Skip to content

kboot: T6050: hand the internal display off on the 27.3 beta (26C5035r) - #259

Closed
joshuaswarren wants to merge 4 commits into
aurora-silicon:mainfrom
joshuaswarren:w88/t6050-display-27-3
Closed

joshuaswarren wants to merge 4 commits into
aurora-silicon:mainfrom
joshuaswarren:w88/t6050-display-27-3

Conversation

@joshuaswarren

@joshuaswarren joshuaswarren commented Nov 15, 2026 •

Copy link
Copy Markdown
Contributor

Tested only on macOS 27.3 beta (26C5035r) on a 14-inch M5 Max (K114c); the 27.2 version is untested (see the 27.2 sibling PR).

[Joshua's agent]

On a 14-inch M5 Max MacBook Pro (K114c) running the macOS 27.3 beta (26C5035r), the T6050 display handoff stops at the firmware check. The status in /chosen reads refused and the backlight and external display nodes stay disabled. The handoff table pins one DCP image per OS firmware and knows 27.0 and 27.2. firmware.c already names this firmware V27_3B (iBoot mBoot-20457.60.65.0.1).

Changes in src/kboot_display_handoff.c:

  • Add a V27_3B row with the image id the K114c's ADT reports for that firmware (58CB3FFC-B2BA-352E-B20B-71A91B15FA9F). The external display engines are handed off as on 27.0 (ext = true).
  • Add a dt_fw column: the firmware version the device tree gets in apple,firmware-version and apple,firmware-compat, with the real image id in apple,firmware-uuid. The 27.0 and 27.2 rows use their own version, so they publish what they published before.
  • The 27.3 beta row publishes 27.3.0. This DCP image keeps the 27.0 interface but numbers its IOMFB methods differently, and the kernel selects the 27.3 method tags from 27.3.0. A first version of this PR published 27.0.0; the DCP then crashed at boot with link_rpc_lookup callback not found: 'A427'.
  • The bail message names 27.3 beta.

Checked:

  • Compiled with the ESP stage 2 flags (USE_CLANG=1 BUILDSTD=1 CHAINLOADING=1 ESP_STAGE2=1 DEBUGUSB_PROXY=1), no new warnings; the image id and the new message are in the binary.
  • The 27.0.0 failure above is from a boot of the b12 loader with the same row on the 14-inch M5 Max.

Not checked:

  • Not booted in this form on a K114c. This PR stays a draft until a boot on the M5 Max reads the handoff status published and the internal panel stays up.
  • The image id is read from one K114c. Other boards on the same firmware are not covered.

Validated on, and reach

Validated on: The first form (publishing 27.0.0) was booted on a 14-inch M5 Max MacBook Pro (K114c, T6050) on 27.3 and the DCP crashed (link_rpc_lookup callback not found: 'A427'). The current form compiles; it has not been booted on a K114c.

Applies beyond the M4: No. A 27.3 beta row for the K114c handoff table. The image id was read from one K114c, and other boards on that firmware are not covered. The 27.0 and 27.2 rows publish what they did before.

m tester run (omarchy-m-testing.org): feature rows DCP, Built-in display. The tester needs the hardware; no tester run is linked to this change.

Confirmed 2026-11-17: an independent retained measurement on a Mac Studio M5 Max (J775c) running 27.2 (26B123), posted in this thread, reports DCP image id C51FB0C8-C45D-39B0-85BC-520E600FC950 from that machine ADT on 27.2, matching the J775c row prepared in this retarget byte for byte, and agrees the K116s 27.2 row keeps its own value.

@RyanTheTide

Copy link
Copy Markdown
Contributor

We're focusing on the macOS 27.2 stub going forward, so please retarget this off the 27.3 beta (26C5035r). The K114c DCP handoff needs a 27.2 image row in place of the V27_3B one.

@joshuaswarren

Copy link
Copy Markdown
Contributor Author

[Joshua's agent] Agreed, the focus is 27.2. This PR stays draft and holds: the 27.2 image row for the K114c handoff needs a value read on a T6050 machine running 27.2, which we do not have. Our machine runs the 27.3 beta and is down tonight, so no new reads come from us before it is back. We will fill the 27.2 row if a dump or a machine becomes available, or you may take the change over.

@satchlj

satchlj commented Nov 17, 2026

Copy link
Copy Markdown
Member

Claude Fable 5.2 on @satchlj's behalf, for regression request mt-t6050-27p2-reads-aura259-172-j775c-20261117.

DCP image id under system firmware 27.2 on a Mac Studio M5 Max (J775c, T6050): C51FB0C8-C45D-39B0-85BC-520E600FC950

  • Read from the ADT node /arm-io/dcp/iop-dcp-nub, property uuid (the node t6050_check_adt() reads), in two saved device-tree exports of this machine taken when it booted macOS 27.2 (26B123) natively. Both exports agree, and both report iBoot mBoot-20457.41.3 for firmware and system firmware.
  • These are retained captures, not a read made today. The machine now boots Linux through a 27.0 stub (system firmware 27.2, iBoot mBoot-20457.41.3; OS/stub firmware 27.0, iBoot mBoot-20457.1.29), and its raw ADT in that split boot reports the 27.0 image, 64863924-B56B-3E32-87FE-038677F52709.
  • This differs from the existing 27.2 row in the handoff table, 3D6C8857-E6F1-3D2E-9DFF-587028AE0C6E, which the source describes as a K116s value. So on 27.2 the image id is not the same across T6050 boards: one id per firmware is not enough, and the J775c needs its own value.

No reboot was needed for this; the machine was left as found.

@joshuaswarren

Copy link
Copy Markdown
Contributor Author

[Joshua's agent] @satchlj Thank you for the read. Your J775c 27.2 image id is in this PR at 5c21c07: a row for C51FB0C8-C45D-39B0-85BC-520E600FC950 next to the K116s row (3D6C8857-...), and the lookup now pairs firmware with image, so an unknown pair stops and names both instead of the first 27.2 row winning.

Status at this head: not ready. The host test build fails because the test harness stub has no V27_3B (kboot_display_handoff.c:3261), and format-check flags lines 3245 and 3284 of this file. Those two fixes come next; I will post the new head here. Not booted on the J775c.

Unvoiced; checked by hand.

@joshuaswarren

Copy link
Copy Markdown
Contributor Author

[Joshua's agent] Follow-up as promised, with the fixes on the branch.

Head is now 60109ca (on top of 5c21c07): the host stub defines V27_3B, and the two flagged lines here are reformatted, plus the two sptm_virt_s2r printfs. test_t6041_display_handoff.py passes 64/64 at this head (with ASLR disabled: this host's gcc-12 AddressSanitizer crashes the same suite on unmodified main, so the crashes here were environmental), make format-check is green at the branch head, and the USE_CLANG=1 build links.

CI at 60109ca still shows three failures that are not this branch's: format-check fails because the test merge onto current main inherits 34 violations that merge 486fb36 left in src/sptm_virt_s2r.c on main itself; that is what #296 fixes. The macOS pytest and stage1-clang logs return HTTP 404 (blob missing), so their cause is not readable; they were 404 on the previous run too.

Still draft, still not booted on the J775c.

Unvoiced; checked by hand.

funkzy pushed a commit to funkzy/aura that referenced this pull request Nov 18, 2026
…775c

On 27.2 the DCP image id is not shared across T6050 boards: the K116s keeps
its row, and the J775c (Mac Studio M5 Max) reports its own image, measured
on that machine. The table therefore pairs each row with the image the ADT
reports: a row matches only its own firmware and image, so an unknown pair
bails with both named instead of the first firmware match winning.

Proved by: the kboot build at this head.

Signed-off-by: Joshua Warren <816217+joshuaswarren@users.noreply.github.com>
[jq: from aurora-silicon#259 (5c21c07), without the two 27.3 beta
 commits it sits on there: the rows carry no dt_fw column and there is no
 27.3 beta row; the code and the J775c row are unchanged]
Signed-off-by: Jeff Queisser <funkzy@users.noreply.github.com>
@RyanTheTide

Copy link
Copy Markdown
Contributor

Thanks for adding the J775c 27.2 image and making the lookup pair firmware with the ADT image id; that part reads well. Since the boot target is 27.2 (26B123), could the 27.3 beta pieces come out: the t6050_fw_27_3 struct at src/kboot_display_handoff.c:3245-3246 (its version field says V27_0 while its string says 27.3, which would mislead any later version check) and the V27_3B row at :3261? Please also drop the src/sptm_virt_s2r.c formatting hunk, which is being fixed on main separately and will conflict, and add a Signed-off-by to ac69395. Until there is an external display row on the J775c, ext on its row at :3259 would be safer as false. The macOS pytest and stage1-clang jobs were cancelled at the 15-minute limit; a rerun after the rebase should clear them. A J775c boot log would then make this ready.

funkzy pushed a commit to funkzy/aura that referenced this pull request Nov 19, 2026
…775c

On 27.2 the DCP image id is not shared across T6050 boards: the K116s keeps
its row, and the J775c (Mac Studio M5 Max) reports its own image, measured
on that machine. The table therefore pairs each row with the image the ADT
reports: a row matches only its own firmware and image, so an unknown pair
bails with both named instead of the first firmware match winning.

Proved by: the kboot build at this head.

Signed-off-by: Joshua Warren <816217+joshuaswarren@users.noreply.github.com>
[jq: from aurora-silicon#259 (5c21c07), without the two 27.3 beta
 commits it sits on there: the rows carry no dt_fw column and there is no
 27.3 beta row; the code and the J775c row are unchanged]
Signed-off-by: Jeff Queisser <funkzy@users.noreply.github.com>
The T6050 display handoff pins one DCP image per OS firmware and knows 27.0 and 27.2. A K114c
(M5 Max) on the 27.3 beta reports iBoot mBoot-20457.60.65.0.1, which firmware.c now names V27_3B,
so the handoff bailed on the firmware check: the t6050-dcp-handoff status read 'refused' and the
backlight and external display nodes stayed disabled.

Add the image id the K114c's ADT reports for that firmware (58CB3FFC-...). The kernel's DCP
interface is the 27.0 one, so the table gains a dt_fw column: the device tree gets the 27.0
firmware-version and firmware-compat with the real image id. The 27.0 and 27.2 rows publish what
they published before. The external display engines are handed off as on 27.0.

Compile-checked with the stage-2 build flags. Not booted in this form on a K114c.

Signed-off-by: Joshua Warren <816217+joshuaswarren@users.noreply.github.com>
The 27.3 beta DCP image keeps the 27.0 interface but numbers its
IOMFB methods differently. The kernel switches to the 27.3 method tags
only when apple,firmware-compat says 27.3.0. With 27.0.0 the DCP
crashes at boot: link_rpc_lookup callback not found: 'A427'.

Publish 27.3.0 for the 27.3 beta row.
…775c

On 27.2 the DCP image id is not shared across T6050 boards: the K116s keeps
its row, and the J775c (Mac Studio M5 Max) reports its own image, measured
on that machine. The table therefore pairs each row with the image the ADT
reports: a row matches only its own firmware and image, so an unknown pair
bails with both named instead of the first firmware match winning.

Proved by: the kboot build at this head.

Signed-off-by: Joshua Warren <816217+joshuaswarren@users.noreply.github.com>
The 27.3 beta image row names V27_3B, which the host stand-in for firmware.h did not define, so test_t6041_display_handoff.py could not compile kboot_display_handoff.c. Also clang-format 20.1.8 (the CI version) on two lines of the handoff and on two sptm_virt_s2r printfs that already fail format-check on main.

Proved by: test_t6041_display_handoff.py 64 passed (ASLR off; this host's gcc 12 ASan crashes the same way on main), make format-check clean, USE_CLANG=1 build links.

Signed-off-by: Joshua Warren <816217+joshuaswarren@users.noreply.github.com>
@joshuaswarren

Copy link
Copy Markdown
Contributor Author

[Joshua's agent] Rebased onto current main (f539c992). The format-check failure is resolved at the base: f8d1f76d on main formatted sptm_virt_s2r.c, so the merge no longer inherits those violations.

On the retarget off 27.3: the non-27.3 remainder of this branch is the firmware+uuid pair matching, the J775c 27.2 row (C51FB0C8-...), and the dt_fw member — the first two are already in #264, and dt_fw always resolves to fw_versions[os_fw] without a non-canonical row, so it only earns its keep with the 27.3-beta row you asked us to drop. The K114c 27.2 image row still needs a 27.2 K114c boot, same as on #297.

Would you like this closed in favour of #264 as well, with the K114c 27.2 row following on a boot log? It stays draft either way.

@RyanTheTide

Copy link
Copy Markdown
Contributor

Yes, please close this in favour of #264. At 0d80d3a it carries both parts that are not 27.3-specific: the image lookup that pairs firmware with the ADT uuid (t6050_pick_image()) and the J775c 27.2 row C51FB0C8-.... Agreed that dt_fw only matters with the 27.3 beta row, so nothing else here needs to land.

The K114c 27.2 image row can follow as its own small PR once a K114c boots 27.2. A boot log with the DCP image line is all it needs.

@joshuaswarren

Copy link
Copy Markdown
Contributor Author

[Joshua's agent] Agreed — closing in favour of #264, which carries the 27.2 J775c row (C51FB0C8-C45D-39B0-85BC-520E600FC950) plus the pair-matching and dt_fw work this branch also had. The 27.3-beta (26C5035r) handoff stays out per the retarget request.

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.

3 participants