Re: [PATCH v4 0/3] mm/page_owner: add filter infrastructure for print_mode and NUMA filtering

[zhen.ni]

在 2026/5/1 02:22, Andrew Morton 写道:
> On Fri,  1 May 2026 00:32:44 +0800 Zhen Ni <zhen.ni@xxxxxxxxxxxx> wrote:
>
> > This patch series introduces filtering capabilities to the page_owner
> > feature to address storage and performance challenges in production
> > environments.
>
> AI review asks a couple of reasonable-sounding questions:
> 	https://sashiko.dev/#/patchset/20260430163247.13628-1-zhen.ni@xxxxxxxxxxxx

Will this cause KCSAN splats?

While the practical impact is minimal (debugfs interface, infrequent
writes, torn reads only cause temporary debug output inconsistency),
we should properly handle this to avoid KCSAN warnings.

I'm wondering if using the __data_racy qualifier would be appropriate
here? Something like:


struct page_owner_filter {
    ...
    nodemask_t __data_racy nid_mask;
};


Is it necessary to evaluate nodes_empty(mask) inside this loop?

I'll fix this by moving the check outside the loop.


Best regards,
Zhen