Re: [RFC 2/4] filemap: use minimum order while allocating folios

From: Pankaj Raghav
Date: Wed Jun 21 2023 - 06:26:10 EST


>> index 47afbca1d122..090b810ddeed 100644
>> --- a/mm/readahead.c
>> +++ b/mm/readahead.c
>> @@ -245,7 +245,8 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
>>               continue;
>>           }
>>   -        folio = filemap_alloc_folio(gfp_mask, 0);
>> +        folio = filemap_alloc_folio(gfp_mask,
>> +                        mapping_min_folio_order(mapping));
>>           if (!folio)
>>               break;
>>           if (filemap_add_folio(mapping, folio, index + i,
>> @@ -259,7 +260,8 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
>>           if (i == nr_to_read - lookahead_size)
>>               folio_set_readahead(folio);
>>           ractl->_workingset |= folio_test_workingset(folio);
>> -        ractl->_nr_pages++;
>> +        ractl->_nr_pages += folio_nr_pages(folio);
>> +        i += folio_nr_pages(folio) - 1;
>>       }
>>         /*
> This is incomplete, as the loop above has some exit statements which blindly step backwards by one
> page.
>
> I found it better to rework the 'for' into a 'while' loop; please check the attached patch.
>
Taken from your patch:

@@ -240,8 +240,8 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
* not worth getting one just for that.
*/
read_pages(ractl);
- ractl->_index++;
- i = ractl->_index + ractl->_nr_pages - index - 1;
+ ractl->_index += folio_nr_pages(folio);
+ i = ractl->_index + ractl->_nr_pages - index;

IIUC, we don't need to update the _index after read_pages() as it already modifies it. We just need
to move ractl->_index by 1 to move to the next index.


> Cheers,
>
> Hannes