Re: Folio mapcount

From: David Hildenbrand <david@redhat.com>
To: "Yin, Fengwei" <fengwei.yin@intel.com>, Zi Yan <ziy@nvidia.com>
Cc: Matthew Wilcox <willy@infradead.org>,
	linux-mm@kvack.org, Vishal Moola <vishal.moola@gmail.com>,
	Hugh Dickins <hughd@google.com>, Rik van Riel <riel@surriel.com>
Subject: Re: Folio mapcount
Date: Sun, 2 Jul 2023 13:45:48 +0200	[thread overview]
Message-ID: <6cec6f68-248e-63b4-5615-9e0f3f819a0a@redhat.com> (raw)
In-Reply-To: <957ea888-a96b-89cc-29e2-973bb9e36f40@intel.com>

On 02.07.23 11:50, Yin, Fengwei wrote:
> 
> 
> On 7/1/2023 9:17 AM, Zi Yan wrote:
>> In kernel, almost all code only cares: 1) if a page/folio has extra pins
>> by checking if mapcount is equal to refcount + extra, and 2)
>> if a page/folio is mapped multiple times. A single mapcount can meet
>> these two needs.
> For 2, how can we know whether a page/folio is mapped multiple times from
> single mapcount? My understanding is we need two counts as folio could be
> partial mapped.

Yes, a single mapcount is most probably insufficient. I started 
analyzing all existing users and use cases, trying to avoid walking page 
tables.

If we want to get rid of all of (most) sub-page mapcounts, we'd probably 
want:

(1) Total mapcount (compound + any sub-page): page_mapped(), pagecount
     vs. refcount games, ...

(2) Compound mapcount (for PMD/PUD-mappale THP only): (2) - (1) tells
     you if it's only PMD mapped or also PTE-mapped. For example, for
     statistics but also swapout code.

(3) Mapcount of first (or any other) subpage (compount+subpage): for
     folio_estimated_sharers().

For anon pages, I'm thinking about remembering an additional

(1) Page/folio creator (MM pointer/identification)
(2) Page/folio creator mapcount

When optimizing a PTE-mapped THP (especially not- pmd-mappale) for the 
fork()+exec() case, we'd have to walk page tables to see if all folio 
references come from this MM. The page/folio creator exactly avoids that 
completely. We might need a mechanism to synchronize against 
mapping/unmapping of this folio from the creator concurrently (relevant 
when mapped into multiple page tables).

Further, for (1) we'd want a 64bit mapcount for large folios, which 
implies a 64bit refcount. For smallish folios, we don't really care.

We should most probably use a bi-weekly MM meeting to discuss that.

Hopefully, I have a full understanding of all use cases and requirements 
until then. Don't have sufficient time to look into all the nasty 
details right now.

-- 
Cheers,

David / dhildenb