Re: [PATCH v5 2/8] mm/lru: replace pgdat lru_lock with lruvec lock

[Alex Shi <alex.shi@xxxxxxxxxxxxxxxxx>]

在 2019/12/10 下午9:41, Matthew Wilcox 写道:
> On Tue, Dec 10, 2019 at 07:46:18PM +0800, Alex Shi wrote:
>> -static void lock_page_lru(struct page *page, int *isolated)
>> +static struct lruvec *lock_page_lru(struct page *page, int *isolated)
>>  {
>> -	pg_data_t *pgdat = page_pgdat(page);
>> +	struct lruvec *lruvec = lock_page_lruvec_irq(page);
>>  
>> -	spin_lock_irq(&pgdat->lru_lock);
>>  	if (PageLRU(page)) {
>> -		struct lruvec *lruvec;
>>  
>> -		lruvec = mem_cgroup_page_lruvec(page, pgdat);
>>  		ClearPageLRU(page);
>>  		del_page_from_lru_list(page, lruvec, page_lru(page));
>>  		*isolated = 1;
>>  	} else
>>  		*isolated = 0;
>> +
>> +	return lruvec;
>>  }
> 
> I still don't understand how this is supposed to work for !PageLRU
> pages.  Which lruvec have you locked if this page isn't on an LRU?
> 

Good question. We could just fold it under PageLRU and no meaning changes.
Is this better to has this patch?

Thanks
Alex

commit 0f4b66d4a42397d57638352b738c3f9658003e44
Author: Alex Shi <alex.shi@xxxxxxxxxxxxxxxxx>
Date:   Wed Dec 11 11:31:53 2019 +0800

    mm/memcg: fold lock in lock_page_lru

    According to the calling path of commit_charge, the lrucare is bound
    with PageLRU, so we could just fold it under PageLRU. This has no
    functional change.

    Signed-off-by: Alex Shi <alex.shi@xxxxxxxxxxxxxxxxx>
    Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
    Cc: Michal Hocko <mhocko@xxxxxxxxxx>
    Cc: Vladimir Davydov <vdavydov.dev@xxxxxxxxx>
    Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
    Cc: cgroups@xxxxxxxxxxxxxxx
    Cc: linux-mm@xxxxxxxxx
    Cc: linux-kernel@xxxxxxxxxxxxxxx

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 833df0ce1bc1..4fe2252cf437 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -2622,9 +2622,10 @@ static void cancel_charge(struct mem_cgroup *memcg, unsigned int nr_pages)

 static struct lruvec *lock_page_lru(struct page *page, int *isolated)
 {
-       struct lruvec *lruvec = lock_page_lruvec_irq(page);
+       struct lruvec *lruvec = NULL;

        if (PageLRU(page)) {
+               lruvec = lock_page_lruvec_irq(page);

                ClearPageLRU(page);
                del_page_from_lru_list(page, lruvec, page_lru(page));
@@ -2638,17 +2639,18 @@ static struct lruvec *lock_page_lru(struct page *page, int *isolated)
 static void unlock_page_lru(struct page *page, int isolated,
                                struct lruvec *locked_lruvec)
 {
-       struct lruvec *lruvec;
+       if (isolated) {
+               struct lruvec *lruvec;

-       unlock_page_lruvec_irq(locked_lruvec);
-       lruvec = lock_page_lruvec_irq(page);
+               if (locked_lruvec)
+                       unlock_page_lruvec_irq(locked_lruvec);
+               lruvec = lock_page_lruvec_irq(page);

-       if (isolated) {
                VM_BUG_ON_PAGE(PageLRU(page), page);
                SetPageLRU(page);
                add_page_to_lru_list(page, lruvec, page_lru(page));
+               unlock_page_lruvec_irq(lruvec);
        }
-       unlock_page_lruvec_irq(lruvec);
 }

 static void commit_charge(struct page *page, struct mem_cgroup *memcg,