Re: [PATCH v2] mm/migrate: don't batch the TLB flush when there is nothing to batch

[Zi Yan]

On Sun Nov 8, 2026 at 7:52 AM EST, Huang, Ying wrote:
> Zhiteng Qiu <qiuzhiteng@xxxxxxxx> writes:
>
<snip>
>
> IMHO, the description above is a bit too long.  Could you try to make it
> more concise?

I agree. It looks like AI generated.

>
>> Signed-off-by: Zhiteng Qiu <qiuzhiteng@xxxxxxxx>
>> ---

<snip>

>> @@ -1344,7 +1344,19 @@ static int migrate_folio_unmap(new_folio_t get_new_folio,
>>  		/* Establish migration ptes */
>>  		VM_BUG_ON_FOLIO(folio_test_anon(src) &&
>>  			       !folio_test_ksm(src) && !anon_vma, src);
>> -		try_to_migrate(src, mode == MIGRATE_ASYNC ? TTU_BATCH_FLUSH : 0);
>> +		enum ttu_flags ttu = mode == MIGRATE_ASYNC ? TTU_BATCH_FLUSH : 0;
>> +
>> +		/*
>> +		 * Batching only pays off when the rmap walk can meet more than
>> +		 * one PTE.  A lone order-0 folio with a single mapping has
>> +		 * exactly one: batching it gains nothing, and it leaves
>> +		 * mm->tlb_flush_batched pending, so later page table changes
>> +		 * in this mm (e.g. NUMA scanning) flush the whole mm.

This comment can be simplified.

>> +		 */
>> +		if (single_folio && !folio_test_large(src) &&
>> +		    !folio_maybe_mapped_shared(src))
>> +			ttu &= ~TTU_BATCH_FLUSH;
>
> Is it possible to use a simplified version as below?
>
>         enum ttu_flags ttu = 0;
>
>         if (mode == MIGRATE_ASYNC &&
>             (!single_folio || folio_mapcount(src) > 1))

LGTM, but I would keep folio_maybe_mapped_shared() to clarify the
intent. Would (!single_folio || folio_maybe_mapped_shared(src)) work?

>                 ttu = TTU_BATCH_FLUSH;
>

-- 
Best Regards,
Yan, Zi