Re: [PATCH]oom-kill: direct hardware access processes should get bonus

[David Rientjes <rientjes@google.com>]

On Tue, 2 Nov 2010, Figo.zhang wrote:

> the victim should not directly access hardware devices like Xorg server,
> because the hardware could be left in an unpredictable state, although 
> user-application can set /proc/pid/oom_score_adj to protect it. so i think
> those processes should get 3% bonus for protection.
> 

Which applications are you referring to that cannot gracefully exit if 
killed?

> Signed-off-by: Figo.zhang <figo1802@gmail.com>
> ---
> mm/oom_kill.c |    8 +++++---
>  1 files changed, 5 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> index 4029583..df6a9da 100644
> --- a/mm/oom_kill.c
> +++ b/mm/oom_kill.c
> @@ -195,10 +195,12 @@ unsigned int oom_badness(struct task_struct *p, struct mem_cgroup *mem,
>  	task_unlock(p);
>  
>  	/*
> -	 * Root processes get 3% bonus, just like the __vm_enough_memory()
> -	 * implementation used by LSMs.
> +	 * Root and direct hardware access processes get 3% bonus, just like the
> +	 * __vm_enough_memory() implementation used by LSMs.

LSM's have this bonus for CAP_SYS_ADMIN, but not for CAP_SYS_RAWIO, so 
this comment is incorrect.

>  	 */
> -	if (has_capability_noaudit(p, CAP_SYS_ADMIN))
> +	if (has_capability_noaudit(p, CAP_SYS_ADMIN) ||
> +	    has_capability_noaudit(p, CAP_SYS_RESOURCE) ||
> +	    has_capability_noaudit(p, CAP_SYS_RAWIO))
>  		points -= 30;
>  
>  	/*

CAP_SYS_RAWIO had a much more dramatic impact in the previous heuristic to 
such a point that it would often allow memory hogging tasks to elude the 
oom killer at the expense of innocent tasks.  I'm not sure this is the 
best way to go.