Re: [PATCH v2] mm: fault in complete folios instead of individual pages for tmpfs

[Baolin Wang <baolin.wang@linux.alibaba.com>]

On 2025/7/5 06:18, Andrew Morton wrote:
> On Fri,  4 Jul 2025 11:19:26 +0800 Baolin Wang <baolin.wang@linux.alibaba.com> wrote:
> 
>> After commit acd7ccb284b8 ("mm: shmem: add large folio support for tmpfs"),
>> tmpfs can also support large folio allocation (not just PMD-sized large
>> folios).
>>
>> However, when accessing tmpfs via mmap(), although tmpfs supports large folios,
>> we still establish mappings at the base page granularity, which is unreasonable.
>>
>> We can map multiple consecutive pages of a tmpfs folios at once according to
>> the size of the large folio. On one hand, this can reduce the overhead of page
>> faults; on the other hand, it can leverage hardware architecture optimizations
>> to reduce TLB misses, such as contiguous PTEs on the ARM architecture.
>>
>> Moreover, tmpfs mount will use the 'huge=' option to control large folio
>> allocation explicitly. So it can be understood that the process's RSS statistics
>> might increase, and I think this will not cause any obvious effects for users.
>>
>> Performance test:
>> I created a 1G tmpfs file, populated with 64K large folios, and write-accessed it
>> sequentially via mmap(). I observed a significant performance improvement:
> 
> That doesn't sound like a crazy thing to do.
> 
>> Before the patch:
>> real	0m0.158s
>> user	0m0.008s
>> sys	0m0.150s
>>
>> After the patch:
>> real	0m0.021s
>> user	0m0.004s
>> sys	0m0.017s
> 
> And look at that.
> 
>> diff --git a/mm/memory.c b/mm/memory.c
>> index 0f9b32a20e5b..9944380e947d 100644
>> --- a/mm/memory.c
>> +++ b/mm/memory.c
>> @@ -5383,10 +5383,10 @@ vm_fault_t finish_fault(struct vm_fault *vmf)
>>   
>>   	/*
>>   	 * Using per-page fault to maintain the uffd semantics, and same
>> -	 * approach also applies to non-anonymous-shmem faults to avoid
>> +	 * approach also applies to non shmem/tmpfs faults to avoid
>>   	 * inflating the RSS of the process.
>>   	 */
>> -	if (!vma_is_anon_shmem(vma) || unlikely(userfaultfd_armed(vma)) ||
>> +	if (!vma_is_shmem(vma) || unlikely(userfaultfd_armed(vma)) ||
>>   	    unlikely(needs_fallback)) {
>>   		nr_pages = 1;
>>   	} else if (nr_pages > 1) {
> 
> and that's it?
> 
> I'm itching to get this into -stable, really.  What LTS user wouldn't
> want this?  

This is an improvement rather than a bugfix, so I don't think it needs 
to go into LTS.

Could it be viewed as correcting an oversight in
> acd7ccb284b8?

Yes, I should have added this optimization in the series of the commit 
acd7ccb284b8. But obviously, I missed this :(.