Skip to content

arm64: dts: exynos: google: zumapro: Add the GPIO controllers' interrupts - #275

Open
kdegeek wants to merge 1 commit into
derekross:omamobilefrom
kdegeek:zumapro-pinctrl-irq
Open

kdegeek wants to merge 1 commit into
derekross:omamobilefrom
kdegeek:zumapro-pinctrl-irq

Conversation

@kdegeek

@kdegeek kdegeek commented Nov 18, 2026

Copy link
Copy Markdown

The follow-up promised on #273: the five GPIO interrupts that zumapro.dtsi's pin controllers are missing.

What and why

The PERIC0, PERIC1, HSI1, HSI2 and HSI2UFS pin controllers have no interrupts property. So the driver's GPIO interrupt (EINT) setup fails for each of them at every boot of a zumapro board:

samsung-pinctrl 10840000.pinctrl: irq number not available

This logs once per controller, and those banks have no GPIO interrupts. The commit adds each controller's interrupt the way zuma.dtsi does:

Controller SPI
PERIC0 634
PERIC1 649
HSI1 413
HSI2 432
HSI2UFS 433

The numbers are the IRQ_GPIO_* values in Google's zumapro interrupt header (GPL, linked in the commit). The same header gives the SPIs this file already uses, for example 643 for USI2 on PERIC0, so the numbering matches.

Evidence

  • The cold boot of arm64: dts: exynos: google: Add the Pixel 9 Pro Fold (comet) #273's DT on comet (2026-11-18, 7.3.0-rc6-omamobile-comet-rc2-cand21) logged all five warnings.
  • The Fold's daily kernel runs with these five interrupts in its device tree. /proc/interrupts lists all five parents (10840000.pinctrl … 13060000.pinctrl) as requested, and none of the warnings appears.
  • No node on either zumapro board takes a GPIO interrupt on these banks yet, so the five haven't fired. Every GPIO interrupt user in the Pixel 9a and Fold device trees is on a gpa or gpn bank, which belong to the ALIVE, FAR and CUSTOM controllers.

This reaches the Pixel 9a, through the shared dtsi. It is not booted on a 9a. I'd welcome one boot there: the five warnings should be gone, with nothing else changed.

Checks

🤖 Generated with Claude Code

…upts

The PERIC0, PERIC1, HSI1, HSI2 and HSI2UFS pin controllers have no
interrupts property, so their GPIO interrupt (EINT) setup fails at
every boot of a zumapro board:

  samsung-pinctrl 10840000.pinctrl: irq number not available

once for each of the five. The controllers have no GPIO interrupts
until it succeeds.

Add each controller's GPIO interrupt, as zuma.dtsi does for the same
controllers. The numbers are those of Google's zumapro interrupt
header (IRQ_GPIO_PERIC0_PERIC0 634, IRQ_GPIO_PERIC1_PERIC1 649,
IRQ_GPIO_HSI1_HSI1 413, IRQ_GPIO_HSI2_HSI2 432 and
IRQ_GPIO_HSI2UFS_HSI2 433), which also gives the SPIs this file
already uses, such as 643 for USI2 on PERIC0.

This reaches the Pixel 9a and the Pixel 9 Pro Fold. The Fold's daily
kernel (7.3.0-rc6-omamobile-comet-rc2-cand21) runs with these five
interrupts in its device tree: /proc/interrupts lists all five parent
interrupts requested, and none of the five warnings is logged. No node
on either board takes a GPIO interrupt on these banks yet, so the five
haven't fired there; every GPIO interrupt user in their device trees
is on a gpa or gpn bank, which belong to the ALIVE, FAR and CUSTOM
controllers.

Link: https://android.googlesource.com/kernel/google-modules/soc/gs/+/3045cdcfebf91dc1d3bb3170967d23169d3b65af/include/dt-bindings/interrupt-controller/zumapro.h
Assisted-by: LLM
Signed-off-by: Nate Crawford <nate@kdegeek.com>
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.

1 participant