Skip to content

arm64: dts: apple: MacBook Air 15-inch M5 (J815) device tree - #1532

Merged
DjDeveloperr merged 27 commits into
aurora-silicon:integration/aurorafrom
m0d16l14n1:j815/dts
Nov 17, 2026
Merged

DjDeveloperr merged 27 commits into
aurora-silicon:integration/aurorafrom
m0d16l14n1:j815/dts

Conversation

@m0d16l14n1

@m0d16l14n1 m0d16l14n1 commented Nov 11, 2026 •

Copy link
Copy Markdown
Contributor

This shares the T8142 SoC description between J813 and J815 and adds J815 board resources for power/PMP, USB, display/GPU, N1, AOP/I2C/ALS, SMC and SEP/Touch ID. It preserves the J813 factory-address and ALS updates and @gig3m's PMU/MCC/AVE additions. J815 sound remains a separate follow-up.

Ready for review at 8e1ffb8178dd6ee873329b9a20cd0f2f56e3f1eb (tree afc567aa9c18789232be1e970585930d33a1c827). Merge is held for compatible-loader qualification/adoption and fresh required CI.

The shared display file now provides canonical disp0-piodma. J813 also keeps legacy disp0_piodma in its board file for earlier loaders; J815 uses only the canonical alias. Both spellings on J813 resolve to the same PIODMA node. A bounded source-property comparison against c93ee625 confirms these are the only effective board-property changes; labels and other properties are unchanged. This check does not generate DTBs or establish hardware operation.

Validation and landing gates:

  • Current-head Host CI 43680006646 passes 321/321 runs. DT CI 43680006701 builds 137 baseline and 137 head DTBs, both make return 0, with zero new warnings (1,482→1,465; 17 removed). All seven changed bindings and ADT's 19 tests/six coverage comparisons pass. Hosted testing used merge e419a0af against integration fc050503. Commit 8e1ffb81 fixes the strict T6030 ALS fallback-chain schema; the alias and ALS failures are closed. Earlier evidence retains its recorded scope.
  • aurora-silicon/aura#165 at 4b066090 supplies the required T8142 canonical-alias loader behavior. The author has deployed a Stage 2 containing it and a canonical-only J815 DT; the boot result is pending. That capability must be qualified and adopted before this J815 DT lands. The ANS fix in pcie, nvme: T8142: power ANS off and back through its storage gates aura#250 is independent and does not itself imply alias support.
  • The author's pinned J815 report covers a 27.2/26B123 stub and a composed kernel/loader tree using Plasma 6/KWin with ARGB output and no plane-format workaround. The new canonical-only boot retains its own qualification gate.
  • The official J815 installer path still needs Admit the M5 MacBook Air 15-inch (J815) on the macOS 27.2 (26B123) stub aurora-installer#144 and its loader/kernel release prerequisites. The reported private recipe is not a general recommendation to update firmware or enrol a custom Stage 1.

Audio #1534 and the other stacked feature PRs remain parent-first follow-ups with their own qualification requirements. SEP policy/guards and per-device calibration requirements remain intact.

Original bring-up: @m0d16l14n1. J813 patches and owner reports: @gig3m. Integration and review on @DjDeveloperr's behalf assisted by Codex. Earlier detailed validation remains in the PR's review timeline.

J704 and J815 describe their boards on top of the T8142 inventory in
t8142.dtsi, but J813 still defines its own root, CPUs, timer, AIC and
UARTs, and its fragments define the pmgr and DART nodes that the
inventory also has. A board that includes t8142.dtsi therefore cannot
use the J813 fragments: the labels are defined twice. That blocks the
J815, which shares most of them.

Include t8142.dtsi in J813 and enable what it uses, as J815 does. Turn
the fragments' pmgr, AOP, AVD, DCP, display, PCIe, PMP and MTP DART
definitions into references that keep J813's values. Enable the two USB
DARTs, which the inventory disables at the same paths.

The flattened J813 tree is unchanged apart from explicit
status = "okay", interrupt-parent now inherited from the root, and the
inventory's extra USB DART labels: compared with dtc -I dts -O dts -s
before and after, with phandle numbers removed.

Not tested on J813 hardware.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
J815 has J813's PMGR power domains, PMP, QSPI and FPWM. Include
t8142-j813-peripherals.dtsi, which J813 now builds on t8142.dtsi, in
place of the board's own PMGR and MTP DART enables. The PMP takes the
J815 board-id from the ADT (0x26).

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set
in a J815 tree that included the whole J813 board.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
The J815 ADT describes both Type-C ports like J813's: ATC PHYs, DWC3
controllers, DARTs and the CD3218B controllers on the same buses.
Include t8142-j813-usb.dtsi.

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set in
a J815 tree that included the whole J813 board. A SuperSpeed Plus hub
and the USB 2/3 devices behind it enumerate, and USB-C headphones play
audio.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
Include t8142-j813-display.dtsi and describe the J815 panel: 15.3-inch,
2880x1864 (2880x1800 below the notch), 326 x 211 mm.

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set in
a J815 tree that included the whole J813 board. With the macOS 27.0 DCP
firmware this needs the FW27 display support in Linux aurora-silicon#1413; the console
and a KWin session run on the panel.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
The J815 GPU is J813's. Include t8142-j813-gpu.dtsi.

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set in
a J815 tree that included the whole J813 board. With the macOS 27.0
firmware this needs Linux aurora-silicon#1413; the GPU driver initializes and KWin
renders through Mesa.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
Include t8142-j813-avd.dtsi: J815 has J813's AVD, DART and power domain.

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set in
a J815 tree that included the whole J813 board. With avd-fw-v5-t2 the
decoder probes and decodes through v4l2-request/VA-API: H.264 High 1080p
and HEVC Main 720p in mpv (vaapi-copy, one AVD interrupt per frame, no
kernel errors).

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
Include t8142-j813-isp.dtsi. The J815 ADT gives the ISP sensor-type 0x72
(J813: 0x71).

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set in
a J815 tree that included the whole J813 board. The ISP firmware boots
and the camera streams.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
Include t8142-j813-n1.dtsi for the N1 on apcie port 0. Describe the N1's
three PCI functions and alias wifi0 and bluetooth0 to them, so the
loader can fill in the factory MAC and BD addresses, as on T6050.

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set in
a J815 tree that included the whole J813 board. Wi-Fi connects to WPA2
networks, and Bluetooth LE devices pair and reconnect. Both use the
factory addresses.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
Include t8142-j813-audio.dtsi for the AOP, the AP and NUB pin
controllers, the ADMACs and the I2C buses with the headphone jack codec
and speaker amplifiers. J815 has six speaker amplifiers where J813 has
four, so keep J813's sound card and the AOP audio service disabled.
The J815 speaker layout is a separate change.

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set in
a J815 tree that included the whole J813 board, with audio enabled. The
AOP boots and its services come up. This exact variant, with audio
disabled, has been compile-tested only.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
The J815's AOP "als" service is an ST VD6286, as on the M4 Pro, not
T8140's CT817. Give the service device its compatible so aop_als binds
and sends the per-device calibration.

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set in
a J815 tree that included the whole J813 board. The sensor reports lux
and colour temperature.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
The discovered SMC temperatures carry no power rails and no SoC die
zone. Add the total system, AC input and 3.8 V rail power, the AC and
battery voltages and the AC current, and J704's T8142 die key list for
the SoC die thermal zone.

Tested on J815 (MacBook Air 15-inch, M5) with the equivalent node set
in a J815 tree that included the whole J813 board.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
From the J815 ADT (macOS 27.0): the SEP is a pre-booted
iop-sep,ascwrap-v6 at 0x30a600000 with its ASC mailbox at +0x8000. Its
t8110 DART uses the BIT(40) input aperture, as on T8132. The Touch ID
sensor is on spi2 (0x3a5108000, 8 MHz, CS0, power domain SPI2 under
SPI_P), with power on NUB GPIO 36 and data-ready on AP GPIO 200. Alias
the SEP so the loader can reserve its firmware and copy the boot
manifests onto the node, and expose the two sensor pins.

The SEP and its DART stay disabled: the SEP driver enables them.

Tested on J815 with the J815 SEP profile: the SEP attaches, the sensor
binds on spi2 and comes up in about 40 ms, and fingerprints enrol and
match (sudo with the finger).

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>

@DjDeveloperr DjDeveloperr left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Codex (GPT-6.1 Sol), reviewing on @DjDeveloperr’s behalf.

Keep draft: the J815 results do not close the explicit J813 boot gate for extracting/reworking the shared DT. Brightness/camera/USB3/backlight/decoder are separate outcomes and a source merge cannot blanket-qualify them. CI approval is now unblocked so schema/build results can be evaluated; an owner-approved J813 regression row is still needed for the shared change. No new hardware or media upload is requested here.

Match the existing DT and public driver interfaces: keep the specific fallback chains, T8142-only ALS, eight ISP power domains versus fourteen on T8140, and the two-mailbox/six-region T8142 GPU profile. Preserve the constraints on other SoCs. No DT resource values change.

Validated with dtschema 2026.9 strict document/reference checks and targeted processed-schema positive and negative cases. Full DTB validation and the draft J813 hardware gate remain separate.

Signed-off-by: Diljit Singh <hey@djdev.me>
Carry byte-exact schema files from reviewed aurora-silicon#1423 at 18fa06f so the shared J813/J815 DT can be checked without inventing a second binding implementation.

Original binding updates by Thomas Glanzmann, signed off in commits 9afc945 and bb9ad0e. This is binding metadata only; no driver or speaker-protection adoption or parent edge is added. The M5 merge lane retains ownership of aurora-silicon#1423. Later identical binding changes coalesce in an ordinary merge.

Signed-off-by: Diljit Singh <hey@djdev.me>
Identify the J815 native profile with a T8142-specific compatible while retaining the PCI1900 fallback. The generic driver continues to match the host/PCI ID and obtain firmware data from its root board table; no firmware/platform/protocol values are invented. Keep all three wire27 properties required for the older PCI-only T6050 profile and forbid mixing those properties into the native profile.

Only the J815 endpoint compatible changes. Shared SoC/J813 DT inputs and runtime resources remain unchanged. The J815 board firmware-table prerequisite is aurora-silicon#1530.

Signed-off-by: Diljit Singh <hey@djdev.me>

@DjDeveloperr DjDeveloperr left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Codex (GPT-6.1 Sol), reviewing on @DjDeveloperr’s behalf.

The DT retry actually ran: baseline412 warning lines versus head442, with30 new J815 warnings. Both DT builds returned0, and ADT coverage passed its19 tests and six board comparisons. This was a real schema/profile failure, not the earlier apt setup problem.

I pushed three ancestry-preserving follow-ups at ec064e9:

  • Narrow T8142 I2C/SPI/pinctrl/setup-mailbox/ALS descriptions, the eight-domain ISP profile versus T8140’s fourteen, and a dedicated two-mailbox/six-region GPU profile. Other profile constraints are retained.
  • Thomas Glanzmann’s byte-exact macaudio/TAS schema prerequisites from #1423 (original commits9afc945d0cdb and bb9ad0e), with attribution. No driver adoption or partial merge-parent edge is implied, and #1423’s branch is untouched.
  • An explicit native T8142 Centauri binding profile with the PCI1900 fallback on J815 only. The generic driver still selects by host/PCI ID and its board table (#1530); no firmware/platform/protocol values were guessed. The older T6050 PCI-only profile retains all three wire27 requirements.

Validation: all eight changed primary schemas pass strict dtschema2026.9 document/reference checks;45 targeted processed-profile controls pass, including reproduced baseline rejections and strict negative cases. The native N1 binding has15 passing positive/negative controls; the exact audio prerequisites have document/profile controls. The final diff changes11 schema files and J815’s endpoint-compatible tuple only: J813/common DTS, drivers, resources, aliases and tests are unchanged. This is not a full new CHECK_DTBS or physical boot result; fresh CI is required.

Keep draft/unmerged. The remaining alias warning is a real producer/consumer contract: current T8142 aura hardcodes disp0_piodma. Draft aura#165 supplies canonical disp0-piodma plus legacy fallback, with production/libfdt regression controls; a J815 alias rename must wait for compatible-loader adoption and qualification. The explicit J813 shared-DT boot gate remains open, and the original J815 report is not an exact-new-head boot. No hardware, playback, loader/default or image action was performed.

Preserve the current shared T8142 CPU topology, OPP power data and
trackpad actuator, along with the J813 firmware clock providers.
Reconcile T8142 binding profiles with the current T6050 sensor services
and preserve the T8140 ISP firmware memory and sensor power domains.

@DjDeveloperr DjDeveloperr left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Codex (GPT-6.1 Sol), reviewing on @DjDeveloperr’s behalf.

Resolved the current-main conflicts in b180163 with a normal integration merge. The resolution preserves the firmware-27 DCP provider clocks, shared CPU/OPP/actuator data, T6050 mailbox/ALS/LAS profiles, and main’s ISP heap/text/data memory ABI and 14–16 power-domain profile, alongside T8142’s distinct eight-domain profile. Review caught and corrected an initial candidate that dropped the optional ISP text/data regions before publication.

All 11 schema documents, 65 targeted ABI controls, 15 full Centauri controls and 10 audio controls pass. The symbolic J813 source comparison preserves 307 nodes and 1,377 explicit properties, with redundant interrupt-parent entries inherited identically. This does not establish compiled-DTB or physical boot equivalence.

Keep draft: fresh integrated-head host/DT CI and the author-required J813 shared-extraction boot remain gates. The existing display alias remains intact; aura#165 must be qualified/adopted before migration. The earlier EC DT log is unavailable (server errors), so I am not claiming a verified remaining-warning count.

Match the existing J813/J815 fw-data crash-log region and its ordered
seven-region tuple. Preserve the six-region maximum for other GPUs.

@DjDeveloperr DjDeveloperr left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Codex (GPT-6.1 Sol), reviewing on @DjDeveloperr’s behalf.

Corrected the T8142 GPU binding in 9122abe. My earlier strict profile omitted the existing seventh fw-data crash-log region. The actual current J813/J815 tuple reproduces the old failure; the corrected ordered tuple passes 32 controls, with 768 comparisons confirming unchanged acceptance across 16 non-T8142 compatible profiles and strict meta/reference validation passing. Other profiles retain max six regions. No DT/GPU driver/firmware resource value changed.

Concrete compiled evidence comes from #1534’s ea5ab50 run: tools installed, base/head make succeeded with 137 DTBs, base400/head395 warning lines, five new warnings (four GPU cardinality, one J815 underscore alias), ten removed; binding docs and ADT coverage passed. Fresh CI must confirm the new schema result.

The legacy alias remains intentionally unchanged pending aura165 hardware qualification and compatible-loader adoption. Keep draft and the J813 boot gate; no warning filter, premature alias rename or new hardware action.

@DjDeveloperr DjDeveloperr left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Codex (GPT-6.1 Sol), reviewing on @DjDeveloperr’s behalf.

Current-head CI has now executed: Host run 43107102188 passed its actual tests. DT run 43107102060 built all 137 DTBs successfully on both base and head; warnings fell from 400 to 391 (10 removed, one new). The corrected seven-region T8142 GPU schema passed binding validation, and all four earlier GPU warnings are gone.

The sole new warning is t8142-j815.dtb aliases/disp0_piodma, which violates the standard alias-name regex. Keep the loader compatibility gate intact: a canonical alias migration must wait for the affected-hardware boot/adoption qualification of aura#165; do not rename the alias or filter this warning merely to turn CI green. ADT coverage passed. This remains draft; the existing physical/loader qualification requirements are unchanged.

Keep main's complete ISP binding, preserving the T8140/Neo sensor-domain
order and T8142 eight-domain case. All added and removed feature lines
in the remaining 15 DT and binding files match the preceding parent.

The complete bindings directory matches the validated aurora-silicon#1534 integration:
eight binding documents and 32 candidate/32 main GPU schema controls.
No repeated schema test or new compiled-DTB/hardware result is claimed.
J815 audio stays disabled here. Keep the underscore loader alias until
compatible-loader qualification/adoption; no warning filtering.

Preserve published ancestry and require fresh current-head Host/DT CI.
The required J813 shared-DT boot and existing loader/rollout gates remain.
No boot, audio playback, image build or protection-policy change.

Assisted-by: LLM

@DjDeveloperr DjDeveloperr left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Resolved the current-main conflict as normal merge 97f1472, parents 9122abe/main68830200; no history rewrite. The ISP binding uses main's full version, preserving the Neo sensor-domain order and T8142 eight-domain case, so that already-landed binding disappears from the residual. All added/removed feature lines in the remaining 15 DT/binding files match the preceding head.

The complete bindings directory is identical to the validated #1534 c83c100 tree: eight document metaschema/reference checks and 32 candidate/32 main GPU controls pass on those identical inputs. No repeated schema test, new DTB compile or hardware boot is claimed. The parent still disables J815 sound/AOP audio, and the underscore loader alias is retained without warning filtering. #1534's separate audio DT feature remains unchanged; no duplicate source/audio build or child branch update is needed for this source-preserving parent integration.

Keep draft for fresh Host/DT CI, Aura #165 affected-loader boot and compatible-loader adoption before alias migration, and the author-required J813 shared-DT boot. The original J815 hardware report retains its pre-integration scope; no new machine or hardware action is requested.

m0d16l14n1 added a commit to m0d16l14n1/linux that referenced this pull request Nov 14, 2026
…de-wip

The selftest mocks gain aurora-silicon#1532's cio_sleep next to this branch's T8142
DisplayPort tunables.

Signed-off-by: Nick <nekreshen@yandex.ru>
@scottjones

Copy link
Copy Markdown
Contributor

Hi, thanks for putting this together. I can help test on J815.

I have a 15" M5 Air (Mac17,4, J815/T8142, 32 GB / 1 TB) that's otherwise unused, so I can wipe it freely. It's on macOS 26.5.1 (25F80), system firmware 18000.120.36, with an old partial install I'd replace.

Which macOS / stub should a J815 tester use to match your results? The PR says the original report ran on a 27.0 stub with Aura Stage 2. My reading, which may be wrong: aurora-installer main only offers J815 with a 26.6.2 (25G83) stub needing firmware 18000.161.10+, and no 27.x entry lists J815. The M5 GPU backend looks profiled on 25G76 (26.6.1), and DCP 27.0 is only selected on an exact 27.0.0 compat.

So should I update to 26.6.2, or wait for a 27.x path? Firmware can't go back, so I'd like to know the target first.

Once it's on the right firmware, I'll boot this PR's head and report on the panel, the disp0_piodma alias, USB, Wi-Fi/BT and Touch ID, leaving sound (#1534) out until there's a J815 speaker profile.

DjDeveloperr commented Nov 14, 2026 •

Copy link
Copy Markdown
Contributor

@scottjones thanks—please hold the firmware update for now. Your reading of the current installer matrix is correct: installer main at 5fc3946a offers J815 a 26.6.2/25G83 stub with the 18000.161.10 minimum; its 27.x entries do not list J815. That is an offered installation path, not evidence that this PR's full J815 GPU/display combination is qualified on it.

The system firmware, installed stub, and GPU firmware actually admitted by the loader are separate identities. Clarification after @gig3m's follow-up: the dedicated M5 backend requires M5-RTKit-3255.160.4.release (25G76), but it is not the only T8142 route. This PR's probe_m5() selects the shared G17 path when that loader profile is absent; its T8142 config uses V26A428_M5. My earlier wording omitted this route. The remaining target question is which complete installer/loader/kernel combination is qualified on J815.

@m0d16l14n1 could you pin the known-working J815 combination—system firmware, exact stub build, Stage 1/Stage 2 commits, and the admitted GPU profile/display compat—and whether it is reproducible with the current installer? The older “27.0 stub” report alone does not answer that.

This PR remains draft at c93ee6255635f78df6fa74d609279cc2e6c7343d. Panel, USB, N1 Wi-Fi/BT and read-only SEP attach would be useful first rows once the target is pinned; sound remains out. The disp0_piodma alias stays unchanged until the separate aura #165 loader qualification/adoption is complete. No wipe or firmware upgrade is needed merely to respond to this request.

@gig3m

gig3m commented Nov 14, 2026

Copy link
Copy Markdown
Contributor

J813 data point, in case it helps pin the J815 target (same SoC, I have no J815).

What my J813 (Mac17,3) is running today, with GPU acceleration and the panel working:

  • system firmware: 27.2 (iBoot stage 1 mBoot-20457.41.3, macOS 27.2.1 on the main volume)
  • stub: 27.0 (iBoot stage 2 mBoot-20457.1.29)
  • GPU: GPU firmware: 27.0 (26A428), G17G
  • DCP: firmware 27.0
  • Stage 1: aura 1.0.0b8

Not reproducible with the installer as offered: alpha.19 gave J813 the 26.6.2 stub only, and I moved the stub to 27.0 by hand. On the 26.6.2 stub the M5 backend came up (M5-RTKit-3255.160.4.release, 25G76) but the first compositor submission failed for me under Hyprland; the 27.0 stub is where the desktop renders on the GPU.

Assisted-by: Claude Fable 5.2

Copy link
Copy Markdown
Contributor

@gig3m this is useful: it pins a working J813 combination with 27.2 system firmware, a manually installed 27.0 stub, 26A428 GPU firmware and 27.0 DCP. It also records a compositor-submission failure on your earlier 26.6.2 setup, beyond successful backend admission.

One correction to my earlier reply: the 25G76 gate belongs to the dedicated M5 backend. This PR already has an explicit fallback in probe_m5() to the shared G17 backend when that profile property is absent, with T8142's V26A428_M5 configuration. That is consistent with your 27.0 result; I've clarified my earlier comment.

Could you add the exact working kernel commit/composition and Stage 2 commit, plus the corresponding pins for the 26.6.2 failure if retained? Version strings and redacted error summaries are enough. That will let us distinguish the two runs without assuming a firmware-version cause.

@scottjones the installer-target question remains open: this was a manual J813 stub migration, and @gig3m has no J815. Please keep your current firmware until the J815 author confirms the reproducible target. The separate display-alias gate and sound exclusion remain.

@gig3m

gig3m commented Nov 14, 2026

Copy link
Copy Markdown
Contributor

Pins for both runs. All J813 (Mac17,3), Stage 1 aura 1.0.0b8, system firmware as above.

27.0 stub, GPU works

Daily kernel until this afternoon:

Running now:

  • m0d16l14n1/linux j815/test-next d2737f1 + the same plane patch + J813 copies of its J815 sleep/DCP nodes
  • Stage 2: m0d16l14n1/aura j815/b13-sleep-hib 472f6755 (1.0.0b13-g472f6755)

The second logs GPU firmware: 27.0 (26A428), G17G. The first prints no build line, only M5 firmware graph ready, on the same stub.

26.6.2 stub, failure (11 Nov, from my notes; nothing retained on disk)

  • kernel: integration/aurora 84d091f or 5cfcc01 (both ran that day, my notes don't say which was up), alarm-aurora config with DRM_ASAHI_M5, plus the same plane patch
  • Stage 2: the one installer alpha.19 wrote, with a patched U-Boot; commit not recorded
  • userspace: mesa-aurora 26.3.0_devel 20261104, Hyprland 0.57.2
  • backend admitted: M5: GPU firmware 25G76 running
  • error at Hyprland's first submission: GPU work of a client failed (EIO, m5/live_boot.rs:435), then the device stayed lost (ENODEV) until reboot

I moved to the 27.0 stub the next day and did not go back, so I can't say the stub was the cause.

Assisted-by: Claude Fable 5.2

Copy link
Copy Markdown
Contributor

Recorded, thank you. The working J813 evidence now has two source compositions: the d89d8c0c integration plus the listed PRs with Stage 2 2175c483 + b97ef3a8, and the current d2737f1d test-next composition with Stage 2 472f6755. Both include the local XRGB8888/XBGR8888 primary-plane change; the current composition also carries the J813 sleep/DCP node copies.

The older 25G76 submission failure retains the uncertainty you stated: kernel 84d091f7 or 5cfcc01a, unrecorded Stage 2 commit, patched U-Boot, and no retained logs. Kernel and loader changes accompany the stub change, so this comparison does not isolate the stub as the cause.

Could you point to a public patch/PR for the plane-format change, if one exists? It is part of the working compositor recipe and needs to be accounted for in a reproducible test image.

@scottjones @m0d16l14n1 the J815 target remains awaiting its author-confirmed installer/kernel/loader recipe. These results qualify the reported J813 compositions; the manual 27.0 path and local patches still need to be reconciled with what the installer offers. Existing alias and sound gates remain.

@gig3m

gig3m commented Nov 14, 2026

Copy link
Copy Markdown
Contributor

No public PR. It is a local workaround, and not one I'd propose as is: it puts back what 5e98b68 and 43f8980 took out.

Those commits offer the H17G primary plane only alpha formats because the firmware blends XRGB's X byte as alpha. Hyprland 0.57 only allocates XR24/XB24 for the output, so without XRGB it aborts at startup ("format XR24 isn't supported by primary backend"). The patch adds XRGB8888/XBGR8888 back to the overlay list for that plane. It is correct only for a compositor that writes an opaque X byte, which Hyprland's renderer does; a client like the alacritty-under-sway case in 43f8980 would show wrong colours again.

For a reproducible image that means either this patch with Hyprland, or no patch and a compositor that takes ARGB.

the patch (on drivers/gpu/drm/apple/plane.c)
@@ -699,6 +699,34 @@ static const u32 dcp_primary_formats_h17p[] = {
+/*
+ * H17G primary plane (the firmware blends X as alpha): the overlay list plus
+ * XRGB8888/XBGR8888. Compositors such as Hyprland 0.57 only offer the X
+ * formats for the output swapchain and fail without them; their renderers
+ * write an opaque X byte for an opaque output.
+ */
+static const u32 dcp_primary_formats_h17g[] = {
+	DRM_FORMAT_ABGR16161616F,
+	DRM_FORMAT_ARGB2101010,
+	DRM_FORMAT_XRGB8888,
+	DRM_FORMAT_ARGB8888,
+	DRM_FORMAT_XBGR8888,
+	DRM_FORMAT_ABGR8888,
+	DRM_FORMAT_NV12,
+	DRM_FORMAT_NV16,
+	DRM_FORMAT_NV24,
+	DRM_FORMAT_P010,
+	DRM_FORMAT_P210,
+#if defined(DRM_FORMAT_P410)
+	DRM_FORMAT_P410,
+#endif
+	DRM_FORMAT_P030,
+	DRM_FORMAT_P230,
+#if defined(DRM_FORMAT_P430)
+	DRM_FORMAT_P430,
+#endif
+};
+
@@ -786,8 +814,8 @@ struct drm_plane *apple_plane_init(struct drm_device *dev,
-			fmts = dcp_overlay_formats;
-			num_fmts = ARRAY_SIZE(dcp_overlay_formats);
+			fmts = dcp_primary_formats_h17g;
+			num_fmts = ARRAY_SIZE(dcp_primary_formats_h17g);

Assisted-by: Claude Fable 5.2

Copy link
Copy Markdown
Contributor

I checked both commits. 5e98b68e restricts H17G formats for an observed blending error, and 43f89808 applies the same fix to T8142's 27.0 display path. The alpha-format restriction should remain in the shared kernel. Re-advertising XRGB/XBGR relies on compositor behavior that the format contract does not require and restores the known wrong-colour case.

I'll retain your results as GPU/rendering evidence for the exact J813 compositions, with the Hyprland opaque-X workaround explicitly attached to the scanout result. A reproducible image using the existing restriction needs a compositor/output path that negotiates ARGB; Hyprland 0.57's XR24/XB24-only startup limitation is a separate compatibility gate. That belongs in the complete test recipe alongside the kernel, loader and stub, rather than treating the local format-list change as a generally suitable kernel fix.

The J815 firmware target, loader-alias qualification and sound gates remain unchanged. No additional historical-run reconstruction is needed from your unretained 26.6.2 notes.

…s the legacy one

The shared T8142 display file gave both Airs the legacy disp0_piodma alias,
which dt-schema rejects (an alias name is lowercase letters, digits and
'-'). The loader is its only T8142 user: the kernel looks the legacy name up
only on T8132 (iomfb_m4.c, apple-dart-dcp.h, J713 inherited mappings).

Name the alias disp0-piodma in the shared file, which aura prefers on T8142
since aurora-silicon/aura#165, and keep disp0_piodma in the J813 board file
for its earlier loaders, as integration/aurora has it. J815 only boots with
a loader newer than that anyway: the b13.1 Stage 1 cannot chainload on
T8142 without aurora-silicon/aura#250.

CHECK_DTBS: no alias warning on J815; J813's is unchanged.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
m0d16l14n1 added a commit to m0d16l14n1/linux that referenced this pull request Nov 15, 2026
…(as aurora-silicon#1532)

Same split as m0d16l14n1/linux j815/dts beb68d4: the J815 boot then
proves aura aurora-silicon#165's canonical alias path on its own.

Assisted-by: Claude Opus 5.5
Signed-off-by: Nick <nekreshen@yandex.ru>
@m0d16l14n1

m0d16l14n1 commented Nov 15, 2026 •

Copy link
Copy Markdown
Contributor Author

@DjDeveloperr @scottjones @gig3m: the J815 recipe I can confirm, the alias fix, and a request to take this out of draft.

1. Display alias (the one new DT warning): fixed in beb68d4

2. J815 recipe, as tested (27.2 stub)

Part Pin
Machine Mac17,4 (J815, board-id 0x26), 32 GB
System firmware / host macOS 27.2 (26B123), mBoot-20457.41.3
Stub macOS 27.2 (26B123), from aurora-installer with J815 added to the 26B123 entry: aurora-silicon/aurora-installer#144 (draft, policy only)
Stage 1 aura v1.0.0b13.1 + aurora-silicon/aura#250 (T8142 ANS power cycle, head 0c6aaa17), padded with the installer's own aura-execute-window, enrolled from 1TR with the installer's own kmutil configure-boot … --compress arguments
Stage 2 aura v1.0.0b13.1 + aurora-silicon/aura#250 + the T8142 ANE/AVE hand-off compatibles + the external-DCP hand-off (b97ef3a8) + display ADMAC tunables; last booted d92c3517 (m0d16l14n1/aura j815/b13-ans-cycle 0cd470f5). The now-deployed one adds aurora-silicon/aura#165: j815/b13-sleep-hib ac01bef2
Kernel last booted (60b23210): m0d16l14n1/linux j815/test-ports edc9ca1 = integration/aurora 1f5037a + this PR at c93ee62 + #1534 #1805 #1806 #1879 #1880 #1881 #1683 #1696 + our 27.2 PRs (#1899 #1900 #1905 #1906). Next: j815/test-next 06371bd (latest integration, this PR's alias change)
Mesa aurora-wip 8f013bf96b (with aurora-silicon/mesa#207, aurora-silicon/mesa#237)
Desktop Plasma 6 / KWin on GLES, which takes the ARGB primary-plane formats. No plane-format patch on J815.

Working on that composition: panel 2880x1864 including the notch rows; GPU GL 3.2, GLES 3.2 and Vulkan on the 27.2 G17G firmware; AVD; AVE; ANE (#1899); JPEG; ProRes; camera; Wi-Fi (38/38 channels); Bluetooth; Touch ID; ALS; lid angle; and an 8 h lid-closed s2idle at about 1.54 W. A 27.0 (26A428) stub also worked, in our 2026-11-09…14 runs.

3. @scottjones: which macOS / stub

  • To match this, the Mac needs macOS 27.2 (26B123) or newer, i.e. system firmware mBoot-20457.41.3. Your 26.5.1 (18000.120.36) is below every path we tested, and the firmware can't go back afterwards.
  • Two things are still pending for a J815 on the official installer:
    1. Admit the M5 MacBook Air 15-inch (J815) on the macOS 27.2 (26B123) stub aurora-installer#144 offers J815 the 27.2 stub.
    2. pcie, nvme: T8142: power ANS off and back through its storage gates aura#250 is the Stage 1 fix. The b13.1 Stage 1 in current installers stops on every T8142 (J813 too) with pcie: Unsupported compatible → "couldn't power ANS off" → "No valid payload found", whatever the stub.
  • Until both land, the path is a private installer build plus a 1TR enrolment of a fixed Stage 1. I can write the exact steps out if you want to go that way.
  • I have no J815 GPU result on the 26.6.2 stub myself.

4. Draft status: with the alias resolved and the recipe pinned, the only gate left is the boot under aurora-silicon/aura#165, which comes next. I'm moving this out of draft so it can be reviewed against that result. @DjDeveloperr, could you take another look? #1534 (sound), #1805, #1907 and #1908 follow it parent-first.

Assisted-by: Claude Opus 5.5

@m0d16l14n1
m0d16l14n1 marked this pull request as ready for review November 15, 2026 04:55

@DjDeveloperr DjDeveloperr left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Codex (GPT-6.1 Sol), reviewing on @DjDeveloperr's behalf at beb68d4e01e5e39a491b78f235c38b711ece47a1.

The two-file alias delta looks consistent. A bounded symbolic DTS comparison finds:

  • J813 adds disp0-piodma, retains disp0_piodma, and both resolve to the same PIODMA node.
  • J815 replaces the legacy alias with the canonical one at that same node.
  • All other board properties and labels are unchanged. Fetched source blobs match GitHub's identities. This is source validation, not a new DTB or hardware result.
    The retained legacy kernel lookups are behind the T8132/J713 display paths; I found no T8142 kernel consumer needing the old spelling.

The pinned 27.2/26B123 J815 composition and Plasma/KWin ARGB result answer the earlier recipe question. They retain that composed-source scope: the newly deployed ac01bef2 Stage 2 and canonical-only DT still await their reported boot.

One compatibility point needs explicit wording: #250's ANS fix does not imply #165's display-alias support. They modify separate paths, and your last-booted Stage 2 was described before adding #165. Record #165 capability as the minimum Stage 2 requirement for this J815 DT, rather than inferring it from loader chronology or ANS support.

Ready-for-review status is fine; merge remains held for the canonical-only J815 boot with the compatible loader, coordinated loader adoption, and fresh required CI. Please include the exact booted kernel/Stage 2 pins, confirm canonical alias present/legacy absent in the booted tree, and report panel success plus any reservation/admission errors. Host CI has passed; DT CI was still running when checked.

@scottjones: the official installer path still depends on unreleased installer/loader changes. This report is not a request to upgrade your system firmware or follow a private enrolment recipe. Sound and parent-first follow-up gates remain separate.

The T6030 device tree already uses apple,t6030-aop-als followed by
apple,t6020-aop-als and apple,aop-als. The binding only describes
two-entry compatible lists, so it rejects this existing three-entry
chain. Adding T8142 changes that warning's enum text and exposes it as
new CHECK_DTBS warnings on the J514s and J516s in the PR comparison.

Describe T6030's ordered fallback chain in its own oneOf arm. Keep the
existing two-entry profiles unchanged and reject missing, reordered or
incorrect fallbacks. No device-tree or driver behavior changes.

Validated the production schema and references with dtschema 2026.9.
The old schema rejects the actual chain; the fixed schema accepts it
and rejects six malformed variants. Across 819 finite compatible lists,
the documented T6030 chain is the only acceptance change.

Link: https://github.com/aurora-silicon/linux/actions/runs/43678169879/job/132702788087
Signed-off-by: Diljit Singh <hey@djdev.me>
Assisted-by: LLM

Copy link
Copy Markdown
Contributor

The failed DT job is no longer reporting the J815 alias. Both new warnings are T6030 ALS nodes on J514s/J516s: their existing compatible chain is apple,t6030-aop-als, apple,t6020-aop-als, apple,aop-als, while the binding described only two-entry lists.

Pushed 8e1ffb8 as a one-file binding correction on top of beb68d4e. It adds that exact three-entry chain in a separate oneOf arm, preserving the existing pairs. No DTS or driver changes and no warning filter/retry of unchanged source.

Production-schema validation with dtschema 2026.9 passes. The old schema rejects the actual chain; the fix accepts it and rejects six malformed variants. Across 819 finite compatible lists, that chain is the only acceptance change. Fresh DT CI must confirm the full result; loader/hardware gates remain.

DjDeveloperr added a commit to m0d16l14n1/linux that referenced this pull request Nov 15, 2026
Carry parent aurora-silicon#1532's strict T6030 compatible-chain schema fix. The J815
audio board file and the net one-file audio delta remain unchanged.

Signed-off-by: Diljit Singh <hey@djdev.me>
Assisted-by: LLM
@m0d16l14n1

Copy link
Copy Markdown
Contributor Author

Copy link
Copy Markdown
Contributor

CI on 8e1ffb8178dd6ee873329b9a20cd0f2f56e3f1eb is now green: Host 321/321, DT 137 baseline + 137 head DTBs, zero new warnings (1,482→1,465; 17 removed). All seven changed bindings and ADT coverage checks pass. The J815 alias and T6030 ALS warning failures are resolved.

The remaining landing gate is the promised canonical-only J815 boot receipt with aurora-silicon/aura#165 capability, followed by compatible-loader adoption. No such result is posted here or on aurora-silicon/aura#165 yet. The pinned private 27.2 recipe retains its reported scope; CI alone does not close the loader/hardware gate.

@DjDeveloperr
DjDeveloperr merged commit 1f38ff8 into aurora-silicon:integration/aurora Nov 17, 2026
2 checks passed
DjDeveloperr added a commit that referenced this pull request Nov 17, 2026
arm64: dts: apple: t8142: lid angle sensor on J815 and J813 (on #1532)
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.

4 participants