Re: [PATCH v1 2/4] mm: Expose abnormal new_pte during move_ptes

[Pu Lehui <pulehui@huaweicloud.com>]

On 2025/5/30 3:19, Andrew Morton wrote:
> On Thu, 29 May 2025 15:56:48 +0000 Pu Lehui <pulehui@huaweicloud.com> wrote:
> 
>> From: Pu Lehui <pulehui@huawei.com>
>>
>> When executing move_ptes, the new_pte must be NULL, otherwise it will be
>> overwritten by the old_pte, and cause the abnormal new_pte to be leaked.
>> In order to make this problem to be more explicit, let's add
>> WARN_ON_ONCE when new_pte is not NULL.
>>
>> ...
>>
>> --- a/mm/mremap.c
>> +++ b/mm/mremap.c
>> @@ -237,6 +237,8 @@ static int move_ptes(struct pagetable_move_control *pmc,
>>   
>>   	for (; old_addr < old_end; old_pte++, old_addr += PAGE_SIZE,
>>   				   new_pte++, new_addr += PAGE_SIZE) {
>> +		WARN_ON_ONCE(!pte_none(*new_pte));
>> +
>>   		if (pte_none(ptep_get(old_pte)))
>>   			continue;
>>   
> 
> We now have no expectation that this will trigger, yes?  It's a sanity

Hi Andrew,

This can sanitize abnormal new_pte. It is expected that uprobe would not 
come in later, but others, uncertain🤔? So it will be a good alert. And 
after patch 1 it will not trigger WARNING.

> check that patch [1/4] is working?  Perhaps VM_WARN_ON_ONCE() would be

Agree, should I respin one more?

> more appropriate.  And maybe even a comment:
> 
> 	/* temporary, remove this one day */
>