Re: [PATCH v7 05/11] mm: split a folio in minimum folio order chunks

[Zi Yan <ziy@nvidia.com>]

[-- Attachment #1: Type: text/plain, Size: 2252 bytes --]

On 7 Jun 2024, at 16:30, Pankaj Raghav (Samsung) wrote:

> On Fri, Jun 07, 2024 at 12:58:33PM -0400, Zi Yan wrote:
>> Hi Pankaj,
>>
>> Can you use ziy@nvidia.com instead of zi.yan@sent.com? Since I just use the latter
>> to send patches. Thanks.
>
> Got it!
>
>>
>> On 7 Jun 2024, at 10:58, Pankaj Raghav (Samsung) wrote:
>>
>>> From: Luis Chamberlain <mcgrof@kernel.org>
>>>
>>> split_folio() and split_folio_to_list() assume order 0, to support
>>> minorder for non-anonymous folios, we must expand these to check the
>>> folio mapping order and use that.
>>>
>>> Set new_order to be at least minimum folio order if it is set in
>>> split_huge_page_to_list() so that we can maintain minimum folio order
>>> requirement in the page cache.
>>>
>>> Update the debugfs write files used for testing to ensure the order
>>> is respected as well. We simply enforce the min order when a file
>>> mapping is used.
>>>
>>> Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
>>> Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
>>> ---
>>>  include/linux/huge_mm.h | 14 ++++++++---
>>>  mm/huge_memory.c        | 55 ++++++++++++++++++++++++++++++++++++++---
>>>  2 files changed, 61 insertions(+), 8 deletions(-)
>>>
>>
>> <snip>
>>
>>>
>>> +int split_folio_to_list(struct folio *folio, struct list_head *list)
>>> +{
>>> +	unsigned int min_order = 0;
>>> +
>>> +	if (!folio_test_anon(folio)) {
>>> +		if (!folio->mapping) {
>>> +			count_vm_event(THP_SPLIT_PAGE_FAILED);
>>
>> You should only increase this counter when the input folio is a THP, namely
>> folio_test_pmd_mappable(folio) is true. For other large folios, we will
>> need a separate counter. Something like MTHP_STAT_FILE_SPLIT_FAILED.
>> See enum mthp_stat_item in include/linux/huge_mm.h.
>>
> Hmm, but we don't have mTHP support for non-anonymous memory right? In
> that case it won't be applicable for file backed memory?

Large folio support in page cache precedes mTHP (large anonymous folio),
thanks to willy's work. mTHP is more like a subset of large folio.
There is no specific counters for page cache large folio. If you think
it is worth tracking folios with orders between 0 and 9 (exclusive),
you can add counters. Matthew, what is your take on this?


--
Best Regards,
Yan, Zi

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 854 bytes --]