Skip to contentSkip to conversation

vmscan: don't use return value trick when oom_killer_disabled

PATCH In review KOSAKI Motohiro 2 participants 9 messages Sep 1, 2010 – Sep 2, 2010

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 vmscan: don't use return value trick when oom_killer_disabled

M. Vefa Bicakci reported 2.6.35 kernel hang up when hibernation on his 32bit 3GB mem machine. (https://bugzilla.kernel.org/show_bug.cgi?id=16771)
Also he was bisected first bad commit is below

commit bb21c7ce18eff8e6e7877ca1d06c6db719376e3c
Author: KOSAKI Motohiro <[email protected]>
Date: Fri Jun 4 14:15:05 2010 -0700

     vmscan: fix do_try_to_free_pages() return value when priority==0 reclaim failure

At first impression, this seemed very strange because the above commit only chenged function return value and hibernate_preallocate_memory() ignore return value of shrink_all_memory(). But it's related.

Now, page allocation from hibernation code may enter infinite loop if the system has highmem.

The reasons are two. 1) hibernate_preallocate_memory() call alloc_pages() wrong order 2) vmscan don't care enough OOM case when oom_killer_disabled.

This patch only fix (2). Why is oom_killer_disabled so special? because when hibernation case, zone->all_unreclaimable never be turned on. hibernation freeze all tasks at first, then kswapd can't works in this case, and zone->all_unreclaimable is only turned from kswapd.

Cc Johannes Weiner <[email protected]>
Cc Rik van Riel <[email protected]>
Cc "Rafael J. Wysocki" <[email protected]>
Cc M. Vefa Bicakci <[email protected]>
---
 mm/vmscan.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
mm/vmscan.c +2 -1
index c391c32..1919d8a 100644
@@ -40,6 +40,7 @@
 #include <linux/memcontrol.h>
 #include <linux/delayacct.h>
 #include <linux/sysctl.h>
+#include <linux/oom.h>
 
 #include <asm/tlbflush.h>
 #include <asm/div64.h>
@@ -1931,7 +1932,7 @@ out:
 		return sc->nr_reclaimed;
 
 	/* top priority shrink_zones still had more to do? don't OOM, then */
-	if (scanning_global_lru(sc) && !all_unreclaimable)
+	if (scanning_global_lru(sc) && !all_unreclaimable && !oom_killer_disabled)
 		return 1;
 
 	return 0;
Signature · 9 lines
1.6.5.2



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#1 lore raw

Hi KOSAKI,

Quoted 17 lines · KOSAKI Motohiro

M. Vefa Bicakci reported 2.6.35 kernel hang up when hibernation on his 32bit 3GB mem machine. (https://bugzilla.kernel.org/show_bug.cgi?id=16771)
Also he was bisected first bad commit is below

 commit bb21c7ce18eff8e6e7877ca1d06c6db719376e3c
 Author: KOSAKI Motohiro <[email protected]>
 Date:   Fri Jun 4 14:15:05 2010 -0700

    vmscan: fix do_try_to_free_pages() return value when priority==0 reclaim failure

At first impression, this seemed very strange because the above commit only chenged function return value and hibernate_preallocate_memory() ignore return value of shrink_all_memory(). But it's related.

Now, page allocation from hibernation code may enter infinite loop if the system has highmem.

Quoted 8 lines

The reasons are two. 1) hibernate_preallocate_memory() call alloc_pages() wrong order 2) vmscan don't care enough OOM case when oom_killer_disabled.

This patch only fix (2). Why is oom_killer_disabled so special? because when hibernation case, zone->all_unreclaimable never be turned on. hibernation freeze all tasks at first, then kswapd can't works in this case, and zone->all_unreclaimable is only turned from kswapd.

Nice catch!!
There is some comment below.

Quoted 33 lines
Cc Johannes Weiner <[email protected]>
Cc Rik van Riel <[email protected]>
Cc "Rafael J. Wysocki" <[email protected]>
Cc M. Vefa Bicakci <[email protected]>

---
 mm/vmscan.c |    3 ++-

 1 files changed, 2 insertions(+), 1 deletions(-)
mm/vmscan.c +2 -2
index c391c32..1919d8a 100644
@@ -40,6 +40,7 @@
  #include <linux/memcontrol.h>
  #include <linux/delayacct.h>
  #include <linux/sysctl.h>
+#include <linux/oom.h>
 
  #include <asm/tlbflush.h>
  #include <asm/div64.h>
@@ -1931,7 +1932,7 @@ out:
                return sc->nr_reclaimed;
 
        /* top priority shrink_zones still had more to do? don't OOM, then */
-       if (scanning_global_lru(sc) && !all_unreclaimable)
+       if (scanning_global_lru(sc) && !all_unreclaimable && !oom_killer_disabled)
                return 1;
 
        return 0;
--
Quoted 2 lines

1.6.5.2

I don't like use oom_killer_disabled directly. That's because we have wrapper inline functions to handle the variable(ex, oom_killer_[disable/enable]).
It means we are reluctant to use the global variable directly. So should we make new function as is_oom_killer_disable?

I think NO.

As I read your description, this problem is related to only hibernation. Since hibernation freezes all processes(include kswapd), this problem happens. Of course, now oom_killer_disabled is used by only hibernation. But it can be used others in future(Off-topic : I don't want it). Others can use it without freezing processes. Then kswapd can set zone->all_unreclaimable and the problem can't happen.

So I want to use sc->hibernation_mode which is already used do_try_to_free_pages instead of oom_killer_disabled.

Signature · 2 lines
Kind regards,
Minchan Kim
#2 lore raw

Hi

Thank you for good commenting!

Quoted 15 lines

I don't like use oom_killer_disabled directly. That's because we have wrapper inline functions to handle the variable(ex, oom_killer_[disable/enable]).
It means we are reluctant to use the global variable directly. So should we make new function as is_oom_killer_disable?

I think NO.

As I read your description, this problem is related to only hibernation. Since hibernation freezes all processes(include kswapd), this problem happens. Of course, now oom_killer_disabled is used by only hibernation. But it can be used others in future(Off-topic : I don't want it). Others can use it without freezing processes. Then kswapd can set zone->all_unreclaimable and the problem can't happen.

Quoted 2 lines

So I want to use sc->hibernation_mode which is already used do_try_to_free_pages instead of oom_killer_disabled.

Unfortunatelly, It's impossible. shrink_all_memory() turn on sc->hibernation_mode. but other hibernation caller merely call alloc_pages(). so we don't have any hint.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#3 lore raw
Quoted 20 lines · KOSAKI Motohiro

Hi

Thank you for good commenting!

Quoted 15 lines

I don't like use oom_killer_disabled directly. That's because we have wrapper inline functions to handle the variable(ex, oom_killer_[disable/enable]).
It means we are reluctant to use the global variable directly. So should we make new function as is_oom_killer_disable?

I think NO.

As I read your description, this problem is related to only hibernation. Since hibernation freezes all processes(include kswapd), this problem happens. Of course, now oom_killer_disabled is used by only hibernation. But it can be used others in future(Off-topic : I don't want it). Others can use it without freezing processes. Then kswapd can set zone->all_unreclaimable and the problem can't happen.

Quoted 7 lines
Quoted 2 lines

So I want to use sc->hibernation_mode which is already used do_try_to_free_pages instead of oom_killer_disabled.

Unfortunatelly, It's impossible. shrink_all_memory() turn on sc->hibernation_mode. but other hibernation caller merely call alloc_pages(). so we don't have any hint.

Ahh.. True. Sorry for that.
I will think some better method.
if I can't find it, I don't mind this patch. :)

Thanks.

Signature · 8 lines
Kind regards,
Minchan Kim

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#4 lore raw
Quoted 25 lines · Minchan Kim
Quoted 23 lines · KOSAKI Motohiro

Hi

Thank you for good commenting!

Quoted 17 lines

I don't like use oom_killer_disabled directly. That's because we have wrapper inline functions to handle the variable(ex, oom_killer_[disable/enable]).
It means we are reluctant to use the global variable directly. So should we make new function as is_oom_killer_disable?

I think NO.

As I read your description, this problem is related to only hibernation. Since hibernation freezes all processes(include kswapd), this problem happens. Of course, now oom_killer_disabled is used by only hibernation. But it can be used others in future(Off-topic : I don't want it). Others can use it without freezing processes. Then kswapd can set zone->all_unreclaimable and the problem can't happen.

So I want to use sc->hibernation_mode which is already used do_try_to_free_pages instead of oom_killer_disabled.

Quoted 7 lines
Quoted 4 lines

Unfortunatelly, It's impossible. shrink_all_memory() turn on sc->hibernation_mode. but other hibernation caller merely call alloc_pages(). so we don't have any hint.

Ahh.. True. Sorry for that.
I will think some better method.
if I can't find it, I don't mind this patch. :)

It seems that the poblem happens following as.
(I might miss something since I just read theyour description)

hibernation
oom_disable
alloc_pages
do_try_to_free_pages

        if (scanning_global_lru(sc) && !all_unreclaimable)
                return 1;

If kswapd is not freezed, it would set zone->all_unreclaimable to 1 and then shrink_zones maybe return true. so alloc_pages could go to _nopage_. If it is, it's no problem.
Right?

I think the problem would come from shrink_zones.
It set false to all_unreclaimable blindly even though shrink_zone can't reclaim any page. It doesn't make sense.
How about this?
I think we need this regardless of the problem.
What do you think about?

mm/vmscan.c +2 -1
index d8fd87d..22017b3 100644
@@ -1901,7 +1901,8 @@ static bool shrink_zones(int priority, struct zonelist *zonelist,
                }
 
                shrink_zone(priority, zone, sc);
-               all_unreclaimable = false;
+               if (sc->nr_reclaimed)
+                       all_unreclaimable = false;
        }
        return all_unreclaimable;
 }
Signature · 8 lines
Kind regards,
Minchan Kim

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#5 lore raw
Quoted 60 lines · Minchan Kim
Quoted 25 lines
Quoted 23 lines · KOSAKI Motohiro

Hi

Thank you for good commenting!

Quoted 17 lines

I don't like use oom_killer_disabled directly. That's because we have wrapper inline functions to handle the variable(ex, oom_killer_[disable/enable]).
It means we are reluctant to use the global variable directly. So should we make new function as is_oom_killer_disable?

I think NO.

As I read your description, this problem is related to only hibernation. Since hibernation freezes all processes(include kswapd), this problem happens. Of course, now oom_killer_disabled is used by only hibernation. But it can be used others in future(Off-topic : I don't want it). Others can use it without freezing processes. Then kswapd can set zone->all_unreclaimable and the problem can't happen.

So I want to use sc->hibernation_mode which is already used do_try_to_free_pages instead of oom_killer_disabled.

Quoted 7 lines
Quoted 4 lines

Unfortunatelly, It's impossible. shrink_all_memory() turn on sc->hibernation_mode. but other hibernation caller merely call alloc_pages(). so we don't have any hint.

Ahh.. True. Sorry for that.
I will think some better method.
if I can't find it, I don't mind this patch. :)

It seems that the poblem happens following as.
(I might miss something since I just read theyour description)

hibernation
oom_disable
alloc_pages
do_try_to_free_pages

        if (scanning_global_lru(sc) && !all_unreclaimable)
                return 1;

If kswapd is not freezed, it would set zone->all_unreclaimable to 1 and then shrink_zones maybe return true. so alloc_pages could go to _nopage_. If it is, it's no problem.
Right?

I think the problem would come from shrink_zones.
It set false to all_unreclaimable blindly even though shrink_zone can't reclaim any page. It doesn't make sense.
How about this?
I think we need this regardless of the problem.
What do you think about?

mm/vmscan.c +0 -0
index d8fd87d..22017b3 100644

mm/vmscan.c static bool shrink_zones(int priority, struct zonelist *zonelist, · line 1908

Quoted 9 lines
mm/vmscan.c +2 -1
@@ -1901,7 +1901,8 @@ static bool shrink_zones(int priority, struct zonelist *zonelist,
                }
 
                shrink_zone(priority, zone, sc);
-               all_unreclaimable = false;
+               if (sc->nr_reclaimed)
+                       all_unreclaimable = false;
        }
        return all_unreclaimable;
 }

here is brief of shrink_zones().

        for_each_zone_zonelist_nodemask(zone, z, zonelist,
                                        gfp_zone(sc->gfp_mask), sc->nodemask) {
                if (!populated_zone(zone))
                        continue;
                if (zone->all_unreclaimable && priority != DEF_PRIORITY)
                            continue;       /* Let kswapd poll it */
                shrink_zone(priority, zone, sc);
                all_unreclaimable = false;
        }

That said,

	all zone's zone->all_unreclaimable are true
		-> all_unreclaimable local variable become true.
	otherwise
		-> all_unreclaimable local variable become false.

The intention is, we don't want to invoke oom-killer if there are !all_unreclaimable zones. So your patch makes big design change and seems to increase OOM risk.

I don't want to send risky patch to -stable.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#6 lore raw
Quoted 88 lines · KOSAKI Motohiro
Quoted 60 lines · Minchan Kim
Quoted 25 lines
Quoted 23 lines · KOSAKI Motohiro

Hi

Thank you for good commenting!

Quoted 17 lines

I don't like use oom_killer_disabled directly. That's because we have wrapper inline functions to handle the variable(ex, oom_killer_[disable/enable]).
It means we are reluctant to use the global variable directly. So should we make new function as is_oom_killer_disable?

I think NO.

As I read your description, this problem is related to only hibernation. Since hibernation freezes all processes(include kswapd), this problem happens. Of course, now oom_killer_disabled is used by only hibernation. But it can be used others in future(Off-topic : I don't want it). Others can use it without freezing processes. Then kswapd can set zone->all_unreclaimable and the problem can't happen.

So I want to use sc->hibernation_mode which is already used do_try_to_free_pages instead of oom_killer_disabled.

Quoted 7 lines
Quoted 4 lines

Unfortunatelly, It's impossible. shrink_all_memory() turn on sc->hibernation_mode. but other hibernation caller merely call alloc_pages(). so we don't have any hint.

Ahh.. True. Sorry for that.
I will think some better method.
if I can't find it, I don't mind this patch. :)

It seems that the poblem happens following as.
(I might miss something since I just read theyour description)

hibernation
oom_disable
alloc_pages
do_try_to_free_pages
        if (scanning_global_lru(sc) && !all_unreclaimable) return 1;
If kswapd is not freezed, it would set zone->all_unreclaimable to 1 and then shrink_zones maybe return true. so alloc_pages could go to _nopage_. If it is, it's no problem.
Right?

I think the problem would come from shrink_zones.
It set false to all_unreclaimable blindly even though shrink_zone can't reclaim any page. It doesn't make sense.
How about this?
I think we need this regardless of the problem.
What do you think about?

mm/vmscan.c +0 -0
index d8fd87d..22017b3 100644

mm/vmscan.c static bool shrink_zones(int priority, struct zonelist *zonelist, · line 1908

Quoted 9 lines
mm/vmscan.c +2 -1
@@ -1901,7 +1901,8 @@ static bool shrink_zones(int priority, struct zonelist *zonelist,
                 }
 
                 shrink_zone(priority, zone, sc);
-               all_unreclaimable = false;
+               if (sc->nr_reclaimed)
+                       all_unreclaimable = false;
         }
         return all_unreclaimable;
  }

here is brief of shrink_zones().

       for_each_zone_zonelist_nodemask(zone, z, zonelist, gfp_zone(sc->gfp_mask), sc->nodemask) { if (!populated_zone(zone))
                       continue;
               if (zone->all_unreclaimable && priority != DEF_PRIORITY) continue;       /* Let kswapd poll it */ shrink_zone(priority, zone, sc);
               all_unreclaimable = false;
       }

That said,
       all zone's zone->all_unreclaimable are true
               -> all_unreclaimable local variable become true. otherwise
               -> all_unreclaimable local variable become false.

Quoted 3 lines

The intention is, we don't want to invoke oom-killer if there are !all_unreclaimable zones. So your patch makes big design change and seems to increase OOM risk.

Right. Thanks for pointing me out.

Quoted 1 line

I don't want to send risky patch to -stable.

Still I don't want to use oom_killer_disabled magic.
But I don't want to prevent urgent stable patch due to my just nitpick.

This is my last try(just quick patch, even I didn't tried compile test). If this isn't good, first of all, let's try merge yours.
And then we can fix it later.

Thanks for comment.

-- CUT HERE --

Why do we check zone->all_unreclaimable in only kswapd? If kswapd is freezed in hibernation, OOM can happen. Let's the check in direct reclaim path, too.

mm/vmscan.c +1 -2
index 3109ff7..41493ba 100644
@@ -1878,12 +1878,11 @@ static void shrink_zone(int priority, struct zone *zone,
 * If a zone is deemed to be full of pinned pages then just give it a light
  * scan then give up on it.
  */
-static bool shrink_zones(int priority, struct zonelist *zonelist,
+static void shrink_zones(int priority, struct zonelist *zonelist,
                                        struct scan_control *sc)
 {
        struct zoneref *z;
        struct zone *zone;
-       bool all_unreclaimable = true;
 
        for_each_zone_zonelist_nodemask(zone, z, zonelist,
                                        gfp_zone(sc->gfp_mask), sc->nodemask) {
@@ -1901,8 +1900,25 @@ static bool shrink_zones(int priority, struct

zonelist *zonelist,

                }

                shrink_zone(priority, zone, sc);
-               all_unreclaimable = false;
        }

+}
+
+static inline int all_unreclaimable(struct zonelist *zonelist, struct scan_control *sc)
+{

+       struct zoneref *z;
+       struct zone *zone;
+       bool all_unreclaimable = true;

+
+ for_each_zone_zonelist_nodemask(zone, z, zonelist,
+ gfp_zone(sc->gfp_mask), sc->nodemask) {
+ if (!populated_zone(zone))
+ continue;
+ if (zone->pages_scanned < (zone_reclaimable_pages(zone) * 6)) {
+ all_unreclaimable = false;
+ break;
+ }
+ }
+

        return all_unreclaimable;

}

@@ -1926,7 +1942,6 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
struct scan_control *sc)

 {
        int priority;
-       bool all_unreclaimable;
        unsigned long total_scanned = 0;
        struct reclaim_state *reclaim_state = current->reclaim_state;
        struct zoneref *z;
@@ -1943,7 +1958,7 @@ static unsigned long do_try_to_free_pages(struct
zonelist *zonelist,
                sc->nr_scanned = 0;
                if (!priority)
                        disable_swap_token();

- all_unreclaimable = shrink_zones(priority, zonelist, sc);
+ shrink_zones(priority, zonelist, sc);
/*
* Don't shrink slabs when reclaiming memory from
* over limit cgroups
@@ -2005,7 +2020,7 @@ out:

                return sc->nr_reclaimed;

        /* top priority shrink_zones still had more to do? don't OOM, then */
-       if (scanning_global_lru(sc) && !all_unreclaimable)
+       if (scanning_global_lru(sc) && !all_unreclaimable(zonelist, sc))
                return 1;

        return 0;
Signature · 2 lines
Kind regards,
Minchan Kim
mm/vmscan.c +21 -6
index 3109ff7..41493ba 100644
@@ -1878,12 +1878,11 @@ static void shrink_zone(int priority, struct zone *zone,
  * If a zone is deemed to be full of pinned pages then just give it a light
  * scan then give up on it.
  */
-static bool shrink_zones(int priority, struct zonelist *zonelist,
+static void shrink_zones(int priority, struct zonelist *zonelist,
 					struct scan_control *sc)
 {
 	struct zoneref *z;
 	struct zone *zone;
-	bool all_unreclaimable = true;
 
 	for_each_zone_zonelist_nodemask(zone, z, zonelist,
 					gfp_zone(sc->gfp_mask), sc->nodemask) {
@@ -1901,8 +1900,25 @@ static bool shrink_zones(int priority, struct zonelist *zonelist,
 		}
 
 		shrink_zone(priority, zone, sc);
-		all_unreclaimable = false;
 	}
+}
+
+static inline int all_unreclaimable(struct zonelist *zonelist, struct scan_control *sc)
+{
+	struct zoneref *z;
+	struct zone *zone;
+	bool all_unreclaimable = true;
+
+	for_each_zone_zonelist_nodemask(zone, z, zonelist,
+					gfp_zone(sc->gfp_mask), sc->nodemask) {
+		if (!populated_zone(zone))
+			continue;
+		if (zone->pages_scanned < (zone_reclaimable_pages(zone) * 6)) {
+			all_unreclaimable = false;
+			break;
+		}
+	}
+
 	return all_unreclaimable;
 }
 
@@ -1926,7 +1942,6 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 					struct scan_control *sc)
 {
 	int priority;
-	bool all_unreclaimable;
 	unsigned long total_scanned = 0;
 	struct reclaim_state *reclaim_state = current->reclaim_state;
 	struct zoneref *z;
@@ -1943,7 +1958,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		sc->nr_scanned = 0;
 		if (!priority)
 			disable_swap_token();
-		all_unreclaimable = shrink_zones(priority, zonelist, sc);
+		shrink_zones(priority, zonelist, sc);
 		/*
 		 * Don't shrink slabs when reclaiming memory from
 		 * over limit cgroups
@@ -2005,7 +2020,7 @@ out:
 		return sc->nr_reclaimed;
 
 	/* top priority shrink_zones still had more to do? don't OOM, then */
-	if (scanning_global_lru(sc) && !all_unreclaimable)
+	if (scanning_global_lru(sc) && !all_unreclaimable(zonelist, sc))
 		return 1;
 
 	return 0;
#7 lore raw
Quoted 6 lines
Quoted 1 line

I don't want to send risky patch to -stable.

Still I don't want to use oom_killer_disabled magic.
But I don't want to prevent urgent stable patch due to my just nitpick.

This is my last try(just quick patch, even I didn't tried compile test).

Looks like conceptually correct. If you will test it and fix whitespace damage, I'll ack this one gladly.

Thanks.

Quoted 92 lines

If this isn't good, first of all, let's try merge yours. And then we can fix it later.

Thanks for comment.

-- CUT HERE --

Why do we check zone->all_unreclaimable in only kswapd? If kswapd is freezed in hibernation, OOM can happen. Let's the check in direct reclaim path, too.

mm/vmscan.c +19 -3
index 3109ff7..41493ba 100644
@@ -1878,12 +1878,11 @@ static void shrink_zone(int priority, struct zone *zone,
 * If a zone is deemed to be full of pinned pages then just give it a light
  * scan then give up on it.
  */
-static bool shrink_zones(int priority, struct zonelist *zonelist,
+static void shrink_zones(int priority, struct zonelist *zonelist,
                                        struct scan_control *sc)
 {
        struct zoneref *z;
        struct zone *zone;
-       bool all_unreclaimable = true;
 
        for_each_zone_zonelist_nodemask(zone, z, zonelist,
                                        gfp_zone(sc->gfp_mask), sc->nodemask) {
@@ -1901,8 +1900,25 @@ static bool shrink_zones(int priority, struct
 zonelist *zonelist,
                }
 
                shrink_zone(priority, zone, sc);
-               all_unreclaimable = false;
        }
+}
+
+static inline int all_unreclaimable(struct zonelist *zonelist, struct
 scan_control *sc)
+{
+       struct zoneref *z;
+       struct zone *zone;
+       bool all_unreclaimable = true;
+
+       for_each_zone_zonelist_nodemask(zone, z, zonelist,
+                                       gfp_zone(sc->gfp_mask), sc->nodemask) {
+               if (!populated_zone(zone))
+                       continue;
+               if (zone->pages_scanned < (zone_reclaimable_pages(zone) * 6)) {
+                       all_unreclaimable = false;
+                       break;
+               }
+       }
+
        return all_unreclaimable;

}

mm/vmscan.c +0 -1
@@ -1926,7 +1942,6 @@ static unsigned long do_try_to_free_pages(struct
 zonelist *zonelist,
                                        struct scan_control *sc)
 {
        int priority;
-       bool all_unreclaimable;
        unsigned long total_scanned = 0;
        struct reclaim_state *reclaim_state = current->reclaim_state;
        struct zoneref *z;
mm/vmscan.c +1 -1
@@ -1943,7 +1958,7 @@ static unsigned long do_try_to_free_pages(struct
 zonelist *zonelist,
                sc->nr_scanned = 0;
                if (!priority)
                        disable_swap_token();
-               all_unreclaimable = shrink_zones(priority, zonelist, sc);
+               shrink_zones(priority, zonelist, sc);
                /*
                 * Don't shrink slabs when reclaiming memory from

* over limit cgroups

mm/vmscan.c +1 -1
@@ -2005,7 +2020,7 @@ out:
                return sc->nr_reclaimed;
 
        /* top priority shrink_zones still had more to do? don't OOM, then */
-       if (scanning_global_lru(sc) && !all_unreclaimable)
+       if (scanning_global_lru(sc) && !all_unreclaimable(zonelist, sc))
                return 1;
 
        return 0;
Signature · 3 lines
-- 
Kind regards,
Minchan Kim

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#8 lore raw
Quoted 5 lines · KOSAKI Motohiro
Quoted 5 lines
Quoted 1 line

I don't want to send risky patch to -stable.

Still I don't want to use oom_killer_disabled magic.
But I don't want to prevent urgent stable patch due to my just nitpick.

Quoted 7 lines
Quoted 1 line

This is my last try(just quick patch, even I didn't tried compile test).

Looks like conceptually correct. If you will test it and fix whitespace damage, I'll ack this one gladly.

Thanks.

I will resend formal patch ASAP. Maybe tonight after out of office. Thanks for quick reply.

Signature · 8 lines
Kind regards,
Minchan Kim

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>