mm: migrate: do not touch page->mem_cgroup of live pages

Changing a page's memcg association complicates dealing with the page,
so we want to limit this as much as possible.  Page migration e.g.  does
not have to do that.  Just like page cache replacement, it can forcibly
charge a replacement page, and then uncharge the old page when it gets
freed.  Temporarily overcharging the cgroup by a single page is not an
issue in practice, and charging is so cheap nowadays that this is much
preferrable to the headache of messing with live pages.

The only place that still changes the page->mem_cgroup binding of live
pages is when pages move along with a task to another cgroup.  But that
path isolates the page from the LRU, takes the page lock, and the move
lock (lock_page_memcg()).  That means page->mem_cgroup is always stable
in callers that have the page isolated from the LRU or locked.  Lighter
unlocked paths, like writeback accounting, can use lock_page_memcg().

[akpm@linux-foundation.org: fix build]
[vdavydov@virtuozzo.com: fix lockdep splat]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vladimir Davydov <vdavydov@virtuozzo.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Greg Thelen <gthelen@google.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

4 files changed, 18 insertions, 14 deletions

diff --git a/mm/filemap.c b/mm/filemap.c
index ee8140c..d8317ca 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -577,7 +577,7 @@ int replace_page_cache_page(struct page *old, struct page *new, gfp_t gfp_mask)
 			__inc_zone_page_state(new, NR_SHMEM);
 		spin_unlock_irqrestore(&mapping->tree_lock, flags);
 		unlock_page_memcg(memcg);
-		mem_cgroup_replace_page(old, new);
+		mem_cgroup_migrate(old, new);
 		radix_tree_preload_end();
 		if (freepage)
 			freepage(old);
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 864e237..64506b2 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -4457,7 +4457,7 @@ static int mem_cgroup_move_account(struct page *page,
 	VM_BUG_ON(compound && !PageTransHuge(page));
 
 	/*
-	 * Prevent mem_cgroup_replace_page() from looking at
+	 * Prevent mem_cgroup_migrate() from looking at
 	 * page->mem_cgroup of its source page while we change it.
 	 */
 	ret = -EBUSY;
@@ -5486,16 +5486,17 @@ void mem_cgroup_uncharge_list(struct list_head *page_list)
 }
 
 /**
- * mem_cgroup_replace_page - migrate a charge to another page
- * @oldpage: currently charged page
- * @newpage: page to transfer the charge to
+ * mem_cgroup_migrate - charge a page's replacement
+ * @oldpage: currently circulating page
+ * @newpage: replacement page
  *
- * Migrate the charge from @oldpage to @newpage.
+ * Charge @newpage as a replacement page for @oldpage. @oldpage will
+ * be uncharged upon free.
  *
  * Both pages must be locked, @newpage->mapping must be set up.
  * Either or both pages might be on the LRU already.
  */
-void mem_cgroup_replace_page(struct page *oldpage, struct page *newpage)
+void mem_cgroup_migrate(struct page *oldpage, struct page *newpage)
 {
 	struct mem_cgroup *memcg;
 	unsigned int nr_pages;
diff --git a/mm/migrate.c b/mm/migrate.c
index 432ecd0..848327d 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -326,12 +326,13 @@ int migrate_page_move_mapping(struct address_space *mapping,
 			return -EAGAIN;
 
 		/* No turning back from here */
-		set_page_memcg(newpage, page_memcg(page));
 		newpage->index = page->index;
 		newpage->mapping = page->mapping;
 		if (PageSwapBacked(page))
 			SetPageSwapBacked(newpage);
 
+		mem_cgroup_migrate(page, newpage);
+
 		return MIGRATEPAGE_SUCCESS;
 	}
 
@@ -373,7 +374,6 @@ int migrate_page_move_mapping(struct address_space *mapping,
 	 * Now we know that no one else is looking at the page:
 	 * no turning back from here.
 	 */
-	set_page_memcg(newpage, page_memcg(page));
 	newpage->index = page->index;
 	newpage->mapping = page->mapping;
 	if (PageSwapBacked(page))
@@ -428,6 +428,8 @@ int migrate_page_move_mapping(struct address_space *mapping,
 	}
 	local_irq_enable();
 
+	mem_cgroup_migrate(page, newpage);
+
 	return MIGRATEPAGE_SUCCESS;
 }
 
@@ -458,9 +460,9 @@ int migrate_huge_page_move_mapping(struct address_space *mapping,
 		return -EAGAIN;
 	}
 
-	set_page_memcg(newpage, page_memcg(page));
 	newpage->index = page->index;
 	newpage->mapping = page->mapping;
+
 	get_page(newpage);
 
 	radix_tree_replace_slot(pslot, newpage);
@@ -468,6 +470,9 @@ int migrate_huge_page_move_mapping(struct address_space *mapping,
 	page_unfreeze_refs(page, expected_count - 1);
 
 	spin_unlock_irq(&mapping->tree_lock);
+
+	mem_cgroup_migrate(page, newpage);
+
 	return MIGRATEPAGE_SUCCESS;
 }
 
@@ -775,7 +780,6 @@ static int move_to_new_page(struct page *newpage, struct page *page,
 	 * page is freed; but stats require that PageAnon be left as PageAnon.
 	 */
 	if (rc == MIGRATEPAGE_SUCCESS) {
-		set_page_memcg(page, NULL);
 		if (!PageAnon(page))
 			page->mapping = NULL;
 	}
@@ -1842,8 +1846,7 @@ fail_putback:
 	}
 
 	mlock_migrate_page(new_page, page);
-	set_page_memcg(new_page, page_memcg(page));
-	set_page_memcg(page, NULL);
+	mem_cgroup_migrate(page, new_page);
 	page_remove_rmap(page, true);
 	set_page_owner_migrate_reason(new_page, MR_NUMA_MISPLACED);
 
diff --git a/mm/shmem.c b/mm/shmem.c
index 440e2a7..1acfdbc 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1116,7 +1116,7 @@ static int shmem_replace_page(struct page **pagep, gfp_t gfp,
 		 */
 		oldpage = newpage;
 	} else {
-		mem_cgroup_replace_page(oldpage, newpage);
+		mem_cgroup_migrate(oldpage, newpage);
 		lru_cache_add_anon(newpage);
 		*pagep = newpage;
 	}