Skip to content

display: hand off the T6041 display on J614C too - #132

Merged
DjDeveloperr merged 1 commit into
mainfrom
display/t6041-max-dcp-handoff
Nov 9, 2026
Merged

DjDeveloperr merged 1 commit into
mainfrom
display/t6041-max-dcp-handoff

Conversation

@DjDeveloperr

@DjDeveloperr DjDeveloperr commented Nov 9, 2026 •

Copy link
Copy Markdown
Contributor

Why

aurora-silicon/linux#1332: on joske's J614c (MacBook Pro 14-inch, M4 Max), the internal panel stays on simpledrm. aura's T6041 M4 Max display handoff serves only apple,j616c. With apple,j614c added locally, joske's boot got through every ADT and hardware check of the J616C entry and stopped only at the device tree:

DCP: T6041: display power state (ADT /arm-io/dcp): DISP_CPU 0x502290000 = 0x1f0003ff, on
DCP: T6041: CPU_CONTROL 0x10 (running) STATUS 0xec; PROTECT dart-dcp 0x1ff dart-disp0 0x1f9
DCP: T6041: display is not status "disabled"
DCP: T6041: armed, but handoff NOT published; display nodes untouched

That boot shows the J614c ADT matching the J616C entry: chip, 26.6.2 firmware, DCP and DART addresses, the running DCP, both DARTs locked, the firmware segments, and the SID 23/0/16 page tables. The refusal came from the kernel tree, which enabled /display-subsystem itself. aurora-silicon/linux#1345 keeps those nodes disabled, as the J616s tree does, and makes the M4 Max display gate adopt this handoff.

What

  • dcp_soc_t6041_j616c becomes dcp_soc_t6041_max, serving apple,j616c and apple,j614c. Chip, firmware gate, addresses and SIDs are unchanged, and nothing changes for other boards or chips.
  • Host test (tests/t6041_display_host.c):
    • j614c, j614c-external and j614c-on-t6040 scenarios, matching the J616C ones;
    • the external regions are now checked in full for both M4 Max scenarios;
    • a display-enabled scenario for the refusal above: with an enabled display subsystem, nothing is published.

Order with the kernel (aurora-silicon/linux#1345)

Either can land first. With a kernel tree that still enables the display subsystem, this refuses exactly as above, so the panel stays on simpledrm as it does today.

Testing

  • pytest tests/python/test_t6041_display_handoff.py: 63 passed (59 before, plus 4 new scenarios).
  • Builds with clang and lld on an aarch64 Linux host (USE_CLANG=1 BUILDSTD=1): CHAINLOADING=1, and RELEASE=1 CHAINLOADING=1 ESP_STAGE2=1 BUILTIN_LOGO=aurora. No new warnings. clang-format --dry-run --Werror is clean on the changed sources.
  • Not run on hardware. No J614c is available here. The ADT evidence is joske's log above. The repository policy asks for a hardware test before merge; the owner asked for this to be merged and then tested by joske on the next build.

🤖 Generated with Claude Code

The J614C (MacBook Pro 14-inch, M4 Max) has the J616C's T6041 display
block. On joske's J614C (linux#1332), with apple,j614c added to the
J616C entry, every ADT and hardware check of that entry passed: the
DISP_CPU power state, the running DCP, dart-dcp and dart-disp0 locked,
the segments and the SID 23, 0 and 16 page tables. It stopped only at
the device tree check, because that kernel's J614c/J616c tree enabled
/display-subsystem itself. The kernel tree now keeps those nodes
disabled for the handoff, as the J616s tree does (linux#1332).

Serve both M4 Max laptops from the one T6041 entry: its board list
becomes apple,j616c and apple,j614c, renamed dcp_soc_t6041_max. Chip,
firmware, addresses and SIDs stay as they were; nothing changes for any
other board or chip.

The host test runs the J616C scenarios for the J614C too (a boot
publishes, the external memories come from the ADT, a J614C tree on a
T6040 does nothing), checks the external regions of both M4 Max
scenarios in full, and adds the refusal the J614C hit: an enabled
display subsystem publishes nothing.

Link: aurora-silicon/linux#1332
Signed-off-by: Diljit Singh <hey@djdev.me>
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.

1 participant