[PATCH rh6] mm: skip zombie in OOM-killer

[PATCH rh6] mm: skip zombie in OOM-killer

[Andrey Vagin]

A parent may wait a memory and zombie will prevent killing another task.

Signed-off-by: Andrey Vagin <avagin@openvz.org>
---
 mm/oom_kill.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index 7dcca55..2fc554e 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -311,7 +311,7 @@ static struct task_struct *select_bad_process(unsigned int *ppoints,
 		 * blocked waiting for another task which itself is waiting
 		 * for memory. Is there a better alternative?
 		 */
-		if (test_tsk_thread_flag(p, TIF_MEMDIE))
+		if (test_tsk_thread_flag(p, TIF_MEMDIE) && p->mm)
 			return ERR_PTR(-1UL);
 
 		/*
-- 
1.7.1

Re: [PATCH rh6] mm: skip zombie in OOM-killer

[David Rientjes]

On Sat, 5 Mar 2011, Andrey Vagin wrote:

> diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> index 7dcca55..2fc554e 100644
> --- a/mm/oom_kill.c
> +++ b/mm/oom_kill.c
> @@ -311,7 +311,7 @@ static struct task_struct *select_bad_process(unsigned int *ppoints,
>  		 * blocked waiting for another task which itself is waiting
>  		 * for memory. Is there a better alternative?
>  		 */
> -		if (test_tsk_thread_flag(p, TIF_MEMDIE))
> +		if (test_tsk_thread_flag(p, TIF_MEMDIE) && p->mm)
>  			return ERR_PTR(-1UL);
>  
>  		/*

I think it would be better to just do

	if (!p->mm)
		continue;

after the check for oom_unkillable_task() because everything that follows 
this really depends on p->mm being non-NULL to actually do anything 
useful.

Re: [PATCH rh6] mm: skip zombie in OOM-killer

[avagin]

On 03/05/2011 02:41 AM, David Rientjes wrote:
> On Sat, 5 Mar 2011, Andrey Vagin wrote:
>
>> diff --git a/mm/oom_kill.c b/mm/oom_kill.c
>> index 7dcca55..2fc554e 100644
>> --- a/mm/oom_kill.c
>> +++ b/mm/oom_kill.c
>> @@ -311,7 +311,7 @@ static struct task_struct *select_bad_process(unsigned int *ppoints,
>>   		 * blocked waiting for another task which itself is waiting
>>   		 * for memory. Is there a better alternative?
>>   		 */
>> -		if (test_tsk_thread_flag(p, TIF_MEMDIE))
>> +		if (test_tsk_thread_flag(p, TIF_MEMDIE)&&  p->mm)
>>   			return ERR_PTR(-1UL);
>>
>>   		/*
>
> I think it would be better to just do
>
> 	if (!p->mm)
> 		continue;
>
> after the check for oom_unkillable_task() because everything that follows
> this really depends on p->mm being non-NULL to actually do anything
> useful.
Yes. You are right. Thank you.

Re: [PATCH rh6] mm: skip zombie in OOM-killer

[KOSAKI Motohiro]

> On Sat, 5 Mar 2011, Andrey Vagin wrote:
> 
> > diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> > index 7dcca55..2fc554e 100644
> > --- a/mm/oom_kill.c
> > +++ b/mm/oom_kill.c
> > @@ -311,7 +311,7 @@ static struct task_struct *select_bad_process(unsigned int *ppoints,
> >  		 * blocked waiting for another task which itself is waiting
> >  		 * for memory. Is there a better alternative?
> >  		 */
> > -		if (test_tsk_thread_flag(p, TIF_MEMDIE))
> > +		if (test_tsk_thread_flag(p, TIF_MEMDIE) && p->mm)
> >  			return ERR_PTR(-1UL);
> >  
> >  		/*
> 
> I think it would be better to just do
> 
> 	if (!p->mm)
> 		continue;
> 
> after the check for oom_unkillable_task() because everything that follows 
> this really depends on p->mm being non-NULL to actually do anything 
> useful.

I'm glad you join to review MM patches. It is worth effort for making
solid kernel. But, please look at a current code at first.

Re: [PATCH rh6] mm: skip zombie in OOM-killer

[KOSAKI Motohiro]

> A parent may wait a memory and zombie will prevent killing another task.
> 
> Signed-off-by: Andrey Vagin <avagin@openvz.org>
> ---
>  mm/oom_kill.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> index 7dcca55..2fc554e 100644
> --- a/mm/oom_kill.c
> +++ b/mm/oom_kill.c
> @@ -311,7 +311,7 @@ static struct task_struct *select_bad_process(unsigned int *ppoints,
>  		 * blocked waiting for another task which itself is waiting
>  		 * for memory. Is there a better alternative?
>  		 */
> -		if (test_tsk_thread_flag(p, TIF_MEMDIE))
> +		if (test_tsk_thread_flag(p, TIF_MEMDIE) && p->mm)
>  			return ERR_PTR(-1UL);

OK. Good catch.

	Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>