Repository navigation
arm64: dts: apple: t6041-j616c: describe the GPU (stacked on #914) - #917
Closed
AxxzyWasTaken wants to merge 8 commits into
Closed
AxxzyWasTaken wants to merge 8 commits into
AxxzyWasTaken wants to merge 8 commits into
Conversation
The Chapel PMIC is on the T6030 and T6031 (M3 Pro/Max) SPMI buses, whose trees already use this compatible, and on the T6041 (M4 Max). Signed-off-by: AxxzyWasTaken <axxzytte@gmail.com>
T6041 has the blocks t6040.dtsi describes at the same addresses and interrupts (checked against a J616C ADT). Three things differ: - Two clusters of six P cores instead of five. The boot loader matches the DT's CPU nodes to the ADT's by MPIDR and drops the ones a binned part leaves out (the 14-core J616C has exactly the J616s's cores), so t6040.dtsi adds the sixth core of each cluster when t6041.dtsi asks for them. - The main PMIC is a Chapel at SPMI address 0xb (ADT nub-spmi0 spmi-chapelL1-b), not an Abbey at 0xe. Its nvmem cells and RTC alarm registers (ADT info-rtc_alarm_*) sit at the Abbey offsets. - Up to eight AMCCs with four cache planes each, six on the binned part (ADT amcc-enable-mask 0x3f): the MCC lists the 24 planes every T6041 has, which needs the mcc driver to take more than 16. t6040.dtb files do not change. Signed-off-by: AxxzyWasTaken <axxzytte@gmail.com>
J616C is the J616s chassis with the T6041. Checked against a J616C ADT (macOS 26.6.2, iBoot mBoot-18000.161.10): every power domain of t6040-pmgr.dtsi, the display block, AVD, PMP, AOP, MTP, SMC, NVMe and USB sit at the same addresses, with the same interrupts and DART streams. What differs from J616s: - the trackpad firmware is tpmtfw-j616c.bin, and the factory data mesacal-j616c.bin and msrk-j616c.bin; - nub-spmi1 holds a Chapel at 0xb and a Spitfire at 0x9, nub-spmi2 Portiers at 0xa and 0xb, not Abbeys at 0xe; nothing on nub-spmi2 has a binding yet, so that bus is left out; - five per-unit USB3 PHY values (auspll, cio4pll, aus40cmn-shm), from this machine's ADT; every other tunable matches J616s; - iBoot moves the ISP firmware from boot to boot, so the ISP stays off until the boot loader fills in its placement; - the GPU (ADT gpu,t6041) is not described yet and stays off. Signed-off-by: AxxzyWasTaken <axxzytte@gmail.com>
J616C uses the shared T604x GPU description with the T6041 compatible. T6041 has four GPU clusters, so the operating points carry one voltage cell per cluster (ADT /arm-io/sgx perf-states to perf-states3). The boot loader fills these, the CS and AFR tables and the six memory regions; the GPU stays unusable without that handoff.
This was referenced Nov 3, 2026
Booting the J616C on 7.3 left the keyboard, trackpad and audio dead: RTKit timed out on the MTP and AOP, and dockchannel-hid failed its AFE preflight. - aura only programs a co-processor's DAPF if the device tree has its alias. The shared J614s/J616s file has only pmp, so add aop and mtp. - The J616C has no gp1c key. Its AFE and STM resets are lines 25 and 26 of the SMC's low GPIO block, so describe that block, drop apple,afe-smc and use power method 2. - Use the plain dockchannel-hid compatible: this Mac's MTP firmware runs without the J616s's ring transport on our daily tree. Signed-off-by: AxxzyWasTaken <axxzytte@gmail.com>
Pick up the J616C keyboard, trackpad and audio device tree fixes.
aura only opens a co-processor's DAPF if the device tree has its alias, as with the AOP and MTP. The ISP stays disabled here for now, so this changes nothing yet, but the camera can't start without it once the ISP is enabled. Signed-off-by: AxxzyWasTaken <axxzytte@gmail.com>
Contributor
|
Merge train (MTL, 06/11): can't take this one yet. Commit bd73f99 ("arm64: dts: apple: t6041-j616c: describe the GPU") has no |
Contributor
Author
|
Closing as already covered: |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stacked on #914: only the last commit,
arm64: dts: apple: t6041-j616c: describe the GPU, is new.#914 keeps the J616C GPU off. This turns it on using the shared T604x GPU nodes:
compatible = "apple,agx-t6041"instead of the inherited T6040 one./arm-io/sgxperf-statestoperf-states3). The shared tables have two, for T6040.&gpuand&gfx_mboxare no longer forced off.The loader fills the operating points, the CS and AFR tables and the six memory regions: aura#68. Without that handoff the regions stay disabled and the driver can't probe, so the GPU stays unused.
Testing
t6041-j616c.dtbbuilds.aurora-wip,vkmarkon Wayland kept the fragment engine busy (checked infdinfo), and Sober (x86, through FEX) reached Roblox's home screen, with no recoveries in about 2.5 minutes. No game has been played on it yet.