Re: [PATCH 10/13] ARM: stm32: remove stm32f4/f7/h7 MCU support

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Medium] Incomplete removal of STM32 MCU platform leaves orphaned drivers and dangling Kconfig references
- [Low] Grammatical typo in documentation update
--

commit b3d9699b653730bdc87f076ae9ec5792bb7764a3
Author: Arnd Bergmann <arnd@arndb.de>

    ARM: stm32: remove stm32f4/f7/h7 MCU support
    
    The stm32 platform started out as a Cortex-M based microcontoller
    platform in 2007, with a few of the older products gaining Linux kernel
    support in 2016. These days, the only relevant chips for Linux are the
    stm32mp1 (Cortex-A7) and stm32mp2 (Cortex-A35) product lines, with the
    microcontrollers running smaller RTOSs on new products.
    
    The microcontrollers were marked as deprecated in the linux-7.3LTS
    release, which will remain supported for a few more years and
    is expected to be sufficient for any of the remaining original users.
    
    Remove the MCU support now, leaving only the stm32mp SoCs in the platform.

> diff --git a/Documentation/arch/arm/stm32/overview.rst b/Documentation/arch/arm/stm32/overview.rst
> index 85cfc84107983..b63aa875cb485 100644
> --- a/Documentation/arch/arm/stm32/overview.rst
> +++ b/Documentation/arch/arm/stm32/overview.rst
> @@ -5,15 +5,13 @@ STM32 ARM Linux Overview
>  Introduction
>  ------------
>  
> -The STMicroelectronics STM32 family of Cortex-A microprocessors (MPUs) and
> -Cortex-M microcontrollers (MCUs) are supported by the 'STM32' platform of
> -ARM Linux.
> +The STMicroelectronics STM32 family of Cortex-A microprocessors (MPUs) is
> +supported by the 'STM32' platform of ARM Linux, support for the Cortex-M
> +based MCUs is has been removed.

[Severity: Low]
Is there a grammatical typo here? Should it be "has been removed" instead of
"is has been removed"?

[ ... ]

> diff --git a/arch/arm/mach-stm32/Kconfig b/arch/arm/mach-stm32/Kconfig
> index 4ac3384e397aa..6ca8507f94512 100644
> --- a/arch/arm/mach-stm32/Kconfig
> +++ b/arch/arm/mach-stm32/Kconfig
> @@ -1,17 +1,15 @@
>  # SPDX-License-Identifier: GPL-2.0-only
>  menuconfig ARCH_STM32
>  	bool "STMicroelectronics STM32 family"
> -	depends on ARM_SINGLE_ARMV7M || ARCH_MULTI_V7
> -	select ARMV7M_SYSTICK if ARM_SINGLE_ARMV7M
> -	select HAVE_ARM_ARCH_TIMER if ARCH_MULTI_V7
> -	select ARM_GIC if ARCH_MULTI_V7
> -	select ARM_PSCI if ARCH_MULTI_V7
> +	depends on ARCH_MULTI_V7
> +	select HAVE_ARM_ARCH_TIMER
> +	select ARM_GIC
> +	select ARM_PSCI
>  	select ARM_AMBA
>  	select ARCH_HAS_RESET_CONTROLLER
>  	select CLKSRC_STM32
>  	select PINCTRL
>  	select RESET_CONTROLLER
> -	select STM32_EXTI if ARM_SINGLE_ARMV7M

[Severity: Medium]
This is a pre-existing issue, but does removing this selection of STM32_EXTI
permanently disable the irq-stm32-exti.c driver?

Looking at drivers/irqchip/Kconfig on line 485, STM32_EXTI is defined as an
invisible boolean config symbol. Without being selected here, the driver will
never be built. Could the corresponding driver and Kconfig entry be removed as
well?

>  	select STM32_FIREWALL
>  	help
>  	  Support for STMicroelectronics STM32MP1 processors using Cortex-A7
> @@ -20,32 +18,6 @@ menuconfig ARCH_STM32
>  
>  if ARCH_STM32
>  
> -if ARM_SINGLE_ARMV7M
> -
> -config MACH_STM32F429
> -	bool "STMicroelectronics STM32F429 (DEPRECATED)"
> -	default y
> -
> -config MACH_STM32F469
> -	bool "STMicroelectronics STM32F469 (DEPRECATED)"
> -	default y
> -
> -config MACH_STM32F746
> -	bool "STMicroelectronics STM32F746 (DEPRECATED)"
> -	default y
> -
> -config MACH_STM32F769
> -	bool "STMicroelectronics STM32F769 (DEPRECATED)"
> -	default y
> -
> -config MACH_STM32H743
> -	bool "STMicroelectronics STM32H743 (DEPRECATED)"
> -	default y

[Severity: Medium]
This isn't a bug introduced by this patch, but doesn't removing these MCU
and debug configs leave dangling references in other files?

For instance, in drivers/clk/Kconfig at line 442, COMMON_CLK_STM32F relies
on these now-undefined symbols:

    def_bool COMMON_CLK && (MACH_STM32F429 || MACH_STM32F469 || MACH_STM32F746)

This silently disables the clock driver and leaves it as dead code. Similarly,
drivers/pinctrl/stm32/Kconfig on line 14 uses MACH_STM32F429 in its visibility
condition for PINCTRL_STM32F429. In arch/arm/include/debug/stm32.S on line 7,
the preprocessor directive checks for CONFIG_STM32F4_DEBUG_UART, which was
also removed in this patch.

Could these remaining orphaned drivers and dead references be cleaned up as
well?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260908152808.3928630-1-arnd@kernel.org?part=10