Re: [PATCH 2/8] mm/compaction: correct last_migrated_pfn update in compact_zone

From: Kemeng Shi
Date: Mon Jul 31 2023 - 22:19:48 EST




on 8/1/2023 10:09 AM, Baolin Wang wrote:
>
>
> On 7/29/2023 1:10 AM, Kemeng Shi wrote:
>> We record start pfn of last isolated page block with last_migrated_pfn. And
>> then:
>> 1. We check if we mark the page block skip for exclusive access in
>> isolate_migratepages_block by test if next migrate pfn is still in last
>> isolated page block. If so, we will set finish_pageblock to do the rescan.
>> 2. We check if a full cc->order block is scanned by test if last scan range
>> passes the cc->order block boundary. If so, we flush the pages were freed.
>>
>> We treat cc->migrate_pfn before isolate_migratepages as the start pfn of
>> last isolated page range. However, we always align migrate_pfn to page block
>> or move to another page block in fast_find_migrateblock or in linearly scan
>> forward in isolate_migratepages before do page isolation in
>> isolate_migratepages_block.
>
> Right. But can you describe the impact in detail if the last_migrated_pfn is not set correctly? For example, this will result in rescan not being set correctly to miss a pageblock's rescanning.
>
> Otherwise looks good to me.
>
Sure, the impact will be added in next version. Thanks!
>> Update last_migrated_pfn with pageblock_start_pfn(cc->migrate_pfn - 1)
>> after scan to correctly set start pfn of last isolated page range. > Signed-off-by: Kemeng Shi <shikemeng@xxxxxxxxxxxxxxx>
>> ---
>>   mm/compaction.c | 3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/mm/compaction.c b/mm/compaction.c
>> index ce7841363b12..fb250c6b2b6e 100644
>> --- a/mm/compaction.c
>> +++ b/mm/compaction.c
>> @@ -2482,7 +2482,8 @@ compact_zone(struct compact_control *cc, struct capture_control *capc)
>>               goto check_drain;
>>           case ISOLATE_SUCCESS:
>>               update_cached = false;
>> -            last_migrated_pfn = iteration_start_pfn;
>> +            last_migrated_pfn = max(cc->zone->zone_start_pfn,
>> +                pageblock_start_pfn(cc->migrate_pfn - 1));
>>           }
>>             err = migrate_pages(&cc->migratepages, compaction_alloc,
>

--
Best wishes
Kemeng Shi