Received-SPF: pass (google.com: domain of linux-kernel+bounces-65560-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
From: Zi Yan <ziy@nvidia.com>
To: David Hildenbrand <david@redhat.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>,
 "\"Pankaj Raghav (Samsung)\"" <kernel@pankajraghav.com>, linux-mm@kvack.org,
 "\"Matthew Wilcox (Oracle)\"" <willy@infradead.org>,
 Yang Shi <shy828301@gmail.com>, Yu Zhao <yuzhao@google.com>,
 "\"Kirill A . Shutemov\"" <kirill.shutemov@linux.intel.com>,
 =?utf-8?q?=22Michal_Koutn=C3=BD=22?= <mkoutny@suse.com>,
 Roman Gushchin <roman.gushchin@linux.dev>,
 "\"Zach O'Keefe\"" <zokeefe@google.com>, Hugh Dickins <hughd@google.com>,
 Mcgrof Chamberlain <mcgrof@kernel.org>,
 Andrew Morton <akpm@linux-foundation.org>, linux-kernel@vger.kernel.org,
 cgroups@vger.kernel.org, linux-fsdevel@vger.kernel.org,
 linux-kselftest@vger.kernel.org
Subject: Re: [PATCH v4 0/7] Split a folio to any lower order folios
Date: Wed, 14 Feb 2024 11:35:50 -0500
Message-ID: <3D59E86D-0B66-4829-AE33-280317F913D8@nvidia.com>
In-Reply-To: <46f61262-e197-456c-97f1-d464ad5688c1@redhat.com>
References: <20240213215520.1048625-1-zi.yan@sent.com>
 <659e1abb-40d0-42ba-ba0a-8256d7eb1c5a@redhat.com>
 <F4470D3A-DC2C-4A6A-B65C-1C94D732A60E@nvidia.com>
 <66d4b27f-85e4-458e-8d66-54f800c5c65f@arm.com>
 <46f61262-e197-456c-97f1-d464ad5688c1@redhat.com>
Content-Type: multipart/signed;
 boundary="=_MailMate_F12DAB28-7693-40ED-B067-53D889A5D109_=";
 micalg=pgp-sha512; protocol="application/pgp-signature"
Precedence: bulk
MIME-Version: 1.0
X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1
X-MS-Exchange-AntiSpam-MessageData-0:
	=?us-ascii?Q?72lrftXv0rGSbFRfhAjsOIMClmX48KUJtPewWVXdvbTz9/RQXetk9QnoVo7t?=
 =?us-ascii?Q?2mwNmNF34D3nVNYs5CZ5+xjjGZRMaiJvGKFH4VhkdewC1U/PGkwlNwrSO2iF?=
 =?us-ascii?Q?p471dg0XbZ4Tp4lRjWefkegdjOeZijZl4UB0bYklSdnrVFE2l0TshuZ166XH?=
 =?us-ascii?Q?hg03WB7gX3RInAEV0XWFISzKJAhwjF1mNnVapjlczxuAP2QRcFDt/D16zQnZ?=
 =?us-ascii?Q?mTev8PO7MqJMnHg1KJD2/sg4/sYzlVRfE2BiHtO7bRRO2FpWWbAmn36w13C9?=
 =?us-ascii?Q?iWB2mSRu/WoA4x05JUrcvHMAZaGNqiNau/VqdsgZQ1fUpeW+RrcTT0PuNzCh?=
 =?us-ascii?Q?b3OcbLQy4UvzVKOltwUTJ5iDnvfHmhf+aDeUUBxXw4IdzdeElg5kRtw8Efsv?=
 =?us-ascii?Q?C/Pe527bCWZyT1+UZSrjLaUxU70W0FzgIZncsauVasZqFRQIQPFBuo7YC6Yd?=
 =?us-ascii?Q?5lzaylKG+jDMhU5wW+nyBbfcoF9tXUfInp8Zj4UU39Lr6zYQQqQ26ZOu1a9g?=
 =?us-ascii?Q?r7vUKy4GcVHj7JS+KpCHeXoDfQGyknbnPbm9VB0gwF6S5kklIjVxtLBd89CZ?=
 =?us-ascii?Q?C8+FUrITwP9auxqMSv9mXyZb2iKYwKcgag/tXVajnJRMyeu/Kmx0QsEF3/lO?=
 =?us-ascii?Q?n9Uipkzfe35v73PlyCtAcufu/xYWsPi4OfiRf05bGFDKco4z9D8U6O720ZZs?=
 =?us-ascii?Q?rJ92RkdyTzaYeFcKv2se2TlnG267/5cmipOa1hIMMX98+VEdId9h7XcMcECu?=
 =?us-ascii?Q?ntkRS+zM5W6irroG+/UcWzZ9RMbcynt2j/rLjHCmkT+Sjc7Stjio/aLtgWcm?=
 =?us-ascii?Q?YScobhIPubGkftbzCpkK/BieVfPAS0fsxu+lmyEgMqbf0/Tnh85nqflM6EC6?=
 =?us-ascii?Q?jDYzxxd3AsM8TTd+pOlSSyiLnU/jjUmJ3zdj1LtmQA3XRYBvkc4kxwqhcBfP?=
 =?us-ascii?Q?RhwfzvgMDvjMU1J/7Kvcgg0SD9uov8du9dAd+gPbiRTMkSbL5ymFjT17/zcn?=
 =?us-ascii?Q?3AmnONh1+eyrQXmkUiqK3eP0Z9XhFe3/VhHV8Or7E5xMrdNPZ+CVQCtHg/z1?=
 =?us-ascii?Q?gER2fePCRw6mMxUS/TOC8jPLrDAMH9BkjKuSL3ET10NBQRvpXtFlttxUBY8Y?=
 =?us-ascii?Q?Or6Mms9jPCsI97T7Yq03BoOKQNQC4gPxHjEuQo5JhdtktlTfvojrh2/bqbWr?=
 =?us-ascii?Q?mxsDPRDFXZGpBdAPg6B8cnMZwZ5oNK27+dk5GThoKFfzStVOVTkdAmdj4a0N?=
 =?us-ascii?Q?IjwphT9fjDPUl2Hsyg+p1A293EJt15JF7FbXo/ExH+1L8oh4PcqRYgr/YxGk?=
 =?us-ascii?Q?9yzyRgjPppCwbT9SFB7OBN7yeePGep4Tv44rTKGPgIm02dfCRaobODlrj03I?=
 =?us-ascii?Q?+o/3R+gUbC5OCt+TTUhxmo0yRREozOsAqYOaPcJc0JIVD49booNT5VW/qPUz?=
 =?us-ascii?Q?QfVX7krwvXI0Y5aCcWma2MpcT6iVmK4OIr5PTMkQcerZKG7zsDdrxrxoooDj?=
 =?us-ascii?Q?AVeAHpdrR6NgcZI9PGqcJe2JvZgtYnhZxig3QBAO5ldlk8fY1PCg0T8ImZ9p?=
 =?us-ascii?Q?f4SLOweV3zPiKjOLmoDw/3CQHrTdyMAGonYkXoG6?=
X-OriginatorOrg: Nvidia.com
X-MS-Exchange-CrossTenant-Network-Message-Id: 6c815335-ce51-45c1-d35d-08dc2d7b030a
X-MS-Exchange-CrossTenant-AuthSource: DS7PR12MB5744.namprd12.prod.outlook.com
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 14 Feb 2024 16:35:53.6322
 (UTC)
X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted
X-MS-Exchange-CrossTenant-Id: 43083d15-7273-40c1-b7db-39efd9ccc17a
X-MS-Exchange-CrossTenant-MailboxType: HOSTED
X-MS-Exchange-CrossTenant-UserPrincipalName: 0zxIAs04AC+DjKeMG9MgB1g7Pf0rkzzA9OYKtTjtUgSf5xFOAsp3qPtJUM6XwUNI
X-MS-Exchange-Transport-CrossTenantHeadersStamped: MW5PR12MB5682

--=_MailMate_F12DAB28-7693-40ED-B067-53D889A5D109_=
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

On 14 Feb 2024, at 5:55, David Hildenbrand wrote:

> On 14.02.24 11:50, Ryan Roberts wrote:
>> On 13/02/2024 22:31, Zi Yan wrote:
>>> On 13 Feb 2024, at 17:21, David Hildenbrand wrote:
>>>
>>>> On 13.02.24 22:55, Zi Yan wrote:
>>>>> From: Zi Yan <ziy@nvidia.com>
>>>>>
>>>>> Hi all,
>>>>>
>>>>> File folio supports any order and multi-size THP is upstreamed[1], =
so both
>>>>> file and anonymous folios can be >0 order. Currently, split_huge_pa=
ge()
>>>>> only splits a huge page to order-0 pages, but splitting to orders h=
igher than
>>>>> 0 is going to better utilize large folios. In addition, Large Block=

>>>>> Sizes in XFS support would benefit from it[2]. This patchset adds s=
upport for
>>>>> splitting a large folio to any lower order folios and uses it durin=
g file
>>>>> folio truncate operations.
>>>>>
>>>>> For Patch 6, Hugh did not like my approach to minimize the number o=
f
>>>>> folios for truncate[3]. I would like to get more feedback, especial=
ly
>>>>> from FS people, on it to decide whether to keep it or not.
>>>>
>>>> I'm curious, would it make sense to exclude the "more" controversial=
 parts (i.e., patch #6) for now, and focus on the XFS use case only?
>>>
>>> Sure. Patch 6 was there to make use of split_huge_page_to_list_to_ord=
er().
>>> Now we have multi-size THP and XFS use cases, it can be dropped.
>>
>> What are your plans for how to determine when to split THP and to what=
 order? I
>> don't see anything in this series that would split anon THP to non-zer=
o order?
>>
>> We have talked about using hints from user space in the past (e.g.  mr=
emap,
>> munmap, madvise, etc). But chrome has a use case where it temporarily =
mprotects
>> a single (4K) page as part of garbage collection (IIRC). If you eagerl=
y split on
>> that hint, you will have lost the benefits of the large folio when it =
later
>> mprotects back to the original setting.
>
> Not only that, splitting will make some of these operations more expens=
ive, possibly with no actual benefit.
>
>>
>> I guess David will suggest this would be a good use case for the khuge=
paged-lite
>> machanism we have been talking about. I dunno - it seems wasteful to s=
plit then
>> collapse again.
>
> I agree. mprotect() and even madvise(), ... might not be good candidate=
s for splitting. mremap() likely is, if the folio is mapped exclusively. =
MADV_DONTNEED/munmap()/mlock() might be good candidates (again, if mapped=
 exclusively). This will need a lot of thought I'm afraid (as you say, de=
ferred splitting is another example).

My initial use was for splitting 1GB THP to 2MB THP, but 1GB THP is not u=
pstream
yet. So for now, this might only be used by XFS. For anonymous large foli=
os,
we will use this when we find a justified use case. What I can think of i=
s
when a PMD-mapped THP happens to be split and the resulting order can be =
a HW/SW
favored order, like 64KB or 32KB (to be able to use contig PTE), we split=

to that order, otherwise, we still split to order-0.

--
Best Regards,
Yan, Zi

--=_MailMate_F12DAB28-7693-40ED-B067-53D889A5D109_=
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename=signature.asc
Content-Type: application/pgp-signature; name=signature.asc

-----BEGIN PGP SIGNATURE-----

iQJDBAEBCgAtFiEE6rR4j8RuQ2XmaZol4n+egRQHKFQFAmXM6+cPHHppeUBudmlk
aWEuY29tAAoJEOJ/noEUByhUVosP/0Fd8rJzrwcJQR1fD6IRLWQ2HLeASzM37Uec
5aSfg9tHn9hsNzYXmwaLcJhYjBRWWQ+9SBAfG+pdeYtudAxmOhmBECuf6evnE6uS
Obx4+f6xgXU19SiKe60q1OijOjIS2VAVcItGjA327xVkJ9vniPWYj4Wv0nRYcGaF
S0+z9I/H/7FkW+Tz9VVCYJqC/zDaWyMlImacY3ccX8OvNyiRiLwDI/R8G1jKRFt/
HghzYcPNlPRF2vgiW3fKR6Hrpy4Vo/nZs+0N5JPCNbQY/KHDx2Uc+bZbZetGFqak
4a63k/J3bAcaoRx1pFZeSWRxT1fpjHx0/r5EmCx/lmI78Zsz/HAzllEgAa5m4U+d
S3BST4m3SFpAEBc0oEIIzhr2S+ZRCHAQ7WiRYYtxiLiuDIoG+yA78qvUbHG3duIa
tt+ixRtryYyBidios+X0deLt+AVxsEfQA5ASy7mp4c7wZ7/E6b9dnQLD0YRhQ0Vt
ehguEnv7xtusxQQFS4Qc0wZrqWalkH9Z7beTFRFi3sD1+DZKY64vCoVCoWgSMBo6
R9xRyTk3xSHhLwR23xnQ1RG7FNWLoLa/qhxploO745CkgUD72hdolFSgii1x6rpf
jOgAGi3XHRhaFM//LOW9aVFKIOjVIEIBuT62Yu1p7ltuCi6DjJYEutbB9BCpDKKQ
S2Z7WO/a
=z7b4
-----END PGP SIGNATURE-----

--=_MailMate_F12DAB28-7693-40ED-B067-53D889A5D109_=--