Repository navigation
arm64: dts: apple: t8122-j504: describe the MagSafe PD controller - #2392
RyanTheTide wants to merge 1 commit into
Conversation
The MacBook Pro 14 (M3, J504) charges over a MagSafe 3 port whose cd321x USB PD controller (hpm5) sits on i2c0 next to the two USB-C controllers, but the board DT only described hpm0 and hpm1. Add hpm5 at 0x3a on i2c0 without a connector, the same shape as the MagSafe node on the T600x J314/J316 and T8112 J413/J415. The bus, address and interrupt line (AP GPIO 11, shared with hpm0 and hpm1) come from the board's ADT. The J613 and J615 already describe their MagSafe controller in t8122-usbpd-spmi.dtsi; the J433 and J434 have no MagSafe port. Their DTBs are unchanged. Signed-off-by: Ryan Murray <ryan@aurorasilicon.org>
|
Looks good; approve once the J504 MagSafe row is in. It does not block the merge: the change is DT-only and adds one inert node. I checked the node against the other cd321x MagSafe nodes (t600x-j314-j316.dtsi:438, T8112 J413/J415): the same shape and no connector. IRQ 11 level-low is shared with hpm0/hpm1 (t8122-usbpd-i2c.dtsi:21,58), and tipd requests its IRQ with IRQF_SHARED (drivers/usb/typec/tipd/core.c:4177-4179 on integration/aurora), so a third sharer is fine. I built all 7 T8122 DTBs from this PR merged onto the current integration/aurora: only t8122-j504 changes, and the other 6 are byte-identical to the base. dt-checks and host-tests pass. HW row: J504 on this head: probe with no tipd errors. MagSafe plugged gives a new /sys/class/typec/portN and its tps6598x-source-psy online=1, and unplugged gives online=0. hpm0/hpm1 still enumerate and still deliver a USB-C plug event on the shared IRQ 11. Plus 0 AVC, 0 failed units and 0 new kernel errors. Assisted-by: LLM (Claude:Opus 5.5) |
Problem
The MacBook Pro 14 (M3, J504) has a MagSafe 3 port. Its USB PD controller (hpm5, a cd321x) is on i2c0 with the two USB-C controllers, but t8122-j504.dts and t8122-usbpd-i2c.dtsi only describe hpm0 and hpm1. Without the node, Linux does not see the MagSafe port or its power supply.
Change
hpm5: usb-pd@3aon i2c0,apple,cd321x, no connector. This is the same shape as the MagSafe node on T600x J314/J316 and T8112 J413/J415. The bus, address and interrupt line come from the board's ADT. The interrupt is AP GPIO 11, level low, which hpm0 and hpm1 already share.apple,sn201202x, hpm5 at 0x8) in t8122-usbpd-spmi.dtsi. No change there.Testing
apple,cd321xis already covered by the ti,tps6598x binding.Still to run
/sys/class/typec/portNappears for hpm5, and its/sys/class/power_supply/*/onlinereads 1. Unplug it:onlinereads 0.dmesg | grep -i tipdshows no probe or IRQ errors, and hpm0/hpm1 still work as before (both share IRQ 11).Per-board values and DTB diffs
Values from each board's ADT:
t8122-j504 decompiled DTB diff:
IDENTICAL-DTB: t8122-j433, t8122-j434, t8122-j613, t8122-j613-25g83, t8122-j615, t8122-j615-25g83.
Assisted-by: LLM (Claude:Opus 5.5)