Re: [PATCH v5 0/2] gpio: Add Cortina Systems CS75xx GPIO driver

From: Bartosz Golaszewski

Date: Thu Nov 19 2026 - 08:43:08 EST


On Thu, Nov 19, 2026 at 02:02:48PM +0300, Fil Dunsky wrote:
> The Cortina Systems CS75xx ("Goldengate G2") SoCs have five banks of 32
> GPIO lines, each bank with its own interrupt. This series adds the
> binding and a driver based on gpio-mmio; level and edge triggers are
> supported, both edges are emulated by flipping the polarity.
>
> Each pin is either a GPIO line or belongs to a peripheral (NAND,
> Ethernet, UART, I2C, ...). That choice is made in the global control
> block, which is not part of the GPIO controller: as discussed on v1 [1],
> the driver does not touch the pin mux and leaves it as the boot loader
> set it up, until there is a pin controller driver for that block.
> Testing the reset button on the Securifi Almond+ requires selecting
> GPIO37 in the global pin mux before gpio-keys is probed; the current
> boot path does not provide that setup. The tests used a separate,
> board-specific setup step. The GPIO driver does not program the mux.
>
> The driver used to be part of a series with the interrupt controller,
> I2C and watchdog drivers and the device tree changes. That series is now
> split per subsystem. The driver can be built with COMPILE_TEST without
> the platform series [2], which is under review separately. The other
> Kconfig option, ARCH_CORTINA_ACCESS, is defined by that series in
> arch/arm/mach-goldengate/Kconfig; it is not in mainline yet. There is no
> code dependency on the platform series. The device tree nodes will
> follow with the other CS75xx dts changes once the platform is merged.
>
> There is no public documentation; the register layout comes from the GPL
> vendor kernels. Tested on a Securifi Almond+ (CS7542): the reset button
> (GPIO 37) works as an active-low gpio-keys input, and the SX8650
> touchscreen uses a level-triggered GPIO interrupt.
>
> [1] https://lore.kernel.org/all/20261012-dreamlike-hazily-19bb60326f0a@spud/
> [2] https://lore.kernel.org/all/20261114-b4-cs75xx-v7-0-8f17f38821b4@xxxxxxxxx/
>

The driver looks good but I cannot pick it up while it references a
Kconfig symbol that's not upstream.

Bart