Skip to content

drm/apple: m4: accept the J604 firmware's page-aligned D451 buffers - #1389

Merged
RyanTheTide merged 1 commit into
aurora-silicon:integration/aurorafrom
aeries0x:j604-dcp-pr
Nov 10, 2026
Merged

RyanTheTide merged 1 commit into
aurora-silicon:integration/aurorafrom
aeries0x:j604-dcp-pr

Conversation

@aeries0x

@aeries0x aeries0x commented Nov 9, 2026

Copy link
Copy Markdown
Contributor

Summary

On my J604 (MacBook Pro 14-inch, M4) the internal panel stays on simpledrm, even with
apple_dart.j713_dcp=1 apple_dart.j713_display=1. aura already hands the display over on the J604
("M4 DCP firmware mappings published", "M4 native scanout diagnostic enabled with preserved SID0
mappings"), both display DARTs admit the inherited tables, and the DCP gets through warm RTKit adoption
to DCPLink READY. Then the J604's 26.6.2 firmware asks D451 allocate_buffer for a buffer aligned to
4096, the internal path accepts only an alignment of 4, and the session is quarantined:

apple-dcp 30ae00000.dcp: DCP callback D451 tag=31353444 20/28 remote=0 flag9=0 dispatch failed: -22
apple-dcp 30ae00000.dcp: Retained DCP session quarantined: -22
apple-dcp 30ae00000.dcp: IOMFB startup failed: -22; retained until guarded reboot
apple-drm soc:display-subsystem: failed to bind 30ae00000.dcp (ops dcp_comp_ops [appledrm]): -22

This adds a panel profile flag, alloc_page_align, set only for the J604: its internal DCP may also ask
for an alignment that is a power of two up to a page, the rule the external DCPs already have. Every
other check of the qualified request (options 0x703, size, the NULL output flags) stays as it is. The
J713 and J715 profiles don't set it, so nothing changes for them.

Testing

J604 (Mac16,1), macOS 26.6.2 firmware, tethered RAM-only boots through aura/main 21a70e6e plus #137 and
#139, with apple_dart.j713_dcp=1 apple_dart.j713_display=1 on the command line:

  • integration/aurora e220b4d without this commit: D451 refused as above, desktop on simpledrm.
  • The same plus this commit: IOMFB startup complete: 317 callbacks, 3024x1964 at 804560 kHz,
    KMS scanout activated, appledrmfb on card1 (eDP-1 connected); the login screen and the desktop come
    up on the panel through DCP, and G16 then registers and reaches READY.
  • This branch's base, integration/aurora ad8751c, plus this commit: the same result. With a
    temporary log line (not part of this PR), the firmware's D451 requests during startup were all options
    0x703 with the output flags 0/0/0: 14 x size 0x1000 align 4, 13 x size 0x1000 align 4096, 2 x size
    0x40000 align 4. A DPMS off/on from Hyprland ran a full panel power cycle (power on -> off, M3 down,
    power off -> on, backlight on again) and the desktop came back.

The test builds also carried #1329 (G16 layout), #1367 (MTP) and local changes unrelated to the display
(read-only RTC/NVMEM, sound nodes off, the SMC power-off write limited). Tested on the J604 only; not
tested: sleep/resume, brightness, external displays.

Unrelated to this change but seen on every boot with the opt-in: the DCP DART's validation warns
Missing '#address-cells' in /soc/dcp@30ae00000 (and #size-cells) from of_n_addr_cells() in
apple_dart_dcp_validate(), with and without this commit.

Assisted-by: LLM

On the J604 (MacBook Pro 14-inch, M4) the internal DCP's 26.6.2 firmware
asks D451 allocate_buffer for buffers with an alignment of 4096 as well
as 4. The internal path accepts only 4, so the first such request fails
the callback, the retained session is quarantined and appledrm does not
bind: the panel stays on simpledrm.

Let a panel profile accept an alignment that is a power of two up to a
page on its internal DCP, which every buffer has, as the external DCPs'
rule already does, and set it for the J604 only. Every other check of
the qualified request stays as it is; nothing changes for the J713 or
the J715.

Tested on a J604 (26.6.2) with apple_dart.j713_dcp=1 and
apple_dart.j713_display=1: the DCP starts, appledrm binds the panel at
3024x1964 and the desktop runs on it.

Assisted-by: LLM
Signed-off-by: Friedrich <261542475+aeries0x@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants