Re: [PATCH] mm/migrate: Split only transparent huge pages when allocation fails

[Anshuman Khandual]

On 09/24/2018 08:00 PM, Michal Hocko wrote:
> On Mon 24-09-18 19:44:55, Anshuman Khandual wrote:
>> When unmap_and_move[_huge_page] function fails due to lack of memory, the
>> splitting should happen only for transparent huge pages not for HugeTLB
>> pages. PageTransHuge() returns true for both THP and HugeTLB pages. Hence
>> the conditonal check should test PagesHuge() flag to make sure that given
>> pages is not a HugeTLB one.
> 
> Well spotted! Have you actually seen this happening or this is review
> driven? I am wondering what would be the real effect of this mismatch?
> I have tried to follow to code path but I suspect
> split_huge_page_to_list would fail for hugetlbfs pages. If there is a
> more serious effect then we should mark the patch for stable as well.

split_huge_page_to_list() fails on HugeTLB pages. I was experimenting around
moving 32MB contig HugeTLB pages on arm64 (with a debug patch applied) hit
the following stack trace when the kernel crashed.

[ 3732.462797] Call trace:
[ 3732.462835]  split_huge_page_to_list+0x3b0/0x858
[ 3732.462913]  migrate_pages+0x728/0xc20
[ 3732.462999]  soft_offline_page+0x448/0x8b0
[ 3732.463097]  __arm64_sys_madvise+0x724/0x850
[ 3732.463197]  el0_svc_handler+0x74/0x110
[ 3732.463297]  el0_svc+0x8/0xc
[ 3732.463347] Code: d1000400 f90b0e60 f2fbd5a2 a94982a1 (f9000420)