[PATCH v4 1/3] riscv: io: avoid null-pointer arithmetic in PIO helpers

[bot+bpf-ci at kernel.org]

> diff --git a/arch/riscv/include/asm/io.h b/arch/riscv/include/asm/io.h
> index 09bb5f57a9d3..6f5d70313c83 100644
> --- a/arch/riscv/include/asm/io.h
> +++ b/arch/riscv/include/asm/io.h
> @@ -56,6 +56,8 @@
>  #define __io_pbw()	RISCV_FENCE(iow, o)
>  #define __io_paw()	RISCV_FENCE(o, io)
>
> +#define PCI_IO_ADDR(addr)	((void __iomem *)((unsigned long)PCI_IOBASE + (addr)))
> +
>  /*
>   * Accesses from a single hart to a single I/O address must be ordered.  This
>   * allows us to use the raw read macros, but we still need to fence before and

Arnd Bergmann raised concerns about this approach in his review of v4
(https://lore.kernel.org/all/3720c3a4-cc0d-4ba8-97ae-35def0189e2e@app.fastmail.com/).

He points out that the NULL pointer arithmetic warnings are actually
useful because they indicate real bugs - NULL pointer dereferences that
will crash the kernel if the driver is ever loaded. The proper fix is to
ensure drivers have 'depends on HAS_IOPORT' in their Kconfig entries.

He suggests removing the custom RISC-V PIO accessors (ins*/outs*)
entirely and using the asm-generic versions instead, which already have
the correct #ifdef checks. This would be a cleaner approach than adding
guards to suppress warnings.

[ ... ]

> diff --git a/include/asm-generic/io.h b/include/asm-generic/io.h
> index ca5a1ce6f0f8..d799e5ccc943 100644
> --- a/include/asm-generic/io.h
> +++ b/include/asm-generic/io.h
> @@ -1205,8 +1205,12 @@ static inline void __iomem *ioremap_np(phys_addr_t offset, size_t size)
>  #define ioport_map ioport_map
>  static inline void __iomem *ioport_map(unsigned long port, unsigned int nr)
>  {
> +#ifdef CONFIG_HAS_IOPORT
>  	port &= IO_SPACE_LIMIT;
>  	return (port > MMIO_UPPER_LIMIT) ? NULL : PCI_IOBASE + port;
> +#else
> +	return NULL;
> +#endif
>  }
>  #define ARCH_HAS_GENERIC_IOPORT_MAP
>  #endif

Is this change necessary? The function definition here is already
guarded by the #ifdef CONFIG_HAS_IOPORT_MAP check, which should prevent
it from being visible when CONFIG_HAS_IOPORT is disabled.


---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

CI run summary: https://github.com/kernel-patches/bpf/actions/runs/25361792504