Re: [PATCH 3/3] mm: Optimize mremap() by PTE batching

[Dev Jain <dev.jain@arm.com>]

On 06/05/25 3:40 pm, Anshuman Khandual wrote:
> On 5/6/25 10:30, Dev Jain wrote:
>> Use folio_pte_batch() to optimize move_ptes(). Use get_and_clear_full_ptes()
>> so as to elide TLBIs on each contig block, which was previously done by
>> ptep_get_and_clear().
>>
>> Signed-off-by: Dev Jain <dev.jain@arm.com>
>> ---
>>   mm/mremap.c | 24 +++++++++++++++++++-----
>>   1 file changed, 19 insertions(+), 5 deletions(-)
>>
>> diff --git a/mm/mremap.c b/mm/mremap.c
>> index 1a08a7c3b92f..3621c07d8eea 100644
>> --- a/mm/mremap.c
>> +++ b/mm/mremap.c
>> @@ -176,7 +176,7 @@ static int move_ptes(struct pagetable_move_control *pmc,
>>   	struct vm_area_struct *vma = pmc->old;
>>   	bool need_clear_uffd_wp = vma_has_uffd_without_event_remap(vma);
>>   	struct mm_struct *mm = vma->vm_mm;
>> -	pte_t *old_ptep, *new_ptep, pte;
>> +	pte_t *old_ptep, *new_ptep, old_pte, pte;
>>   	pmd_t dummy_pmdval;
>>   	spinlock_t *old_ptl, *new_ptl;
>>   	bool force_flush = false;
>> @@ -185,6 +185,7 @@ static int move_ptes(struct pagetable_move_control *pmc,
>>   	unsigned long old_end = old_addr + extent;
>>   	unsigned long len = old_end - old_addr;
>>   	int err = 0;
>> +	int nr;
>>   
>>   	/*
>>   	 * When need_rmap_locks is true, we take the i_mmap_rwsem and anon_vma
>> @@ -237,10 +238,14 @@ static int move_ptes(struct pagetable_move_control *pmc,
>>   
>>   	for (; old_addr < old_end; old_ptep++, old_addr += PAGE_SIZE,
>>   				   new_ptep++, new_addr += PAGE_SIZE) {
>> -		if (pte_none(ptep_get(old_ptep)))
>> +		const fpb_t fpb_flags = FPB_IGNORE_DIRTY | FPB_IGNORE_SOFT_DIRTY;
>> +		int max_nr = (old_end - old_addr) >> PAGE_SHIFT;
>> +
>> +		nr = 1;
>> +		old_pte = ptep_get(old_ptep);
>> +		if (pte_none(old_pte))
>>   			continue;
>>   
>> -		pte = ptep_get_and_clear(mm, old_addr, old_ptep);
>>   		/*
>>   		 * If we are remapping a valid PTE, make sure
>>   		 * to flush TLB before we drop the PTL for the
>> @@ -252,8 +257,17 @@ static int move_ptes(struct pagetable_move_control *pmc,
>>   		 * the TLB entry for the old mapping has been
>>   		 * flushed.
>>   		 */
>> -		if (pte_present(pte))
>> +		if (pte_present(old_pte)) {
>> +			if ((max_nr != 1) && maybe_contiguous_pte_pfns(old_ptep, old_pte)) {
> 
> maybe_contiguous_pte_pfns() cost will be applicable for memory
> areas greater than a single PAGE_SIZE (i.e max_nr != 1) ? This
> helper extracts an additional consecutive pte, ensures that it
> is valid mapped and extracts pfn before comparing for the span.
> 
> There is some cost associated with the above code sequence which
> looks justified for sequential access of memory buffers that has
> consecutive physical memory backing.

I did not see any regression for the simple case of mremapping base pages.

> But what happens when such
> buffers are less probable, will those buffers take a performance
> hit for all the comparisons that just turn out to be negative ?

When would that be the case? We are remapping consecutive ptes to 
consecutive ptes.

> 
>> +				struct folio *folio = vm_normal_folio(vma, old_addr, old_pte);
>> +
>> +				if (folio && folio_test_large(folio))
>> +					nr = folio_pte_batch(folio, old_addr, old_ptep,
>> +					old_pte, max_nr, fpb_flags, NULL, NULL, NULL);
>> +			}
>>   			force_flush = true;
>> +		}
>> +		pte = get_and_clear_full_ptes(mm, old_addr, old_ptep, nr, 0);
>>   		pte = move_pte(pte, old_addr, new_addr);
>>   		pte = move_soft_dirty_pte(pte);
>>   
>> @@ -266,7 +280,7 @@ static int move_ptes(struct pagetable_move_control *pmc,
>>   				else if (is_swap_pte(pte))
>>   					pte = pte_swp_clear_uffd_wp(pte);
>>   			}
>> -			set_pte_at(mm, new_addr, new_ptep, pte);
>> +			set_ptes(mm, new_addr, new_ptep, pte, nr);
>>   		}
>>   	}
>>