Re: [PATCH] mm: don't call lru draining in the nested lru_cache_disable

[Minchan Kim <minchan@kernel.org>]

On Mon, Dec 06, 2021 at 03:34:19PM -0800, Minchan Kim wrote:
> On Mon, Dec 06, 2021 at 03:04:21PM -0800, Andrew Morton wrote:
> > On Mon,  6 Dec 2021 14:10:06 -0800 Minchan Kim <minchan@kernel.org> wrote:
> > 
> > > lru_cache_disable involves IPIs to drain pagevec of each core,
> > > which sometimes takes quite long time to complete depending
> > > on cpu's business, which makes allocation too slow up to
> > > sveral hundredth milliseconds. Furthermore, the repeated draining
> > > in the alloc_contig_range makes thing worse considering caller
> > > of alloc_contig_range usually tries multiple times in the loop.
> > > 
> > > This patch makes the lru_cache_disable aware of the fact the
> > > pagevec was already disabled. With that, user of alloc_contig_range
> > > can disable the lru cache in advance in their context during the
> > > repeated trial so they can avoid the multiple costly draining
> > > in cma allocation.
> > 
> > Isn't this racy?
> >  
> > > ...
> > >
> > > @@ -859,7 +869,12 @@ atomic_t lru_disable_count = ATOMIC_INIT(0);
> > >   */
> > >  void lru_cache_disable(void)
> > >  {
> > > -	atomic_inc(&lru_disable_count);
> > > +	/*
> > > +	 * If someone is already disabled lru_cache, just return with
> > > +	 * increasing the lru_disable_count.
> > > +	 */
> > > +	if (atomic_inc_not_zero(&lru_disable_count))
> > > +		return;
> > >  #ifdef CONFIG_SMP
> > >  	/*
> > >  	 * lru_add_drain_all in the force mode will schedule draining on
> > > @@ -873,6 +888,7 @@ void lru_cache_disable(void)
> > >  #else
> > >  	lru_add_and_bh_lrus_drain();
> > >  #endif
> > 
> > There's a window here where lru_disable_count==0 and new pages can get
> > added to lru?
> 
> Indeed. If __lru_add_drain_all in core A didn't run yet but increased
> the disable count already, lru_cache_disable in core B will not see
> those pages in the LRU. Need to be fixed it.
> 
> Thanks, Andrew.

From 0874e108b4708355d703927716a49670b989e960 Mon Sep 17 00:00:00 2001
From: Minchan Kim <minchan@kernel.org>
Date: Mon, 6 Dec 2021 11:59:36 -0800
Subject: [PATCH v2] mm: don't call lru draining in the nested lru_cache_disable

lru_cache_disable involves IPIs to drain pagevec of each core,
which sometimes takes quite long time to complete depending
on cpu's business, which makes allocation too slow up to
sveral hundredth milliseconds. Furthermore, the repeated draining
in the alloc_contig_range makes thing worse considering caller
of alloc_contig_range usually tries multiple times in the loop.

This patch makes the lru_cache_disable aware of the fact the
pagevec was already disabled. With that, user of alloc_contig_range
can disable the lru cache in advance in their context during the
repeated trial so they can avoid the multiple costly draining
in cma allocation.

Signed-off-by: Minchan Kim <minchan@kernel.org>
---
 * from v1 - https://lore.kernel.org/lkml/20211206221006.946661-1-minchan@kernel.org/
   * fix lru_cache_disable race - akpm

 include/linux/swap.h | 14 ++------------
 mm/cma.c             |  5 +++++
 mm/swap.c            | 26 ++++++++++++++++++++++++--
 3 files changed, 31 insertions(+), 14 deletions(-)

diff --git a/include/linux/swap.h b/include/linux/swap.h
index ba52f3a3478e..fe18e86a4f13 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -348,19 +348,9 @@ extern void lru_note_cost_page(struct page *);
 extern void lru_cache_add(struct page *);
 extern void mark_page_accessed(struct page *);
 
-extern atomic_t lru_disable_count;
-
-static inline bool lru_cache_disabled(void)
-{
-	return atomic_read(&lru_disable_count);
-}
-
-static inline void lru_cache_enable(void)
-{
-	atomic_dec(&lru_disable_count);
-}
-
+extern bool lru_cache_disabled(void);
 extern void lru_cache_disable(void);
+extern void lru_cache_enable(void);
 extern void lru_add_drain(void);
 extern void lru_add_drain_cpu(int cpu);
 extern void lru_add_drain_cpu_zone(struct zone *zone);
diff --git a/mm/cma.c b/mm/cma.c
index 995e15480937..60be555c5b95 100644
--- a/mm/cma.c
+++ b/mm/cma.c
@@ -30,6 +30,7 @@
 #include <linux/cma.h>
 #include <linux/highmem.h>
 #include <linux/io.h>
+#include <linux/swap.h>
 #include <linux/kmemleak.h>
 #include <trace/events/cma.h>
 
@@ -453,6 +454,8 @@ struct page *cma_alloc(struct cma *cma, unsigned long count,
 	if (bitmap_count > bitmap_maxno)
 		goto out;
 
+	lru_cache_disable();
+
 	for (;;) {
 		spin_lock_irq(&cma->lock);
 		bitmap_no = bitmap_find_next_zero_area_off(cma->bitmap,
@@ -492,6 +495,8 @@ struct page *cma_alloc(struct cma *cma, unsigned long count,
 		start = bitmap_no + mask + 1;
 	}
 
+	lru_cache_enable();
+
 	trace_cma_alloc_finish(cma->name, pfn, page, count, align);
 
 	/*
diff --git a/mm/swap.c b/mm/swap.c
index af3cad4e5378..edcfcd6cf38e 100644
--- a/mm/swap.c
+++ b/mm/swap.c
@@ -847,7 +847,17 @@ void lru_add_drain_all(void)
 }
 #endif /* CONFIG_SMP */
 
-atomic_t lru_disable_count = ATOMIC_INIT(0);
+static atomic_t lru_disable_count = ATOMIC_INIT(0);
+
+bool lru_cache_disabled(void)
+{
+	return atomic_read(&lru_disable_count) != 0;
+}
+
+void lru_cache_enable(void)
+{
+	atomic_dec(&lru_disable_count);
+}
 
 /*
  * lru_cache_disable() needs to be called before we start compiling
@@ -859,7 +869,17 @@ atomic_t lru_disable_count = ATOMIC_INIT(0);
  */
 void lru_cache_disable(void)
 {
-	atomic_inc(&lru_disable_count);
+	static DEFINE_MUTEX(lock);
+
+	mutex_lock(&lock);
+	/*
+	 * If someone is already disabled lru_cache, just return with
+	 * increasing the lru_disable_count.
+	 */
+	if (atomic_inc_not_zero(&lru_disable_count)) {
+		mutex_unlock(&lock);
+		return;
+	}
 #ifdef CONFIG_SMP
 	/*
 	 * lru_add_drain_all in the force mode will schedule draining on
@@ -873,6 +893,8 @@ void lru_cache_disable(void)
 #else
 	lru_add_and_bh_lrus_drain();
 #endif
+	atomic_inc(&lru_disable_count);
+	mutex_unlock(&lock);
 }
 
 /**
-- 
2.34.1.400.ga245620fadb-goog