Re: [PATCH] mm: hugetlb: independent PMD page table shared count

[David Hildenbrand <david@redhat.com>]

On 17.12.24 03:02, Liu Shixin wrote:
> 
> 
> On 2024/12/16 23:34, David Hildenbrand wrote:
>> On 14.12.24 11:44, Liu Shixin wrote:
>>> The folio refcount may be increased unexpectly through try_get_folio() by
>>> caller such as split_huge_pages. In huge_pmd_unshare(), we use refcount to
>>> check whether a pmd page table is shared. The check is incorrect if the
>>> refcount is increased by the above caller, and this can cause the page
>>> table leaked:
>>
>> Are you sure it is "leaked" ?
>>
>> I assume what happens is that we end up freeing a page table without calling its constructor. That's why page freeing code complains about "nonzero mapcount" (overlayed by something else).
> 
> 1. The page table itself will be discarded after reporting the "nonzero mapcount".
> 
> 2. The HugeTLB page mapped by the page table miss freeing since we treat the page table as shared
>      and a shared page table will not be to unmap.

Ah, the page table still maps something, that makes sense. So we're 
leaking that indeed.

-- 
Cheers,

David / dhildenb