stage2 9b662fca. eDP-1 at 2560x1664. G15G firmware ready with no flags.
• GUI smoke passed, 0 DART faults, 0 regressed.

We also booted a combined trial kernel (#363 + #372 + #388 + #400, with m1n1 #23 + #24 + #25, stage2 v1.6.1-473-gd2b8d44b) on four Macs: M3 Air and J504 Vulkan 780/0 bad each, J514M Hyprland at 3024x1964@120 with Vulkan 694/0 bad (mesa #32 staged), MacBook Neo wifi soak 9/9. 0 DART faults on all four, every Mac restored to its own stock boot.bin, and no regressions vs stock apart from audio.microphone-signal on the J514M, which fails on stock too.

[maralcbr]

Reviewed: J504-only board record and T8122 admission; board checks exact, no change for J613/J615/J433 or other SoCs; CI green and lab results clean. Merging ahead of m1n1#25 per the ship order.