Re: [syzbot ci] Re: mm/hugetlb: fix possible deadlock in __hugetlb_zap_begin

[Lance Yang]

On 2025/11/10 17:15, syzbot ci wrote:
> syzbot ci has tested the following series
>
> [v1] mm/hugetlb: fix possible deadlock in __hugetlb_zap_begin
> https://lore.kernel.org/all/20251110051421.29436-1-lance.yang@xxxxxxxxx
> * [PATCH 1/1] mm/hugetlb: fix possible deadlock in __hugetlb_zap_begin
>
> and found the following issues:
> * possible deadlock in move_hugetlb_page_tables
> * possible deadlock in remove_inode_hugepages
> * possible deadlock in unmap_vmas

Oops, this fix was incomplete, as other code paths with the reverse locking
order still exist, leading to the deadlocks reported by syzbot ...

Let me take a closer look ...