diff mbox series

[1/5] mm/vmscan: Throttle reclaim until some writeback completes if congested

Message ID 20210920085436.20939-2-mgorman@techsingularity.net (mailing list archive)
State New, archived
Headers show
Series Remove dependency on congestion_wait in mm/ | expand

Commit Message

Mel Gorman Sept. 20, 2021, 8:54 a.m. UTC
Page reclaim throttles on wait_iff_congested under the following conditions

o kswapd is encountering pages under writeback and marked for immediate
  reclaim implying that pages are cycling through the LRU faster than
  pages can be cleaned.

o Direct reclaim will stall if all dirty pages are backed by congested
  inodes.

wait_iff_congested is almost completely broken with few exceptions. This
patch adds a new node-based workqueue and tracks the number of throttled
tasks and pages written back since throttling started. If enough pages
belonging to the node are written back then the throttled tasks will wake
early. If not, the throttled tasks sleeps until the timeout expires.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
---
 include/linux/backing-dev.h      |  1 -
 include/linux/mmzone.h           |  9 +++++
 include/trace/events/vmscan.h    | 34 +++++++++++++++++++
 include/trace/events/writeback.h |  7 ----
 mm/backing-dev.c                 | 48 --------------------------
 mm/filemap.c                     |  1 +
 mm/internal.h                    |  9 +++++
 mm/page_alloc.c                  |  1 +
 mm/vmscan.c                      | 58 +++++++++++++++++++++++++++-----
 mm/vmstat.c                      |  1 +
 10 files changed, 105 insertions(+), 64 deletions(-)

Comments

NeilBrown Sept. 20, 2021, 11:19 p.m. UTC | #1
On Mon, 20 Sep 2021, Mel Gorman wrote:
>  
> +void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page);
> +static inline void acct_reclaim_writeback(struct page *page)
> +{
> +	pg_data_t *pgdat = page_pgdat(page);
> +
> +	if (atomic_read(&pgdat->nr_reclaim_throttled))
> +		__acct_reclaim_writeback(pgdat, page);

The first thing __acct_reclaim_writeback() does is repeat that
atomic_read().
Should we read it once and pass the value in to
__acct_reclaim_writeback(), or is that an unnecessary
micro-optimisation?


> +/*
> + * Account for pages written if tasks are throttled waiting on dirty
> + * pages to clean. If enough pages have been cleaned since throttling
> + * started then wakeup the throttled tasks.
> + */
> +void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page)
> +{
> +	unsigned long nr_written;
> +	int nr_throttled = atomic_read(&pgdat->nr_reclaim_throttled);
> +
> +	__inc_node_page_state(page, NR_THROTTLED_WRITTEN);
> +	nr_written = node_page_state(pgdat, NR_THROTTLED_WRITTEN) -
> +		READ_ONCE(pgdat->nr_reclaim_start);
> +
> +	if (nr_written > SWAP_CLUSTER_MAX * nr_throttled)
> +		wake_up_interruptible_all(&pgdat->reclaim_wait);

A simple wake_up() could be used here.  "interruptible" is only needed
if non-interruptible waiters should be left alone.  "_all" is only needed
if there are some exclusive waiters.  Neither of these apply, so I think
the simpler interface is best.


> +}
> +
>  /* possible outcome of pageout() */
>  typedef enum {
>  	/* failed to write page out, page is locked */
> @@ -1412,9 +1453,8 @@ static unsigned int shrink_page_list(struct list_head *page_list,
>  
>  		/*
>  		 * The number of dirty pages determines if a node is marked
> -		 * reclaim_congested which affects wait_iff_congested. kswapd
> -		 * will stall and start writing pages if the tail of the LRU
> -		 * is all dirty unqueued pages.
> +		 * reclaim_congested. kswapd will stall and start writing
> +		 * pages if the tail of the LRU is all dirty unqueued pages.
>  		 */
>  		page_check_dirty_writeback(page, &dirty, &writeback);
>  		if (dirty || writeback)
> @@ -3180,19 +3220,20 @@ static void shrink_node(pg_data_t *pgdat, struct scan_control *sc)
>  		 * If kswapd scans pages marked for immediate
>  		 * reclaim and under writeback (nr_immediate), it
>  		 * implies that pages are cycling through the LRU
> -		 * faster than they are written so also forcibly stall.
> +		 * faster than they are written so forcibly stall
> +		 * until some pages complete writeback.
>  		 */
>  		if (sc->nr.immediate)
> -			congestion_wait(BLK_RW_ASYNC, HZ/10);
> +			reclaim_throttle(pgdat, VMSCAN_THROTTLE_WRITEBACK, HZ/10);
>  	}
>  
>  	/*
>  	 * Tag a node/memcg as congested if all the dirty pages
>  	 * scanned were backed by a congested BDI and

"congested BDI" doesn't mean anything any more.  Is this a good time to
correct that comment.
This comment seems to refer to the test

      sc->nr.dirty && sc->nr.dirty == sc->nr.congested)

a few lines down.  But nr.congested is set from nr_congested which
counts when inode_write_congested() is true - almost never - and when 
"writeback and PageReclaim()".

Is that last test the sign that we are cycling through the LRU to fast?
So the comment could become:

   Tag a node/memcg as congested if all the dirty page were
   already marked for writeback and immediate reclaim (counted in
   nr.congested).

??

Patch seems to make sense to me, but I'm not expert in this area.

Thanks!
NeilBrown
NeilBrown Sept. 21, 2021, 12:13 a.m. UTC | #2
On Mon, 20 Sep 2021, Mel Gorman wrote:
> -long wait_iff_congested(int sync, long timeout)
> -{
> -	long ret;
> -	unsigned long start = jiffies;
> -	DEFINE_WAIT(wait);
> -	wait_queue_head_t *wqh = &congestion_wqh[sync];
> -
> -	/*
> -	 * If there is no congestion, yield if necessary instead
> -	 * of sleeping on the congestion queue
> -	 */
> -	if (atomic_read(&nr_wb_congested[sync]) == 0) {
> -		cond_resched();
> -
> -		/* In case we scheduled, work out time remaining */
> -		ret = timeout - (jiffies - start);
> -		if (ret < 0)
> -			ret = 0;
> -
> -		goto out;
> -	}
> -
> -	/* Sleep until uncongested or a write happens */
> -	prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);

Uninterruptible wait.

....
> +static void
> +reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason,
> +							long timeout)
> +{
> +	wait_queue_head_t *wqh = &pgdat->reclaim_wait;
> +	unsigned long start = jiffies;
> +	long ret;
> +	DEFINE_WAIT(wait);
> +
> +	atomic_inc(&pgdat->nr_reclaim_throttled);
> +	WRITE_ONCE(pgdat->nr_reclaim_start,
> +		 node_page_state(pgdat, NR_THROTTLED_WRITTEN));
> +
> +	prepare_to_wait(wqh, &wait, TASK_INTERRUPTIBLE);

Interruptible wait.

Why the change?  I think these waits really need to be TASK_UNINTERRUPTIBLE.

Thanks,
NeilBrown
Mel Gorman Sept. 21, 2021, 10:58 a.m. UTC | #3
On Tue, Sep 21, 2021 at 10:13:17AM +1000, NeilBrown wrote:
> On Mon, 20 Sep 2021, Mel Gorman wrote:
> > -long wait_iff_congested(int sync, long timeout)
> > -{
> > -	long ret;
> > -	unsigned long start = jiffies;
> > -	DEFINE_WAIT(wait);
> > -	wait_queue_head_t *wqh = &congestion_wqh[sync];
> > -
> > -	/*
> > -	 * If there is no congestion, yield if necessary instead
> > -	 * of sleeping on the congestion queue
> > -	 */
> > -	if (atomic_read(&nr_wb_congested[sync]) == 0) {
> > -		cond_resched();
> > -
> > -		/* In case we scheduled, work out time remaining */
> > -		ret = timeout - (jiffies - start);
> > -		if (ret < 0)
> > -			ret = 0;
> > -
> > -		goto out;
> > -	}
> > -
> > -	/* Sleep until uncongested or a write happens */
> > -	prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
> 
> Uninterruptible wait.
> 
> ....
> > +static void
> > +reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason,
> > +							long timeout)
> > +{
> > +	wait_queue_head_t *wqh = &pgdat->reclaim_wait;
> > +	unsigned long start = jiffies;
> > +	long ret;
> > +	DEFINE_WAIT(wait);
> > +
> > +	atomic_inc(&pgdat->nr_reclaim_throttled);
> > +	WRITE_ONCE(pgdat->nr_reclaim_start,
> > +		 node_page_state(pgdat, NR_THROTTLED_WRITTEN));
> > +
> > +	prepare_to_wait(wqh, &wait, TASK_INTERRUPTIBLE);
> 
> Interruptible wait.
> 
> Why the change?  I think these waits really need to be TASK_UNINTERRUPTIBLE.
> 

Because from mm/ context, I saw no reason why the task *should* be
uninterruptible. It's waiting on other tasks to complete IO and it is not
protecting device state, filesystem state or anything else. If it gets
a signal, it's safe to wake up, particularly if that signal is KILL and
the context is a direct reclaimer.

The original TASK_UNINTERRUPTIBLE is almost certainly a copy&paste from
congestion_wait which may be called because a filesystem operation must
complete before it can return to userspace so a signal waking it up is
pointless.
Mel Gorman Sept. 21, 2021, 11:12 a.m. UTC | #4
On Tue, Sep 21, 2021 at 09:19:07AM +1000, NeilBrown wrote:
> On Mon, 20 Sep 2021, Mel Gorman wrote:
> >  
> > +void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page);
> > +static inline void acct_reclaim_writeback(struct page *page)
> > +{
> > +	pg_data_t *pgdat = page_pgdat(page);
> > +
> > +	if (atomic_read(&pgdat->nr_reclaim_throttled))
> > +		__acct_reclaim_writeback(pgdat, page);
> 
> The first thing __acct_reclaim_writeback() does is repeat that
> atomic_read().
> Should we read it once and pass the value in to
> __acct_reclaim_writeback(), or is that an unnecessary
> micro-optimisation?
> 

I think it's a micro-optimisation but I can still do it.

> 
> > +/*
> > + * Account for pages written if tasks are throttled waiting on dirty
> > + * pages to clean. If enough pages have been cleaned since throttling
> > + * started then wakeup the throttled tasks.
> > + */
> > +void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page)
> > +{
> > +	unsigned long nr_written;
> > +	int nr_throttled = atomic_read(&pgdat->nr_reclaim_throttled);
> > +
> > +	__inc_node_page_state(page, NR_THROTTLED_WRITTEN);
> > +	nr_written = node_page_state(pgdat, NR_THROTTLED_WRITTEN) -
> > +		READ_ONCE(pgdat->nr_reclaim_start);
> > +
> > +	if (nr_written > SWAP_CLUSTER_MAX * nr_throttled)
> > +		wake_up_interruptible_all(&pgdat->reclaim_wait);
> 
> A simple wake_up() could be used here.  "interruptible" is only needed
> if non-interruptible waiters should be left alone.  "_all" is only needed
> if there are some exclusive waiters.  Neither of these apply, so I think
> the simpler interface is best.
> 

You're right.

> 
> > +}
> > +
> >  /* possible outcome of pageout() */
> >  typedef enum {
> >  	/* failed to write page out, page is locked */
> > @@ -1412,9 +1453,8 @@ static unsigned int shrink_page_list(struct list_head *page_list,
> >  
> >  		/*
> >  		 * The number of dirty pages determines if a node is marked
> > -		 * reclaim_congested which affects wait_iff_congested. kswapd
> > -		 * will stall and start writing pages if the tail of the LRU
> > -		 * is all dirty unqueued pages.
> > +		 * reclaim_congested. kswapd will stall and start writing
> > +		 * pages if the tail of the LRU is all dirty unqueued pages.
> >  		 */
> >  		page_check_dirty_writeback(page, &dirty, &writeback);
> >  		if (dirty || writeback)
> > @@ -3180,19 +3220,20 @@ static void shrink_node(pg_data_t *pgdat, struct scan_control *sc)
> >  		 * If kswapd scans pages marked for immediate
> >  		 * reclaim and under writeback (nr_immediate), it
> >  		 * implies that pages are cycling through the LRU
> > -		 * faster than they are written so also forcibly stall.
> > +		 * faster than they are written so forcibly stall
> > +		 * until some pages complete writeback.
> >  		 */
> >  		if (sc->nr.immediate)
> > -			congestion_wait(BLK_RW_ASYNC, HZ/10);
> > +			reclaim_throttle(pgdat, VMSCAN_THROTTLE_WRITEBACK, HZ/10);
> >  	}
> >  
> >  	/*
> >  	 * Tag a node/memcg as congested if all the dirty pages
> >  	 * scanned were backed by a congested BDI and
> 
> "congested BDI" doesn't mean anything any more.  Is this a good time to
> correct that comment.
> This comment seems to refer to the test
> 
>       sc->nr.dirty && sc->nr.dirty == sc->nr.congested)
> 
> a few lines down.  But nr.congested is set from nr_congested which
> counts when inode_write_congested() is true - almost never - and when 
> "writeback and PageReclaim()".
> 
> Is that last test the sign that we are cycling through the LRU to fast?
> So the comment could become:
> 
>    Tag a node/memcg as congested if all the dirty page were
>    already marked for writeback and immediate reclaim (counted in
>    nr.congested).
> 
> ??
> 
> Patch seems to make sense to me, but I'm not expert in this area.
> 

Comments updated.

Diff on top looks like

diff --git a/mm/internal.h b/mm/internal.h
index e25b3686bfab..90764d646e02 100644
--- a/mm/internal.h
+++ b/mm/internal.h
@@ -34,13 +34,15 @@
 
 void page_writeback_init(void);
 
-void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page);
+void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page,
+						int nr_throttled);
 static inline void acct_reclaim_writeback(struct page *page)
 {
 	pg_data_t *pgdat = page_pgdat(page);
+	int nr_throttled = atomic_read(&pgdat->nr_reclaim_throttled);
 
-	if (atomic_read(&pgdat->nr_reclaim_throttled))
-		__acct_reclaim_writeback(pgdat, page);
+	if (nr_throttled)
+		__acct_reclaim_writeback(pgdat, page, nr_throttled);
 }
 
 vm_fault_t do_swap_page(struct vm_fault *vmf);
diff --git a/mm/vmscan.c b/mm/vmscan.c
index b58ea0b13286..2dc17de91d32 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1034,10 +1034,10 @@ reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason,
  * pages to clean. If enough pages have been cleaned since throttling
  * started then wakeup the throttled tasks.
  */
-void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page)
+void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page,
+							int nr_throttled)
 {
 	unsigned long nr_written;
-	int nr_throttled = atomic_read(&pgdat->nr_reclaim_throttled);
 
 	__inc_node_page_state(page, NR_THROTTLED_WRITTEN);
 	nr_written = node_page_state(pgdat, NR_THROTTLED_WRITTEN) -
@@ -3228,9 +3228,8 @@ static void shrink_node(pg_data_t *pgdat, struct scan_control *sc)
 	}
 
 	/*
-	 * Tag a node/memcg as congested if all the dirty pages
-	 * scanned were backed by a congested BDI and
-	 * non-kswapd tasks will stall on reclaim_throttle.
+	 * Tag a node/memcg as congested if all the dirty pages were marked
+	 * for writeback and immediate reclaim (counted in nr.congested).
 	 *
 	 * Legacy memcg will stall in page writeback so avoid forcibly
 	 * stalling in reclaim_throttle().
@@ -3241,8 +3240,8 @@ static void shrink_node(pg_data_t *pgdat, struct scan_control *sc)
 		set_bit(LRUVEC_CONGESTED, &target_lruvec->flags);
 
 	/*
-	 * Stall direct reclaim for IO completions if underlying BDIs
-	 * and node is congested. Allow kswapd to continue until it
+	 * Stall direct reclaim for IO completions if the lruvec is
+	 * node is congested. Allow kswapd to continue until it
 	 * starts encountering unqueued dirty pages or cycling through
 	 * the LRU too quickly.
 	 */
@@ -4427,7 +4426,7 @@ void wakeup_kswapd(struct zone *zone, gfp_t gfp_flags, int order,
 
 	trace_mm_vmscan_wakeup_kswapd(pgdat->node_id, highest_zoneidx, order,
 				      gfp_flags);
-	wake_up_interruptible(&pgdat->kswapd_wait);
+	wake_up_all(&pgdat->kswapd_wait);
 }
 
 #ifdef CONFIG_HIBERNATION
NeilBrown Sept. 21, 2021, 9:27 p.m. UTC | #5
On Tue, 21 Sep 2021, Mel Gorman wrote:
> On Tue, Sep 21, 2021 at 09:19:07AM +1000, NeilBrown wrote:
> > On Mon, 20 Sep 2021, Mel Gorman wrote:
> > >  
> > > +void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page);
> > > +static inline void acct_reclaim_writeback(struct page *page)
> > > +{
> > > +	pg_data_t *pgdat = page_pgdat(page);
> > > +
> > > +	if (atomic_read(&pgdat->nr_reclaim_throttled))
> > > +		__acct_reclaim_writeback(pgdat, page);
> > 
> > The first thing __acct_reclaim_writeback() does is repeat that
> > atomic_read().
> > Should we read it once and pass the value in to
> > __acct_reclaim_writeback(), or is that an unnecessary
> > micro-optimisation?
> > 
> 
> I think it's a micro-optimisation but I can still do it.
> 
> > 
> > > +/*
> > > + * Account for pages written if tasks are throttled waiting on dirty
> > > + * pages to clean. If enough pages have been cleaned since throttling
> > > + * started then wakeup the throttled tasks.
> > > + */
> > > +void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page)
> > > +{
> > > +	unsigned long nr_written;
> > > +	int nr_throttled = atomic_read(&pgdat->nr_reclaim_throttled);
> > > +
> > > +	__inc_node_page_state(page, NR_THROTTLED_WRITTEN);
> > > +	nr_written = node_page_state(pgdat, NR_THROTTLED_WRITTEN) -
> > > +		READ_ONCE(pgdat->nr_reclaim_start);
> > > +
> > > +	if (nr_written > SWAP_CLUSTER_MAX * nr_throttled)
> > > +		wake_up_interruptible_all(&pgdat->reclaim_wait);
> > 
> > A simple wake_up() could be used here.  "interruptible" is only needed
> > if non-interruptible waiters should be left alone.  "_all" is only needed
> > if there are some exclusive waiters.  Neither of these apply, so I think
> > the simpler interface is best.
> > 
> 
> You're right.
> 
> > 
> > > +}
> > > +
> > >  /* possible outcome of pageout() */
> > >  typedef enum {
> > >  	/* failed to write page out, page is locked */
> > > @@ -1412,9 +1453,8 @@ static unsigned int shrink_page_list(struct list_head *page_list,
> > >  
> > >  		/*
> > >  		 * The number of dirty pages determines if a node is marked
> > > -		 * reclaim_congested which affects wait_iff_congested. kswapd
> > > -		 * will stall and start writing pages if the tail of the LRU
> > > -		 * is all dirty unqueued pages.
> > > +		 * reclaim_congested. kswapd will stall and start writing
> > > +		 * pages if the tail of the LRU is all dirty unqueued pages.
> > >  		 */
> > >  		page_check_dirty_writeback(page, &dirty, &writeback);
> > >  		if (dirty || writeback)
> > > @@ -3180,19 +3220,20 @@ static void shrink_node(pg_data_t *pgdat, struct scan_control *sc)
> > >  		 * If kswapd scans pages marked for immediate
> > >  		 * reclaim and under writeback (nr_immediate), it
> > >  		 * implies that pages are cycling through the LRU
> > > -		 * faster than they are written so also forcibly stall.
> > > +		 * faster than they are written so forcibly stall
> > > +		 * until some pages complete writeback.
> > >  		 */
> > >  		if (sc->nr.immediate)
> > > -			congestion_wait(BLK_RW_ASYNC, HZ/10);
> > > +			reclaim_throttle(pgdat, VMSCAN_THROTTLE_WRITEBACK, HZ/10);
> > >  	}
> > >  
> > >  	/*
> > >  	 * Tag a node/memcg as congested if all the dirty pages
> > >  	 * scanned were backed by a congested BDI and
> > 
> > "congested BDI" doesn't mean anything any more.  Is this a good time to
> > correct that comment.
> > This comment seems to refer to the test
> > 
> >       sc->nr.dirty && sc->nr.dirty == sc->nr.congested)
> > 
> > a few lines down.  But nr.congested is set from nr_congested which
> > counts when inode_write_congested() is true - almost never - and when 
> > "writeback and PageReclaim()".
> > 
> > Is that last test the sign that we are cycling through the LRU to fast?
> > So the comment could become:
> > 
> >    Tag a node/memcg as congested if all the dirty page were
> >    already marked for writeback and immediate reclaim (counted in
> >    nr.congested).
> > 
> > ??
> > 
> > Patch seems to make sense to me, but I'm not expert in this area.
> > 
> 
> Comments updated.
> 
> Diff on top looks like
> 
> diff --git a/mm/internal.h b/mm/internal.h
> index e25b3686bfab..90764d646e02 100644
> --- a/mm/internal.h
> +++ b/mm/internal.h
> @@ -34,13 +34,15 @@
>  
>  void page_writeback_init(void);
>  
> -void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page);
> +void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page,
> +						int nr_throttled);
>  static inline void acct_reclaim_writeback(struct page *page)
>  {
>  	pg_data_t *pgdat = page_pgdat(page);
> +	int nr_throttled = atomic_read(&pgdat->nr_reclaim_throttled);
>  
> -	if (atomic_read(&pgdat->nr_reclaim_throttled))
> -		__acct_reclaim_writeback(pgdat, page);
> +	if (nr_throttled)
> +		__acct_reclaim_writeback(pgdat, page, nr_throttled);
>  }
>  
>  vm_fault_t do_swap_page(struct vm_fault *vmf);
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index b58ea0b13286..2dc17de91d32 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1034,10 +1034,10 @@ reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason,
>   * pages to clean. If enough pages have been cleaned since throttling
>   * started then wakeup the throttled tasks.
>   */
> -void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page)
> +void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page,
> +							int nr_throttled)
>  {
>  	unsigned long nr_written;
> -	int nr_throttled = atomic_read(&pgdat->nr_reclaim_throttled);
>  
>  	__inc_node_page_state(page, NR_THROTTLED_WRITTEN);
>  	nr_written = node_page_state(pgdat, NR_THROTTLED_WRITTEN) -
> @@ -3228,9 +3228,8 @@ static void shrink_node(pg_data_t *pgdat, struct scan_control *sc)
>  	}
>  
>  	/*
> -	 * Tag a node/memcg as congested if all the dirty pages
> -	 * scanned were backed by a congested BDI and
> -	 * non-kswapd tasks will stall on reclaim_throttle.
> +	 * Tag a node/memcg as congested if all the dirty pages were marked
> +	 * for writeback and immediate reclaim (counted in nr.congested).
>  	 *
>  	 * Legacy memcg will stall in page writeback so avoid forcibly
>  	 * stalling in reclaim_throttle().
> @@ -3241,8 +3240,8 @@ static void shrink_node(pg_data_t *pgdat, struct scan_control *sc)
>  		set_bit(LRUVEC_CONGESTED, &target_lruvec->flags);
>  
>  	/*
> -	 * Stall direct reclaim for IO completions if underlying BDIs
> -	 * and node is congested. Allow kswapd to continue until it
> +	 * Stall direct reclaim for IO completions if the lruvec is
> +	 * node is congested. Allow kswapd to continue until it
>  	 * starts encountering unqueued dirty pages or cycling through
>  	 * the LRU too quickly.
>  	 */
> @@ -4427,7 +4426,7 @@ void wakeup_kswapd(struct zone *zone, gfp_t gfp_flags, int order,
>  
>  	trace_mm_vmscan_wakeup_kswapd(pgdat->node_id, highest_zoneidx, order,
>  				      gfp_flags);
> -	wake_up_interruptible(&pgdat->kswapd_wait);
> +	wake_up_all(&pgdat->kswapd_wait);

???

That isn't the wake_up that I pointed too.

Other changes look good - thanks.

NeilBrown


>  }
>  
>  #ifdef CONFIG_HIBERNATION
> 
>
NeilBrown Sept. 21, 2021, 9:40 p.m. UTC | #6
On Tue, 21 Sep 2021, Mel Gorman wrote:
> On Tue, Sep 21, 2021 at 10:13:17AM +1000, NeilBrown wrote:
> > On Mon, 20 Sep 2021, Mel Gorman wrote:
> > > -long wait_iff_congested(int sync, long timeout)
> > > -{
> > > -	long ret;
> > > -	unsigned long start = jiffies;
> > > -	DEFINE_WAIT(wait);
> > > -	wait_queue_head_t *wqh = &congestion_wqh[sync];
> > > -
> > > -	/*
> > > -	 * If there is no congestion, yield if necessary instead
> > > -	 * of sleeping on the congestion queue
> > > -	 */
> > > -	if (atomic_read(&nr_wb_congested[sync]) == 0) {
> > > -		cond_resched();
> > > -
> > > -		/* In case we scheduled, work out time remaining */
> > > -		ret = timeout - (jiffies - start);
> > > -		if (ret < 0)
> > > -			ret = 0;
> > > -
> > > -		goto out;
> > > -	}
> > > -
> > > -	/* Sleep until uncongested or a write happens */
> > > -	prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
> > 
> > Uninterruptible wait.
> > 
> > ....
> > > +static void
> > > +reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason,
> > > +							long timeout)
> > > +{
> > > +	wait_queue_head_t *wqh = &pgdat->reclaim_wait;
> > > +	unsigned long start = jiffies;
> > > +	long ret;
> > > +	DEFINE_WAIT(wait);
> > > +
> > > +	atomic_inc(&pgdat->nr_reclaim_throttled);
> > > +	WRITE_ONCE(pgdat->nr_reclaim_start,
> > > +		 node_page_state(pgdat, NR_THROTTLED_WRITTEN));
> > > +
> > > +	prepare_to_wait(wqh, &wait, TASK_INTERRUPTIBLE);
> > 
> > Interruptible wait.
> > 
> > Why the change?  I think these waits really need to be TASK_UNINTERRUPTIBLE.
> > 
> 
> Because from mm/ context, I saw no reason why the task *should* be
> uninterruptible. It's waiting on other tasks to complete IO and it is not
> protecting device state, filesystem state or anything else. If it gets
> a signal, it's safe to wake up, particularly if that signal is KILL and
> the context is a direct reclaimer.

I disagree.  An Interruptible sleep only makes sense if the "was
interrupted" status can propagate up to user-space (or to some in-kernel
handler that will clear the signal).
In particular, if reclaim_throttle() is called in a loop (which it is),
and if that loop doesn't check for signal_pending (which it doesn't),
then the next time around the loop after receiving a signal, it won't
sleep at all.  That would be bad.

In general, if you don't return an error, then you probably shouldn't
sleep Interruptible.

I notice that tasks sleep on kswapd_wait as TASK_INTERRUPTIBLE, but they
don't have any signal handling.  I suspect this isn't actually a defect
because I suspect that is it not even possible to SIGKILL kswapd.  But
the code seems misleading.  I guess I should write a patch.

Unless reclaim knows to abort completely on a signal (__GFP_KILLABLE
???) this must be an UNINTERRUPTIBLE wait.

Thanks,
NeilBrown

> 
> The original TASK_UNINTERRUPTIBLE is almost certainly a copy&paste from
> congestion_wait which may be called because a filesystem operation must
> complete before it can return to userspace so a signal waking it up is
> pointless.
> 
> -- 
> Mel Gorman
> SUSE Labs
> 
>
Dave Chinner Sept. 22, 2021, 6:04 a.m. UTC | #7
On Tue, Sep 21, 2021 at 11:58:31AM +0100, Mel Gorman wrote:
> On Tue, Sep 21, 2021 at 10:13:17AM +1000, NeilBrown wrote:
> > On Mon, 20 Sep 2021, Mel Gorman wrote:
> > > -long wait_iff_congested(int sync, long timeout)
> > > -{
> > > -	long ret;
> > > -	unsigned long start = jiffies;
> > > -	DEFINE_WAIT(wait);
> > > -	wait_queue_head_t *wqh = &congestion_wqh[sync];
> > > -
> > > -	/*
> > > -	 * If there is no congestion, yield if necessary instead
> > > -	 * of sleeping on the congestion queue
> > > -	 */
> > > -	if (atomic_read(&nr_wb_congested[sync]) == 0) {
> > > -		cond_resched();
> > > -
> > > -		/* In case we scheduled, work out time remaining */
> > > -		ret = timeout - (jiffies - start);
> > > -		if (ret < 0)
> > > -			ret = 0;
> > > -
> > > -		goto out;
> > > -	}
> > > -
> > > -	/* Sleep until uncongested or a write happens */
> > > -	prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
> > 
> > Uninterruptible wait.
> > 
> > ....
> > > +static void
> > > +reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason,
> > > +							long timeout)
> > > +{
> > > +	wait_queue_head_t *wqh = &pgdat->reclaim_wait;
> > > +	unsigned long start = jiffies;
> > > +	long ret;
> > > +	DEFINE_WAIT(wait);
> > > +
> > > +	atomic_inc(&pgdat->nr_reclaim_throttled);
> > > +	WRITE_ONCE(pgdat->nr_reclaim_start,
> > > +		 node_page_state(pgdat, NR_THROTTLED_WRITTEN));
> > > +
> > > +	prepare_to_wait(wqh, &wait, TASK_INTERRUPTIBLE);
> > 
> > Interruptible wait.
> > 
> > Why the change?  I think these waits really need to be TASK_UNINTERRUPTIBLE.
> > 
> 
> Because from mm/ context, I saw no reason why the task *should* be
> uninterruptible. It's waiting on other tasks to complete IO and it is not
> protecting device state, filesystem state or anything else. If it gets
> a signal, it's safe to wake up, particularly if that signal is KILL and
> the context is a direct reclaimer.

I disagree. whether the sleep should be interruptable or
not is entirely dependent on whether the caller can handle failure
or not. If this is GFP_NOFAIL, allocation must not fail no matter
what the context is, so signals and the like are irrelevant.

For a context that can handle allocation failure, then it makes
sense to wake on events that will result in the allocation failing
immediately. But if all this does is make the allocation code go
around another retry loop sooner, then an interruptible sleep still
doesn't make any sense at all here...

> The original TASK_UNINTERRUPTIBLE is almost certainly a copy&paste from
> congestion_wait which may be called because a filesystem operation must
> complete before it can return to userspace so a signal waking it up is
> pointless.

Yup, but that AFAICT that same logic still applies. Only now it's
the allocation context that determines whether signal waking is
pointless or not...

Cheer,

Dave.
Mel Gorman Sept. 22, 2021, 8:03 a.m. UTC | #8
On Wed, Sep 22, 2021 at 04:04:47PM +1000, Dave Chinner wrote:
> On Tue, Sep 21, 2021 at 11:58:31AM +0100, Mel Gorman wrote:
> > On Tue, Sep 21, 2021 at 10:13:17AM +1000, NeilBrown wrote:
> > > On Mon, 20 Sep 2021, Mel Gorman wrote:
> > > > -long wait_iff_congested(int sync, long timeout)
> > > > -{
> > > > -	long ret;
> > > > -	unsigned long start = jiffies;
> > > > -	DEFINE_WAIT(wait);
> > > > -	wait_queue_head_t *wqh = &congestion_wqh[sync];
> > > > -
> > > > -	/*
> > > > -	 * If there is no congestion, yield if necessary instead
> > > > -	 * of sleeping on the congestion queue
> > > > -	 */
> > > > -	if (atomic_read(&nr_wb_congested[sync]) == 0) {
> > > > -		cond_resched();
> > > > -
> > > > -		/* In case we scheduled, work out time remaining */
> > > > -		ret = timeout - (jiffies - start);
> > > > -		if (ret < 0)
> > > > -			ret = 0;
> > > > -
> > > > -		goto out;
> > > > -	}
> > > > -
> > > > -	/* Sleep until uncongested or a write happens */
> > > > -	prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
> > > 
> > > Uninterruptible wait.
> > > 
> > > ....
> > > > +static void
> > > > +reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason,
> > > > +							long timeout)
> > > > +{
> > > > +	wait_queue_head_t *wqh = &pgdat->reclaim_wait;
> > > > +	unsigned long start = jiffies;
> > > > +	long ret;
> > > > +	DEFINE_WAIT(wait);
> > > > +
> > > > +	atomic_inc(&pgdat->nr_reclaim_throttled);
> > > > +	WRITE_ONCE(pgdat->nr_reclaim_start,
> > > > +		 node_page_state(pgdat, NR_THROTTLED_WRITTEN));
> > > > +
> > > > +	prepare_to_wait(wqh, &wait, TASK_INTERRUPTIBLE);
> > > 
> > > Interruptible wait.
> > > 
> > > Why the change?  I think these waits really need to be TASK_UNINTERRUPTIBLE.
> > > 
> > 
> > Because from mm/ context, I saw no reason why the task *should* be
> > uninterruptible. It's waiting on other tasks to complete IO and it is not
> > protecting device state, filesystem state or anything else. If it gets
> > a signal, it's safe to wake up, particularly if that signal is KILL and
> > the context is a direct reclaimer.
> 
> I disagree. whether the sleep should be interruptable or
> not is entirely dependent on whether the caller can handle failure
> or not. If this is GFP_NOFAIL, allocation must not fail no matter
> what the context is, so signals and the like are irrelevant.
> 
> For a context that can handle allocation failure, then it makes
> sense to wake on events that will result in the allocation failing
> immediately. But if all this does is make the allocation code go
> around another retry loop sooner, then an interruptible sleep still
> doesn't make any sense at all here...
> 

Ok, between this and Neil's mail on the same topic, I'm convinced.
diff mbox series

Patch

diff --git a/include/linux/backing-dev.h b/include/linux/backing-dev.h
index ac7f231b8825..9fb1f0ae273c 100644
--- a/include/linux/backing-dev.h
+++ b/include/linux/backing-dev.h
@@ -154,7 +154,6 @@  static inline int wb_congested(struct bdi_writeback *wb, int cong_bits)
 }
 
 long congestion_wait(int sync, long timeout);
-long wait_iff_congested(int sync, long timeout);
 
 static inline bool mapping_can_writeback(struct address_space *mapping)
 {
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 6a1d79d84675..ef0a63ebd21d 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -199,6 +199,7 @@  enum node_stat_item {
 	NR_VMSCAN_IMMEDIATE,	/* Prioritise for reclaim when writeback ends */
 	NR_DIRTIED,		/* page dirtyings since bootup */
 	NR_WRITTEN,		/* page writings since bootup */
+	NR_THROTTLED_WRITTEN,	/* NR_WRITTEN while reclaim throttled */
 	NR_KERNEL_MISC_RECLAIMABLE,	/* reclaimable non-slab kernel pages */
 	NR_FOLL_PIN_ACQUIRED,	/* via: pin_user_page(), gup flag: FOLL_PIN */
 	NR_FOLL_PIN_RELEASED,	/* pages returned via unpin_user_page() */
@@ -272,6 +273,10 @@  enum lru_list {
 	NR_LRU_LISTS
 };
 
+enum vmscan_throttle_state {
+	VMSCAN_THROTTLE_WRITEBACK,
+};
+
 #define for_each_lru(lru) for (lru = 0; lru < NR_LRU_LISTS; lru++)
 
 #define for_each_evictable_lru(lru) for (lru = 0; lru <= LRU_ACTIVE_FILE; lru++)
@@ -841,6 +846,10 @@  typedef struct pglist_data {
 	int node_id;
 	wait_queue_head_t kswapd_wait;
 	wait_queue_head_t pfmemalloc_wait;
+	wait_queue_head_t reclaim_wait;	/* wq for throttling reclaim */
+	atomic_t nr_reclaim_throttled;	/* nr of throtted tasks */
+	unsigned long nr_reclaim_start;	/* nr pages written while throttled
+					 * when throttling started. */
 	struct task_struct *kswapd;	/* Protected by
 					   mem_hotplug_begin/end() */
 	int kswapd_order;
diff --git a/include/trace/events/vmscan.h b/include/trace/events/vmscan.h
index 88faf2400ec2..c317f9fe0d17 100644
--- a/include/trace/events/vmscan.h
+++ b/include/trace/events/vmscan.h
@@ -27,6 +27,14 @@ 
 		{RECLAIM_WB_ASYNC,	"RECLAIM_WB_ASYNC"}	\
 		) : "RECLAIM_WB_NONE"
 
+#define _VMSCAN_THROTTLE_WRITEBACK	(1 << VMSCAN_THROTTLE_WRITEBACK)
+
+#define show_throttle_flags(flags)						\
+	(flags) ? __print_flags(flags, "|",					\
+		{_VMSCAN_THROTTLE_WRITEBACK,	"VMSCAN_THROTTLE_WRITEBACK"}	\
+		) : "VMSCAN_THROTTLE_NONE"
+
+
 #define trace_reclaim_flags(file) ( \
 	(file ? RECLAIM_WB_FILE : RECLAIM_WB_ANON) | \
 	(RECLAIM_WB_ASYNC) \
@@ -454,6 +462,32 @@  DEFINE_EVENT(mm_vmscan_direct_reclaim_end_template, mm_vmscan_node_reclaim_end,
 	TP_ARGS(nr_reclaimed)
 );
 
+TRACE_EVENT(mm_vmscan_throttled,
+
+	TP_PROTO(int nid, int usec_timeout, int usec_delayed, int reason),
+
+	TP_ARGS(nid, usec_timeout, usec_delayed, reason),
+
+	TP_STRUCT__entry(
+		__field(int, nid)
+		__field(int, usec_timeout)
+		__field(int, usec_delayed)
+		__field(int, reason)
+	),
+
+	TP_fast_assign(
+		__entry->nid = nid;
+		__entry->usec_timeout = usec_timeout;
+		__entry->usec_delayed = usec_delayed;
+		__entry->reason = 1U << reason;
+	),
+
+	TP_printk("nid=%d usec_timeout=%d usect_delayed=%d reason=%s",
+		__entry->nid,
+		__entry->usec_timeout,
+		__entry->usec_delayed,
+		show_throttle_flags(__entry->reason))
+);
 #endif /* _TRACE_VMSCAN_H */
 
 /* This part must be outside protection */
diff --git a/include/trace/events/writeback.h b/include/trace/events/writeback.h
index 840d1ba84cf5..3bc759b81897 100644
--- a/include/trace/events/writeback.h
+++ b/include/trace/events/writeback.h
@@ -763,13 +763,6 @@  DEFINE_EVENT(writeback_congest_waited_template, writeback_congestion_wait,
 	TP_ARGS(usec_timeout, usec_delayed)
 );
 
-DEFINE_EVENT(writeback_congest_waited_template, writeback_wait_iff_congested,
-
-	TP_PROTO(unsigned int usec_timeout, unsigned int usec_delayed),
-
-	TP_ARGS(usec_timeout, usec_delayed)
-);
-
 DECLARE_EVENT_CLASS(writeback_single_inode_template,
 
 	TP_PROTO(struct inode *inode,
diff --git a/mm/backing-dev.c b/mm/backing-dev.c
index 4a9d4e27d0d9..0ea1a105eae5 100644
--- a/mm/backing-dev.c
+++ b/mm/backing-dev.c
@@ -1041,51 +1041,3 @@  long congestion_wait(int sync, long timeout)
 	return ret;
 }
 EXPORT_SYMBOL(congestion_wait);
-
-/**
- * wait_iff_congested - Conditionally wait for a backing_dev to become uncongested or a pgdat to complete writes
- * @sync: SYNC or ASYNC IO
- * @timeout: timeout in jiffies
- *
- * In the event of a congested backing_dev (any backing_dev) this waits
- * for up to @timeout jiffies for either a BDI to exit congestion of the
- * given @sync queue or a write to complete.
- *
- * The return value is 0 if the sleep is for the full timeout. Otherwise,
- * it is the number of jiffies that were still remaining when the function
- * returned. return_value == timeout implies the function did not sleep.
- */
-long wait_iff_congested(int sync, long timeout)
-{
-	long ret;
-	unsigned long start = jiffies;
-	DEFINE_WAIT(wait);
-	wait_queue_head_t *wqh = &congestion_wqh[sync];
-
-	/*
-	 * If there is no congestion, yield if necessary instead
-	 * of sleeping on the congestion queue
-	 */
-	if (atomic_read(&nr_wb_congested[sync]) == 0) {
-		cond_resched();
-
-		/* In case we scheduled, work out time remaining */
-		ret = timeout - (jiffies - start);
-		if (ret < 0)
-			ret = 0;
-
-		goto out;
-	}
-
-	/* Sleep until uncongested or a write happens */
-	prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
-	ret = io_schedule_timeout(timeout);
-	finish_wait(wqh, &wait);
-
-out:
-	trace_writeback_wait_iff_congested(jiffies_to_usecs(timeout),
-					jiffies_to_usecs(jiffies - start));
-
-	return ret;
-}
-EXPORT_SYMBOL(wait_iff_congested);
diff --git a/mm/filemap.c b/mm/filemap.c
index dae481293b5d..59187787fbfc 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -1605,6 +1605,7 @@  void end_page_writeback(struct page *page)
 
 	smp_mb__after_atomic();
 	wake_up_page(page, PG_writeback);
+	acct_reclaim_writeback(page);
 	put_page(page);
 }
 EXPORT_SYMBOL(end_page_writeback);
diff --git a/mm/internal.h b/mm/internal.h
index cf3cb933eba3..e25b3686bfab 100644
--- a/mm/internal.h
+++ b/mm/internal.h
@@ -34,6 +34,15 @@ 
 
 void page_writeback_init(void);
 
+void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page);
+static inline void acct_reclaim_writeback(struct page *page)
+{
+	pg_data_t *pgdat = page_pgdat(page);
+
+	if (atomic_read(&pgdat->nr_reclaim_throttled))
+		__acct_reclaim_writeback(pgdat, page);
+}
+
 vm_fault_t do_swap_page(struct vm_fault *vmf);
 
 void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index b37435c274cf..d849ddfc1e51 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -7396,6 +7396,7 @@  static void __meminit pgdat_init_internals(struct pglist_data *pgdat)
 
 	init_waitqueue_head(&pgdat->kswapd_wait);
 	init_waitqueue_head(&pgdat->pfmemalloc_wait);
+	init_waitqueue_head(&pgdat->reclaim_wait);
 
 	pgdat_page_ext_init(pgdat);
 	lruvec_init(&pgdat->__lruvec);
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 74296c2d1fed..b58ea0b13286 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1006,6 +1006,47 @@  static void handle_write_error(struct address_space *mapping,
 	unlock_page(page);
 }
 
+static void
+reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason,
+							long timeout)
+{
+	wait_queue_head_t *wqh = &pgdat->reclaim_wait;
+	unsigned long start = jiffies;
+	long ret;
+	DEFINE_WAIT(wait);
+
+	atomic_inc(&pgdat->nr_reclaim_throttled);
+	WRITE_ONCE(pgdat->nr_reclaim_start,
+		 node_page_state(pgdat, NR_THROTTLED_WRITTEN));
+
+	prepare_to_wait(wqh, &wait, TASK_INTERRUPTIBLE);
+	ret = schedule_timeout(timeout);
+	finish_wait(wqh, &wait);
+	atomic_dec(&pgdat->nr_reclaim_throttled);
+
+	trace_mm_vmscan_throttled(pgdat->node_id, jiffies_to_usecs(timeout),
+				jiffies_to_usecs(jiffies - start),
+				reason);
+}
+
+/*
+ * Account for pages written if tasks are throttled waiting on dirty
+ * pages to clean. If enough pages have been cleaned since throttling
+ * started then wakeup the throttled tasks.
+ */
+void __acct_reclaim_writeback(pg_data_t *pgdat, struct page *page)
+{
+	unsigned long nr_written;
+	int nr_throttled = atomic_read(&pgdat->nr_reclaim_throttled);
+
+	__inc_node_page_state(page, NR_THROTTLED_WRITTEN);
+	nr_written = node_page_state(pgdat, NR_THROTTLED_WRITTEN) -
+		READ_ONCE(pgdat->nr_reclaim_start);
+
+	if (nr_written > SWAP_CLUSTER_MAX * nr_throttled)
+		wake_up_interruptible_all(&pgdat->reclaim_wait);
+}
+
 /* possible outcome of pageout() */
 typedef enum {
 	/* failed to write page out, page is locked */
@@ -1412,9 +1453,8 @@  static unsigned int shrink_page_list(struct list_head *page_list,
 
 		/*
 		 * The number of dirty pages determines if a node is marked
-		 * reclaim_congested which affects wait_iff_congested. kswapd
-		 * will stall and start writing pages if the tail of the LRU
-		 * is all dirty unqueued pages.
+		 * reclaim_congested. kswapd will stall and start writing
+		 * pages if the tail of the LRU is all dirty unqueued pages.
 		 */
 		page_check_dirty_writeback(page, &dirty, &writeback);
 		if (dirty || writeback)
@@ -3180,19 +3220,20 @@  static void shrink_node(pg_data_t *pgdat, struct scan_control *sc)
 		 * If kswapd scans pages marked for immediate
 		 * reclaim and under writeback (nr_immediate), it
 		 * implies that pages are cycling through the LRU
-		 * faster than they are written so also forcibly stall.
+		 * faster than they are written so forcibly stall
+		 * until some pages complete writeback.
 		 */
 		if (sc->nr.immediate)
-			congestion_wait(BLK_RW_ASYNC, HZ/10);
+			reclaim_throttle(pgdat, VMSCAN_THROTTLE_WRITEBACK, HZ/10);
 	}
 
 	/*
 	 * Tag a node/memcg as congested if all the dirty pages
 	 * scanned were backed by a congested BDI and
-	 * wait_iff_congested will stall.
+	 * non-kswapd tasks will stall on reclaim_throttle.
 	 *
 	 * Legacy memcg will stall in page writeback so avoid forcibly
-	 * stalling in wait_iff_congested().
+	 * stalling in reclaim_throttle().
 	 */
 	if ((current_is_kswapd() ||
 	     (cgroup_reclaim(sc) && writeback_throttling_sane(sc))) &&
@@ -3208,7 +3249,7 @@  static void shrink_node(pg_data_t *pgdat, struct scan_control *sc)
 	if (!current_is_kswapd() && current_may_throttle() &&
 	    !sc->hibernation_mode &&
 	    test_bit(LRUVEC_CONGESTED, &target_lruvec->flags))
-		wait_iff_congested(BLK_RW_ASYNC, HZ/10);
+		reclaim_throttle(pgdat, VMSCAN_THROTTLE_WRITEBACK, HZ/10);
 
 	if (should_continue_reclaim(pgdat, sc->nr_reclaimed - nr_reclaimed,
 				    sc))
@@ -4286,6 +4327,7 @@  static int kswapd(void *p)
 
 	WRITE_ONCE(pgdat->kswapd_order, 0);
 	WRITE_ONCE(pgdat->kswapd_highest_zoneidx, MAX_NR_ZONES);
+	atomic_set(&pgdat->nr_reclaim_throttled, 0);
 	for ( ; ; ) {
 		bool ret;
 
diff --git a/mm/vmstat.c b/mm/vmstat.c
index 8ce2620344b2..9b2bc9d61d4b 100644
--- a/mm/vmstat.c
+++ b/mm/vmstat.c
@@ -1225,6 +1225,7 @@  const char * const vmstat_text[] = {
 	"nr_vmscan_immediate_reclaim",
 	"nr_dirtied",
 	"nr_written",
+	"nr_throttled_written",
 	"nr_kernel_misc_reclaimable",
 	"nr_foll_pin_acquired",
 	"nr_foll_pin_released",