Re: [PATCH 5/9] mm: Turn folio_test_hugetlb into a PageType

From: Matthew Wilcox <willy@infradead.org>
To: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	linux-mm@kvack.org, David Hildenbrand <david@redhat.com>,
	Miaohe Lin <linmiaohe@huawei.com>,
	Muchun Song <muchun.song@linux.dev>,
	Oscar Salvador <osalvador@suse.de>,
	Luis Chamberlain <mcgrof@kernel.org>
Subject: Re: [PATCH 5/9] mm: Turn folio_test_hugetlb into a PageType
Date: Sat, 23 Mar 2024 03:24:58 +0000	[thread overview]
Message-ID: <Zf5LioPZm8LSQ86i@casper.infradead.org> (raw)
In-Reply-To: <ad3098cb-8766-4f68-8494-da05589d0654@suse.cz>

On Fri, Mar 22, 2024 at 11:19:34AM +0100, Vlastimil Babka wrote:
> Should we add metadata wrt closing the bug report from Luis?
> 
> https://lore.kernel.org/all/8fa1c95c-4749-33dd-42ba-243e492ab109@suse.cz/

Probably a good idea.

> I assume this wouldn't be fun wrt stable...

I don't think it should be too bad?  I think we only need to backport it
as far as v6.6 when I got rid of folio->dtor.  Yes, it's unreliable
before that, but it doesn't cause crashes, just bad decisions.

> > Since hugetlb pages do not use individual page mapcounts (they are always
> > fully mapped and use the entire_mapcount field to record the number
> 
> Wasn't there some discussions to allow partial mappings of hugetlb? What
> would be the implications?

I think I'm hammering another nail into that coffin.  As I understand
it, everyone has given up on that proposal and they're looking to make
THP more reliable so they can use THP.  See Yu Zhao's recent proposals.