Skip to content

arm64: dts: apple: t8152-j833ct: the iMac (24-inch, M6) core board (on #2340) - #2343

Draft
RyanTheTide wants to merge 14 commits into
integration/aurorafrom
dtfill/t8152-j833ct-core
Draft

RyanTheTide wants to merge 14 commits into
integration/aurorafrom
dtfill/t8152-j833ct-core

Conversation

@RyanTheTide

Copy link
Copy Markdown
Contributor

Problem

The iMac 24-inch M6 (J833ct) cannot boot from the tree. t8152-j833ct.dts only includes t8152.dtsi, where every node is disabled, the CPUs, AIC and timer included.

Change

  1. t8152: share the loader-booted core board description: move the SoC-level part of the J804 board file into t8152-common.dtsi, unchanged. That covers the RAM handoff contract, framebuffer and cpuidle, the twelve CPUs with their first two performance states and cluster controllers, the AIC, timer, SMC and its hwmon, the watchdog, the GPIO controller, I2C1/I2C3, the boot NOR, and the PMGR states. t8152-j804.dts keeps only what is J804's: the no-restore GPIO pins, the trackpad AFE reset, the fan/power hwmon keys, the keyboard backlight, the N1, SD and SEP board parts.
  2. t8152-j833ct: describe the iMac (24-inch, M6) core board: the J833ct includes the common file and adds its identity ("all-in-one") and its own no-restore GPIO pins.

Testing

cpp + dtc:

  • J804: the decompiled tree is identical before and after the split.
  • J873g: unchanged.
  • J833ct: builds with no dtc warnings and gains the core board: 12 CPUs, AIC, timer, SMC, watchdog, pinctrl, I2C, NOR and PMGR.

Still to run

No J833ct owner is known. A first row needs someone with the iMac: aura Stage 2, then Linux, with framebuffer console on the panel, every started CPU online, two cpufreq states per cluster, SMC temperatures, NVRAM readable, and no SError at pinctrl probe. The loader also has to accept the board, which it does not yet: it only accepts J873g among T8152 boards.

Details

Assisted-by: LLM (Claude:Opus 5.5)

LiamRay11 and others added 14 commits November 16, 2026 09:04
The T8152 (M6) PMGR blocks are compatible with the T8103 ones. The
J873 device tree already uses this compatible; document it.

Signed-off-by: LiamRay11 <228416644+LiamRay11@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The T8152 (M6) PMGR power state controls are compatible with the
T8103 ones. The J873 device tree already uses this compatible;
document it.

Signed-off-by: LiamRay11 <228416644+LiamRay11@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The I2C controllers of T8152 (M6) are compatible with the T8103 ones.

Signed-off-by: LiamRay11 <228416644+LiamRay11@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The T8152 (M6) watchdog is compatible with the T8103 one.

Signed-off-by: LiamRay11 <228416644+LiamRay11@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The T8152 (M6) FPWM, which drives the keyboard backlight of the
J804 MacBook Pro, is compatible with the S5L one.

Signed-off-by: LiamRay11 <228416644+LiamRay11@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Reading every pin of the T8152 GPIO block to fill a register cache at
probe takes an SError on the J804 MacBook Pro: pins 78 and 208 fault
on an AP read. The T8140 profile accesses the block uncached and
honours gpio-reserved-ranges, so a board that has to keep the CPU off
such pins can use "apple,t8152-pinctrl", "apple,t8140-pinctrl", as
T6050 does with its T8132 fallback. The existing T8103 fallback stays
valid for boards that do not need it.

Signed-off-by: LiamRay11 <228416644+LiamRay11@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
As on T8140, the OS cannot request core power-down on T8152 (M6) and
WFI is the only idle state. Allow "apple,t8152-cpuidle" with the
T8140 compatible as its fallback, which the driver already matches.

Signed-off-by: LiamRay11 <228416644+LiamRay11@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Turn the disabled J804 inventory into a board description for a RAM
boot from the loader, on top of the shared T8152 description:

  - the twelve CPUs (the loader removes those it has not started),
    with the first two performance states of each cluster, which the
    cpufreq driver may use without a thermal policy, as on J873
  - the AIC and the timer, the boot framebuffer and CPU idle
  - the SMC with its mailbox, the pcIO output that resets the
    trackpad's touch controller, and the hardware monitor: fan, power,
    voltage and current keys, with temperatures discovered. The fan
    is monitored only; its speed stays under firmware control
  - the watchdog, the GPIO controller with the pins that fault on an
    AP read reserved, the I2C controllers of the jack codec and the
    speaker amplifiers, and the keyboard backlight PWM
  - the boot NOR and the PMGR block from t8152-j873-qspi.dtsi, shared
    with J873, plus the SIO and I2C power states

Everything else stays described and disabled; the storage, display,
input, wireless and GPU nodes follow with their drivers. The values
come from the J804 ADT.

Tested on a MacBook Pro (M6, J804) with an equivalent standalone device
tree on an earlier base (integration/aurora f5fa539): all twelve CPUs
come up, the watchdog restarts the machine, the SMC hardware monitor
reports about 170 inputs including the fan, the keyboard backlight
lights, and the NVRAM partition of the boot NOR reads. This tree,
rewritten onto the shared T8152 labels, has not been booted.

Signed-off-by: LiamRay11 <228416644+LiamRay11@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The J804 has no apcie host, port 0 or port-0 DART, so the N1 Wi-Fi and
Bluetooth functions that the centauri drivers handle never appear.

The J804 ADT's /arm-io/apcie, apcie/pci-bridge0 (PERST# GPIO 3, CLKREQ#
GPIO 0, maximum link speed 3), its centauri-control/alpha/beta functions
and /arm-io/dart-apcie0 have the same registers, interrupts and pins as
the J873g's. Move the host and port 0 from t8152-j873-n1.dtsi (sithglan)
into a board-neutral t8152-n1.dtsi and include it from both boards. The
J873g keeps its own pinctrl_ap and pcie0_dart in t8152-j873-n1.dtsi; its
DTB is unchanged. The J804 uses pcie0_dart and pinctrl_ap from t8152.dtsi
and its own board file.

On the J804, add the PMGR power states the port and its DART sit in, ADT
pmgr devices APCIE_GP (+0x8c0) and APCIE_SYS_GP (+0xec0, child of
APCIE_GP), kept on because the loader hands over a trained link, and give
them to the host and to pcie0_dart. Describe the Wi-Fi and Bluetooth
functions (0,1 and 0,2) with the wifi0 and bluetooth0 aliases, which the
loader fills from the ADT's /chosen mac-address-wifi0 and
mac-address-bluetooth0.

Signed-off-by: Ryan Murray <ryan@aurorasilicon.org>
The J804 has a GL9755 SD host controller on apcie port 2 (ADT
/arm-io/apcie/pci-bridge2/pcie-sdreader); the Mac mini has a different
device there, so this stays in the board file. The loader switches on the
reader's power (the ADT's SMC function-sd_pwr_en) and trains the link.

Add port 2 to the host: its port registers (+0x18a028000) and PHY lane
(+0x18f018000), from the apcie node's reg; the host interrupts of ports 0
to 2 (ADT apcie interrupts 1571, 1579, 1587), because the driver takes a
port's interrupt by its port number; and bus 2 on dart-apcie2, stream 1.
The port's PERST# is pci-bridge2's function-perst, GPIO 5, and its link
runs at gen 1 (maximum-link-speed 1). The port's MSIs come from the
host's shared range. Card detect and write protect are active high
(card-detect-polarity and write-protect-polarity 1), as in
t6050-j714s.dts and t8142-j704-pcie.dtsi.

Enable pcie2_dart with the apcie power state, APCIE_SYS_GP.

Signed-off-by: Ryan Murray <ryan@aurorasilicon.org>
Describe the T8152 SEP for the 14-inch MacBook Pro (M6).

In t8152-peripherals.dtsi, next to the existing sep_dart:
 - sep@28a600000 (J804 ADT /arm-io/sep, iop-sep,ascwrap-v8, first reg
   0x8a600000 + 0x74000), disabled: iBoot hands over a running SEP and
   apple_sep enables it and its DART when it loads.
 - sep_mbox@28a710080, disabled: on T8152 the SEP talks through its window
   of the post-office block, like the SMC (third sep reg + 0x10080), with
   the ADT interrupts 542 (send-empty) and 541 (recv-not-empty).
 - sep_dart gets the DMA window iBoot programmed (ADT dart-sep vm-base-0
   0x10000004000, vm-size-0 0xffff0000), as on T8142 and T6050.
The J804, J873g and J833ct ADTs describe sep and dart-sep identically.

In t8152-j804.dts: the "sep" alias, which makes the loader reserve the SEP
firmware and publish the boot manifests on the node; the PMGR SEP (+0xc00,
critical) and GPIO (+0xb60) states, both kept on as in
t8142-j813-sep.dtsi; those domains on the SEP and on its DART; and the
mailbox enabled.

The SEP node follows t8140.dtsi and t8142-j813-sep.dtsi; the mailbox
follows the J804 SMC mailbox (LiamRay11).

Signed-off-by: Ryan Murray <ryan@aurorasilicon.org>
…mailboxes

On T8152 the NVMe (ANS), display (DCP), keyboard and trackpad (MTP)
and SEP coprocessors talk through windows of the same post-office
block as the SMC: counted FIFOs sharing one status register, with
interrupt masking left to the interrupt controller. The register
layout is the one apple,t8152-smc-mailbox already describes.

Add a compatible per coprocessor, each with apple,t8152-smc-mailbox as
its fallback, so the existing driver match serves them without a code
change and a later difference can still be keyed on the specific
compatible.

All four carry traffic on a MacBook Pro (M6, J804), with the
single-compatible form of these nodes: NVMe, the keyboard and
trackpad, and Touch ID run through them, and the DCP driver exchanges
its messages through the DCP window.

Signed-off-by: LiamRay11 <228416644+LiamRay11@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Move the parts of the J804 core board description that are SoC-level
into t8152-common.dtsi, so that the J833ct iMac can use them too:

  - the RAM handoff contract, the boot framebuffer and CPU idle
  - the twelve CPUs with the first two performance states of each
    cluster and the cluster performance controllers
  - the AIC and the timer
  - the SMC, its mailbox and SRAM carve-out, and the hardware monitor
    with temperatures discovered
  - the watchdog, the GPIO controller, the I2C controllers of the jack
    codec and the speaker amplifiers with the (disabled) jack codec
  - the boot NOR and PMGR block, plus the SIO, I2C, internal PCIe
    (APCIE_GP, APCIE_SYS_GP), GPIO and SEP power states

Every one of these ADT nodes (/cpus/cpu*, /arm-io/aic, wdt, smc, gpio0,
i2c1, i2c3, i2c1/audio-codec-output, qspi, pmgr) has the same reg,
interrupts and gates in the J804 and J833ct ADTs. t8152-j804.dts keeps
what is J804 only: the GPIO no-restore pins, the trackpad AFE reset
pcIO output, the fan, power and battery hwmon keys, and the keyboard
backlight on fpwm0 (absent from the J833ct ADT).

The J804 DTB decompiles (dtc -s) to the same tree as before; the
common part is the J804 description by LiamRay11, moved unchanged.

Signed-off-by: Ryan Murray <ryan@aurorasilicon.org>
… board

Base the J833ct on the shared loader-booted T8152 description, which
brings up the twelve CPUs (the loader removes those it did not start),
the AIC, timer, boot framebuffer console, SMC, watchdog, GPIO, the
I2C controllers and the boot NOR, as on J804.

The board adds only its GPIO no-restore pins (78, 79 and 208, from the
J833ct ADT /arm-io/gpio0 no-restore-gpio-pins) and its identity, with
the model name from the ADT product description. The keyboard and
trackpad are Bluetooth accessories: the J833ct ADT has no MTP and no
fpwm0, so there is no trackpad reset and no keyboard backlight. The
hardware monitor discovers temperatures only; fan and power keys
follow once read on the machine.

Not booted: no J833ct is available for testing.

Signed-off-by: Ryan Murray <ryan@aurorasilicon.org>
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