[BUG][PATCH] fix oom killer kills current every time if there is memory-less-node

[BUG][PATCH] fix oom killer kills current every time if there is memory-less-node

[KAMEZAWA Hiroyuki]

constrained_alloc(), which is called to detect where oom is from,
checks passed zone_list().
If zone_list includes all nodes, it thinks oom is from mempolicy.

But there is memory-less-node. contstrained_alloc() should get
memory_less_node into count. Otherwise, current process will die
at any time. This patch fix it.

Signed-Off-By: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>



 mm/oom_kill.c |    7 ++++++-
 1 files changed, 6 insertions(+), 1 deletion(-)

Index: devel-2.6.20-rc1-mm1/mm/oom_kill.c
===================================================================
--- devel-2.6.20-rc1-mm1.orig/mm/oom_kill.c	2006-12-16 13:47:59.000000000 +0900
+++ devel-2.6.20-rc1-mm1/mm/oom_kill.c	2006-12-22 12:11:55.000000000 +0900
@@ -174,7 +174,12 @@
 {
 #ifdef CONFIG_NUMA
 	struct zone **z;
-	nodemask_t nodes = node_online_map;
+	nodemask_t nodes;
+	int node;
+	/* node has memory ? */
+	for_each_online_node(node)
+		if (NODE_DATA(node)->node_present_pages)
+			node_set(node, nodes);
 
 	for (z = zonelist->zones; *z; z++)
 		if (cpuset_zone_allowed_softwall(*z, gfp_mask))

Re: [BUG][PATCH] fix oom killer kills current every time if there is memory-less-node

[Paul Jackson]

KAMEZAWA-san wrote:
> But there is memory-less-node. contstrained_alloc() should get
> memory_less_node into count.

This patch looks ok to me.

One line in the patch comment seems backward:

  If zone_list includes all nodes, it thinks oom is from mempolicy.

Shouldn't that be:

  If zone_list doesn't include all nodes, it thinks oom is from mempolicy.

-- 
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <pj@sgi.com> 1.925.600.0401

Re: [BUG][PATCH] fix oom killer kills current every time if there is memory-less-node

[KAMEZAWA Hiroyuki]

On Thu, 21 Dec 2006 21:18:12 -0800
Paul Jackson <pj@sgi.com> wrote:

> KAMEZAWA-san wrote:
> > But there is memory-less-node. contstrained_alloc() should get
> > memory_less_node into count.
> 
> This patch looks ok to me.
> 
> One line in the patch comment seems backward:
> 
>   If zone_list includes all nodes, it thinks oom is from mempolicy.
> 
> Shouldn't that be:
> 
>   If zone_list doesn't include all nodes, it thinks oom is from mempolicy.
> 
Ah, yes. (>_<, Thank you.

-Kame

[BUG][PATCH] fix oom killer kills current every time if there is memory-less-node take2

[KAMEZAWA Hiroyuki]

Fixed patch comment.

-Kame
==
constrained_alloc(), which is called to detect where oom is from,
checks passed zone_list().
If zone_list doesn't include all nodes, it thinks oom is from mempolicy.

But there is memory-less-node. memory-less-node's zones are never included in
zonelist[].

contstrained_alloc() should get memory_less_node into count. 
Otherwise, it always thinks 'oom is from mempolicy'.
This means that current process dies at any time. This patch fix it.

Signed-Off-By: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>



 mm/oom_kill.c |    7 ++++++-
 1 files changed, 6 insertions(+), 1 deletion(-)

Index: devel-2.6.20-rc1-mm1/mm/oom_kill.c
===================================================================
--- devel-2.6.20-rc1-mm1.orig/mm/oom_kill.c	2006-12-16 13:47:59.000000000 +0900
+++ devel-2.6.20-rc1-mm1/mm/oom_kill.c	2006-12-22 12:11:55.000000000 +0900
@@ -174,7 +174,12 @@
 {
 #ifdef CONFIG_NUMA
 	struct zone **z;
-	nodemask_t nodes = node_online_map;
+	nodemask_t nodes;
+	int node;
+	/* node has memory ? */
+	for_each_online_node(node)
+		if (NODE_DATA(node)->node_present_pages)
+			node_set(node, nodes);
 
 	for (z = zonelist->zones; *z; z++)
 		if (cpuset_zone_allowed_softwall(*z, gfp_mask))