Re: [PATCH linux-next] mm: huge_memory: fix misused mapping_large_folio_support() for anon folios

[Zi Yan <ziy@nvidia.com>]

[-- Attachment #1: Type: text/plain, Size: 4259 bytes --]

+Luis and Pankaj, who are working on enable bs > ps in XFS and touch split_huge_page_to_list_to_order().


On 4 Jun 2024, at 6:52, Zi Yan wrote:

> On 4 Jun 2024, at 0:57, David Hildenbrand wrote:
>
>> On 04.06.24 07:47, xu.xin16@zte.com.cn wrote:
>>> From: Ran Xiaokai <ran.xiaokai@zte.com.cn>
>>>
>>> When I did a large folios split test, a WARNING
>>> "[ 5059.122759][  T166] Cannot split file folio to non-0 order"
>>> was triggered. But my test cases are only for anonmous folios.
>>> while mapping_large_folio_support() is only reasonable for page
>>> cache folios.
>>
>> Agreed.
>>
>> I wonder if mapping_large_folio_support() should either
>>
>> a) Complain if used for anon folios, so we can detect the wrong use more easily. (VM_WARN_ON_ONCE())
>
> This is much better.
>
>>
>> b) Return "true" for anonymous mappings, although that's more debatable.
>
> This might fix the warning here, but the function might get wrong uses easily.
>
>>
>>>
>>> In split_huge_page_to_list_to_order(), the folio passed to
>>> mapping_large_folio_support() maybe anonmous folio. The
>>> folio_test_anon() check is missing. So the split of the anonmous THP
>>> is failed. This is also the same for shmem_mapping(). We'd better add
>>> a check for both. But the shmem_mapping() in __split_huge_page() is
>>> not involved, as for anonmous folios, the end parameter is set to -1, so
>>> (head[i].index >= end) is always false. shmem_mapping() is not called.
>>>
>>> Using /sys/kernel/debug/split_huge_pages to verify this, with this
>>> patch, large anon THP is successfully split and the warning is ceased.
>>>
>>> Signed-off-by: Ran Xiaokai <ran.xiaokai@zte.com.cn>
>>> Cc: xu xin <xu.xin16@zte.com.cn>
>>> Cc: Yang Yang <yang.yang29@zte.com.cn>
>>> ---
>>>   mm/huge_memory.c | 38 ++++++++++++++++++++------------------
>>>   1 file changed, 20 insertions(+), 18 deletions(-)
>>>
>>> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
>>> index 317de2afd371..4c9c7e5ea20c 100644
>>> --- a/mm/huge_memory.c
>>> +++ b/mm/huge_memory.c
>>> @@ -3009,31 +3009,33 @@ int split_huge_page_to_list_to_order(struct page *page, struct list_head *list,
>>>   	if (new_order >= folio_order(folio))
>>>   		return -EINVAL;
>>>
>>> -	/* Cannot split anonymous THP to order-1 */
>>> -	if (new_order == 1 && folio_test_anon(folio)) {
>>> -		VM_WARN_ONCE(1, "Cannot split to order-1 folio");
>>> -		return -EINVAL;
>>> -	}
>>> -
>>>   	if (new_order) {
>>>   		/* Only swapping a whole PMD-mapped folio is supported */
>>>   		if (folio_test_swapcache(folio))
>>>   			return -EINVAL;
>>> -		/* Split shmem folio to non-zero order not supported */
>>> -		if (shmem_mapping(folio->mapping)) {
>>> -			VM_WARN_ONCE(1,
>>> -				"Cannot split shmem folio to non-0 order");
>>> -			return -EINVAL;
>>> -		}
>>> -		/* No split if the file system does not support large folio */
>>> -		if (!mapping_large_folio_support(folio->mapping)) {
>>> -			VM_WARN_ONCE(1,
>>> -				"Cannot split file folio to non-0 order");
>>> -			return -EINVAL;
>>> +
>>> +		if (folio_test_anon(folio)) {
>>> +			/* Cannot split anonymous THP to order-1 */
>>> +			if (new_order == 1) {
>>> +				VM_WARN_ONCE(1, "Cannot split to order-1 folio");
>>> +				return -EINVAL;
>>> +			}
>>> +		} else {
>>> +			/* Split shmem folio to non-zero order not supported */
>>> +			if (shmem_mapping(folio->mapping)) {
>>> +				VM_WARN_ONCE(1,
>>> +					"Cannot split shmem folio to non-0 order");
>>> +				return -EINVAL;
>>> +			}
>>> +			/* No split if the file system does not support large folio */
>>> +			if (!mapping_large_folio_support(folio->mapping)) {
>>> +				VM_WARN_ONCE(1,
>>> +					"Cannot split file folio to non-0 order");
>>> +				return -EINVAL;
>>> +			}
>>>   		}
>>>   	}
>>
>> What about the following sequence:
>>
>> if (folio_test_anon(folio)) {
>> 	if (new_order == 1)
>> 		...
>> } else if (new_order) {
>> 	if (shmem_mapping(...))
>> 		...
>> 	...
>> }
>>
>> if (folio_test_swapcache(folio) && new_order)
>> 	return -EINVAL;
>>
>> Should result in less churn and reduce indentation level.
>
> Yeah, this looks better to me.
>
> Best Regards,
> Yan, Zi


Best Regards,
Yan, Zi

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