Re: [PATCH 3/3] parisc/mm: Drop all pxdp_get()

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] mm: unnecessary out-of-line conversion of ptep_test_and_clear_young()
--

commit 6892dae5d3895dcd59fc83328a1aca2f33ba942b
Author: Anshuman Khandual <anshuman.khandual@arm.com>

parisc/mm: Drop all pxdp_get()

This commit drops the redundant platform-specific pxdp_get() helpers in
favor of the generic MM header definitions. It also moves dependent functions
like ptep_test_and_clear_young() out-of-line to avoid build failures.

> diff --git a/arch/parisc/mm/init.c b/arch/parisc/mm/init.c
> index be3380c9bcdae..3da7d8def9435 100644
> --- a/arch/parisc/mm/init.c
> +++ b/arch/parisc/mm/init.c
> @@ -992,3 +992,16 @@ struct execmem_info __init *execmem_arch_setup(void)
>  	return &execmem_info;
>  }
>  #endif /* CONFIG_EXECMEM */
> +
> +bool ptep_test_and_clear_young(struct vm_area_struct *vma,
> +			       unsigned long addr, pte_t *ptep)
> +{
> +	pte_t pte;
> +
> +	pte = ptep_get(ptep);
> +	if (!pte_young(pte))
> +		return false;
> +
> +	set_pte_at(vma->vm_mm, addr, ptep, pte_mkold(pte));
> +	return true;
> +}

[Severity: Medium]
Is there a reason this function was moved out-of-line instead of dropping
the custom implementation entirely?

This implementation perfectly matches the generic inline default provided in
include/linux/pgtable.h. Since this path is called heavily during memory
pressure and page reclaim, such as from page_referenced_one(), moving it
out-of-line introduces function call overhead for every PTE scanned.

Could we instead remove the __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG definition
in arch/parisc/include/asm/pgtable.h? This would allow the compiler to use
the generic inline fallback and avoid the performance regression.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260923043226.331880-1-anshuman.khandual@arm.com?part=3