Re: [PATCH v9 1/3] riscv: implement this_cpu operations

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Low] riscv: percpu: typeof(pcp) retains __percpu and const qualifiers
  in locals
--

commit 9753271dee3a084b7bff8ad7ad29d76cb64e49f4
Author: Yunhui Cui <cuiyunhui@bytedance.com>

riscv: implement this_cpu operations

This commit implements RISC-V-specific this_cpu operations using preemption
disablement and relaxed AMOs (or 32-bit LR/SC for 8-bit and 16-bit accesses
when Zabha is unavailable).

Link: https://lore.kernel.org/r/20260826010748.zphu3wh2hlm4t4o2@offworld

> diff --git a/arch/riscv/include/asm/percpu.h b/arch/riscv/include/asm/percpu.h
> new file mode 100644
> index 0000000000000..280b122b9fe69
> --- /dev/null
> +++ b/arch/riscv/include/asm/percpu.h

[ ... ]

> +#define _pcp_protect_return(op, pcp, args...)				\
> +({									\
> +	typeof(pcp) __retval;						\
> +	preempt_disable_notrace();					\
> +	__retval = (typeof(pcp))op(raw_cpu_ptr(&(pcp)), ##args);	\
> +	preempt_enable_notrace();					\
> +	__retval;							\
> +})
> +
> +#define _pcp_protect_cmpxchg_return(op, pcp, old, new)			\
> +({									\
> +	typeof(pcp) __retval;						\
> +	typeof(pcp) *__pcp_ptr;						\
> +									\
> +	preempt_disable_notrace();					\
> +	__pcp_ptr = raw_cpu_ptr(&(pcp));				\
> +	instrument_atomic_read_write(__pcp_ptr, sizeof(*__pcp_ptr));	\
> +	__retval = (typeof(pcp))op(__pcp_ptr, (unsigned long)(old),	\
> +				     (unsigned long)(new));		\
> +	preempt_enable_notrace();					\
> +	__retval;							\
> +})

[ ... ]

> +#define this_cpu_cmpxchg128(pcp, o, n)					\
> +({									\
> +	u128 old__ = (o), new__ = (n), ret__;				\
> +	typeof(pcp) *ptr__;						\
> +									\
> +	preempt_disable_notrace();					\
> +	ptr__ = raw_cpu_ptr(&(pcp));					\
> +	if (system_has_cmpxchg128())					\
> +		ret__ = cmpxchg128_local(ptr__, old__, new__);		\
> +	else								\
> +		ret__ = __percpu_generic_cmpxchg128(ptr__, old__, new__); \
> +	preempt_enable_notrace();					\
> +	ret__;								\
> +})

[Severity: Low]
Should _pcp_protect_return(), _pcp_protect_cmpxchg_return(), and
this_cpu_cmpxchg128() use TYPEOF_UNQUAL(pcp) instead of typeof(pcp) for
their local variables and pointers?

typeof(pcp) retains the __percpu address-space and const qualifiers from
pcp.  Because raw_cpu_ptr() strips __percpu via PERCPU_PTR() (since commit
6a39fe05ecaa ("percpu: use TYPEOF_UNQUAL() in *_cpu_ptr() accessors")),
declaring __pcp_ptr and ptr__ with typeof(pcp) * leaves them in the
__percpu address space and triggers sparse address-space mismatch warnings
on assignment from raw_cpu_ptr() and when passing them to helpers taking
regular pointers.  Retaining const on __retval also causes a build error if
this_cpu_read() is used on a const per-CPU variable.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261108110406.97495-1-cuiyunhui@bytedance.com?part=1