mm: memcontrol: do not kill uncharge batching in free_pages_and_swap_cache

free_pages_and_swap_cache limits release_pages to PAGEVEC_SIZE chunks.
This is not a big deal for the normal release path but it completely kills
memcg uncharge batching which reduces res_counter spin_lock contention.
Dave has noticed this with his page fault scalability test case on a large
machine when the lock was basically dominating on all CPUs:

    80.18%    80.18%  [kernel]               [k] _raw_spin_lock
                  |
                  --- _raw_spin_lock
                     |
                     |--66.59%-- res_counter_uncharge_until
                     |          res_counter_uncharge
                     |          uncharge_batch
                     |          uncharge_list
                     |          mem_cgroup_uncharge_list
                     |          release_pages
                     |          free_pages_and_swap_cache
                     |          tlb_flush_mmu_free
                     |          |
                     |          |--90.12%-- unmap_single_vma
                     |          |          unmap_vmas
                     |          |          unmap_region
                     |          |          do_munmap
                     |          |          vm_munmap
                     |          |          sys_munmap
                     |          |          system_call_fastpath
                     |          |          __GI___munmap
                     |          |
                     |           --9.88%-- tlb_flush_mmu
                     |                     tlb_finish_mmu
                     |                     unmap_region
                     |                     do_munmap
                     |                     vm_munmap
                     |                     sys_munmap
                     |                     system_call_fastpath
                     |                     __GI___munmap

In his case the load was running in the root memcg and that part has been
handled by reverting 05b843012335 ("mm: memcontrol: use root_mem_cgroup
res_counter") because this is a clear regression, but the problem remains
inside dedicated memcgs.

There is no reason to limit release_pages to PAGEVEC_SIZE batches other
than lru_lock held times.  This logic, however, can be moved inside the
function.  mem_cgroup_uncharge_list and free_hot_cold_page_list do not
hold any lock for the whole pages_to_free list so it is safe to call them
in a single run.

The release_pages() code was previously breaking the lru_lock each
PAGEVEC_SIZE pages (ie, 14 pages).  However this code has no usage of
pagevecs so switch to breaking the lock at least every SWAP_CLUSTER_MAX
(32) pages.  This means that the lock acquisition frequency is
approximately halved and the max hold times are approximately doubled.

The now unneeded batching is removed from free_pages_and_swap_cache().

Also update the grossly out-of-date release_pages documentation.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reported-by: Dave Hansen <dave@sr71.net>
Cc: Vladimir Davydov <vdavydov@parallels.com>
Cc: Greg Thelen <gthelen@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

2 files changed, 23 insertions, 21 deletions

diff --git a/mm/swap.c b/mm/swap.c
index 6b2dc3897cd5..8a12b33936b4 100644
--- a/mm/swap.c
+++ b/mm/swap.c
@@ -887,18 +887,14 @@ void lru_add_drain_all(void)
 	mutex_unlock(&lock);
 }
 
-/*
- * Batched page_cache_release().  Decrement the reference count on all the
- * passed pages.  If it fell to zero then remove the page from the LRU and
- * free it.
- *
- * Avoid taking zone->lru_lock if possible, but if it is taken, retain it
- * for the remainder of the operation.
+/**
+ * release_pages - batched page_cache_release()
+ * @pages: array of pages to release
+ * @nr: number of pages
+ * @cold: whether the pages are cache cold
  *
- * The locking in this function is against shrink_inactive_list(): we recheck
- * the page count inside the lock to see whether shrink_inactive_list()
- * grabbed the page via the LRU.  If it did, give up: shrink_inactive_list()
- * will free it.
+ * Decrement the reference count on all the pages in @pages.  If it
+ * fell to zero, remove the page from the LRU and free it.
  */
 void release_pages(struct page **pages, int nr, bool cold)
 {
@@ -907,6 +903,7 @@ void release_pages(struct page **pages, int nr, bool cold)
 	struct zone *zone = NULL;
 	struct lruvec *lruvec;
 	unsigned long uninitialized_var(flags);
+	unsigned int uninitialized_var(lock_batch);
 
 	for (i = 0; i < nr; i++) {
 		struct page *page = pages[i];
@@ -920,6 +917,16 @@ void release_pages(struct page **pages, int nr, bool cold)
 			continue;
 		}
 
+		/*
+		 * Make sure the IRQ-safe lock-holding time does not get
+		 * excessive with a continuous string of pages from the
+		 * same zone. The lock is held only if zone != NULL.
+		 */
+		if (zone && ++lock_batch == SWAP_CLUSTER_MAX) {
+			spin_unlock_irqrestore(&zone->lru_lock, flags);
+			zone = NULL;
+		}
+
 		if (!put_page_testzero(page))
 			continue;
 
@@ -930,6 +937,7 @@ void release_pages(struct page **pages, int nr, bool cold)
 				if (zone)
 					spin_unlock_irqrestore(&zone->lru_lock,
 									flags);
+				lock_batch = 0;
 				zone = pagezone;
 				spin_lock_irqsave(&zone->lru_lock, flags);
 			}
diff --git a/mm/swap_state.c b/mm/swap_state.c
index ef1f39139b71..154444918685 100644
--- a/mm/swap_state.c
+++ b/mm/swap_state.c
@@ -265,18 +265,12 @@ void free_page_and_swap_cache(struct page *page)
 void free_pages_and_swap_cache(struct page **pages, int nr)
 {
 	struct page **pagep = pages;
+	int i;
 
 	lru_add_drain();
-	while (nr) {
-		int todo = min(nr, PAGEVEC_SIZE);
-		int i;
-
-		for (i = 0; i < todo; i++)
-			free_swap_cache(pagep[i]);
-		release_pages(pagep, todo, false);
-		pagep += todo;
-		nr -= todo;
-	}
+	for (i = 0; i < nr; i++)
+		free_swap_cache(pagep[i]);
+	release_pages(pagep, nr, false);
 }
 
 /*