Re: [PATCH v1 05/29] mm/balloon_compaction: make PageOffline sticky until the page is freed

[David Hildenbrand]

On 01.07.25 08:13, Harry Yoo wrote:
> On Mon, Jun 30, 2025 at 12:14:01PM -0400, Zi Yan wrote:
> > On 30 Jun 2025, at 12:01, Lorenzo Stoakes wrote:
> >
> > > On Mon, Jun 30, 2025 at 02:59:46PM +0200, David Hildenbrand wrote:
> > > > Let the page freeing code handle clearing the page type.
> > >
> > > Why is this advantageous? We want to keep the page marked offline for longer?
> > >
> > > > Acked-by: Zi Yan <ziy@xxxxxxxxxx>
> > > > Acked-by: Harry Yoo <harry.yoo@xxxxxxxxxx>
> > > > Signed-off-by: David Hildenbrand <david@xxxxxxxxxx>
> > >
> > > On assumption this UINT_MAX stuff is sane :)) I mean this is straightforward I
> > > guess:
> >
> > This is how page type is cleared.
> > See: https://elixir.bootlin.com/linux/v6.15.4/source/include/linux/page-flags.h#L1013.
> >
> > I agree with you that patch 4 should have a comment in free_pages_prepare()
> > about what the code is for and why UINT_MAX is used.
>
> Or instead of comment, maybe something like this:
>
> /* Clear any page type */
> static __always_inline void __ClearPageType(struct page *page)
> {
> 	VM_WARN_ON_ONCE_PAGE(!page_has_type(page), page);
> 	page->page_type = UINT_MAX;
> }
>
> in patch 4:
>
> if (unlikely(page_has_type(page)))
> 	__ClearPageType(page);

I don't think we should do that. It's very specialized code that nobody 
should be reusing.

And it will all change once Willy reworks the page_type vs. _mapcount 
overlay.

--
Cheers,

David / dhildenb