Re: [PATCH v3 2/4] mm/mmap/vma_merge: fold curr, next assignment logic

[Vlastimil Babka]

On 3/22/23 10:06, Vlastimil Babka wrote:
> On 3/22/23 08:13, Lorenzo Stoakes wrote:
>> Use find_vma_intersection() and vma_lookup() to both simplify the logic and
>> to fold the end == next->vm_start condition into one block.
>> 
>> This groups all of the simple range checks together and establishes the
>> invariant that, if prev, curr or next are non-NULL then their positions are
>> as expected.
>> 
>> This has no functional impact.
> 
> I'm not so sure...

Disregard that, vma_lookup() is not find_vma(). Next cleanup challenge: more
obvious naming of those two.

>> Signed-off-by: Lorenzo Stoakes <lstoakes@xxxxxxxxx>

Reviewed-by: Vlastimil Babka <vbabka@xxxxxxx>

>> ---
>>  mm/mmap.c | 24 +++++++++++-------------
>>  1 file changed, 11 insertions(+), 13 deletions(-)
>> 
>> diff --git a/mm/mmap.c b/mm/mmap.c
>> index c9834364ac98..dbdbb92493b2 100644
>> --- a/mm/mmap.c
>> +++ b/mm/mmap.c
>> @@ -930,15 +930,14 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
>>  	if (vm_flags & VM_SPECIAL)
>>  		return NULL;
>> 
>> -	curr = find_vma(mm, prev ? prev->vm_end : 0);
>> -	if (curr && curr->vm_end == end)		/* cases 6, 7, 8 */
>> -		next = find_vma(mm, curr->vm_end);
>> -	else
>> -		next = curr;
>> +	/* Does the input range span an existing VMA? (cases 5 - 8) */
>> +	curr = find_vma_intersection(mm, prev ? prev->vm_end : 0, end);
>> 
>> -	/* In cases 1 - 4 there's no CCCC vma */
>> -	if (curr && end <= curr->vm_start)
>> -		curr = NULL;
>> +	if (!curr ||			/* cases 1 - 4 */
>> +	    end == curr->vm_end)	/* cases 6 - 8, adjacent VMA */
>> +		next = vma_lookup(mm, end);
> 
> AFAICS if the next vma is not adjacent to CCCC or ****, but there's a gap,
> this will still give you a non-NULL result?
> 
>> +	else
>> +		next = NULL;		/* case 5 */
>> 
>>  	/* verify some invariant that must be enforced by the caller */
>>  	VM_WARN_ON(prev && addr <= prev->vm_start);
>> @@ -959,11 +958,10 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
>>  		}
>>  	}
>>  	/* Can we merge the successor? */
>> -	if (next && end == next->vm_start &&
> 
> And then without this end == next->vm_start check, merge will be done
> despite the gap.
> 
>> -			mpol_equal(policy, vma_policy(next)) &&
>> -			can_vma_merge_before(next, vm_flags,
>> -					     anon_vma, file, pgoff+pglen,
>> -					     vm_userfaultfd_ctx, anon_name)) {
>> +	if (next && mpol_equal(policy, vma_policy(next)) &&
>> +	    can_vma_merge_before(next, vm_flags,
>> +				 anon_vma, file, pgoff+pglen,
>> +				 vm_userfaultfd_ctx, anon_name)) {
>>  		merge_next = true;
>>  	}
>> 
>> --
>> 2.39.2
>