Re: [PATCH] mm: stricter check on THP migration entry

[Miaohe Lin]

On 2022/6/20 12:48, Matthew Wilcox wrote:
> On Sat, Jun 18, 2022 at 03:32:43PM +0800, Miaohe Lin wrote:
>> When VM_BUG_ON check for THP migration entry, the existing code only check
>> thp_migration_supported case, but not for !thp_migration_supported case.
>> If !thp_migration_supported() and !pmd_present(), the original code may
>> dead loop in theory.  To make the VM_BUG_ON check more consistent, we need
>> to catch both cases.
> 
> This patch makes no sense to me.

Could you please explain it more? This patch is inspired by below commit indeed:
28b0ee3fb350 ("mm/gup.c: stricter check on THP migration entry during follow_pmd_mask").

Thanks.

> 
>> Signed-off-by: Miaohe Lin <linmiaohe@xxxxxxxxxx>
>> ---
>>  mm/memory.c | 7 +++----
>>  1 file changed, 3 insertions(+), 4 deletions(-)
>>
>> diff --git a/mm/memory.c b/mm/memory.c
>> index fee2884481f2..6a9a17c7f58a 100644
>> --- a/mm/memory.c
>> +++ b/mm/memory.c
>> @@ -5043,10 +5043,9 @@ static vm_fault_t __handle_mm_fault(struct vm_area_struct *vma,
>>  
>>  		barrier();
>>  		if (unlikely(is_swap_pmd(vmf.orig_pmd))) {
>> -			VM_BUG_ON(thp_migration_supported() &&
>> -					  !is_pmd_migration_entry(vmf.orig_pmd));
>> -			if (is_pmd_migration_entry(vmf.orig_pmd))
>> -				pmd_migration_entry_wait(mm, vmf.pmd);
>> +			VM_BUG_ON(!thp_migration_supported() ||
>> +				  !is_pmd_migration_entry(vmf.orig_pmd));
>> +			pmd_migration_entry_wait(mm, vmf.pmd);
>>  			return 0;
>>  		}
>>  		if (pmd_trans_huge(vmf.orig_pmd) || pmd_devmap(vmf.orig_pmd)) {
>> -- 
>> 2.23.0
>>
>>
> 
> .
>