Thread (29 messages) 29 messages, 6 authors, 2020-01-14

Re: [PATCH v7 01/10] mm/vmscan: remove unnecessary lruvec adding

From: Konstantin Khlebnikov <hidden>
Date: 2020-01-10 08:40:05
Also in: linux-mm, lkml

On 25/12/2019 12.04, Alex Shi wrote:
quoted hunk ↗ jump to hunk
We don't have to add a freeable page into lru and then remove from it.
This change saves a couple of actions and makes the moving more clear.

The SetPageLRU needs to be kept here for list intergrity. Otherwise:

     #0 mave_pages_to_lru                #1 release_pages
                                         if (put_page_testzero())
     if (put_page_testzero())
	                                    !PageLRU //skip lru_lock
                                                 list_add(&page->lru,);
     else
         list_add(&page->lru,) //corrupt

Signed-off-by: Alex Shi <redacted>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: yun.wang@linux.alibaba.com
Cc: linux-mm@kvack.org
Cc: linux-kernel@vger.kernel.org
---
  mm/vmscan.c | 16 +++++++---------
  1 file changed, 7 insertions(+), 9 deletions(-)
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 572fb17c6273..8719361b47a0 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1852,26 +1852,18 @@ static unsigned noinline_for_stack move_pages_to_lru(struct lruvec *lruvec,
Here is another cleanup: pass only pgdat as argument.

This function reavaluates lruvec for each page under lru lock.
Probably this is redundant for now but could be used in the future (or your patchset already use that).
  	while (!list_empty(list)) {
  		page = lru_to_page(list);
  		VM_BUG_ON_PAGE(PageLRU(page), page);
+		list_del(&page->lru);
  		if (unlikely(!page_evictable(page))) {
-			list_del(&page->lru);
  			spin_unlock_irq(&pgdat->lru_lock);
  			putback_lru_page(page);
  			spin_lock_irq(&pgdat->lru_lock);
  			continue;
  		}
-		lruvec = mem_cgroup_page_lruvec(page, pgdat);
-
Please leave a comment that we must set PageLRU before dropping our page reference.
quoted hunk ↗ jump to hunk
  		SetPageLRU(page);
-		lru = page_lru(page);
-
-		nr_pages = hpage_nr_pages(page);
-		update_lru_size(lruvec, lru, page_zonenum(page), nr_pages);
-		list_move(&page->lru, &lruvec->lists[lru]);
  
  		if (put_page_testzero(page)) {
  			__ClearPageLRU(page);
  			__ClearPageActive(page);
-			del_page_from_lru_list(page, lruvec, lru);
  
  			if (unlikely(PageCompound(page))) {
  				spin_unlock_irq(&pgdat->lru_lock);
@@ -1880,6 +1872,12 @@ static unsigned noinline_for_stack move_pages_to_lru(struct lruvec *lruvec,
  			} else
  				list_add(&page->lru, &pages_to_free);
  		} else {
+			lruvec = mem_cgroup_page_lruvec(page, pgdat);
+			lru = page_lru(page);
+			nr_pages = hpage_nr_pages(page);
+
+			update_lru_size(lruvec, lru, page_zonenum(page), nr_pages);
+			list_add(&page->lru, &lruvec->lists[lru]);
  			nr_moved += nr_pages;
  		}
IMHO It looks better to in this way:

SetPageLRU()

if (unlikely(put_page_testzero())) {
  <free>
  continue;
}

<add>
  	}
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help