Re: [PATCH] mm/oom_kill: fix stale comment in constrained_alloc()

[Michal Hocko]

On Thu 13-08-26 11:06:54, Ye Liu wrote:
> From: Ye Liu <liuye@xxxxxxxxxx>
> 
> The comment above the __GFP_THISNODE check in constrained_alloc()
> is stale and inaccurate: it references __GFP_NOFAIL (which is not
> checked here), claims the OOM killer does "random task kill" (it
> uses badness scores), and mentions CONSTRAINT_THISNODE (which was
> never added to the enum).  Replace it with a description that
> matches the actual code behavior.

The comment is indeed confusing and stale. Your update unfortunatelly as
well. This code is simply unreachable because we never trigger global
OOM with __GFP_THISNODE (see __alloc_pages_may_oom) and for memcg OOM
__GFP_THISNODE is immaterial.

> Signed-off-by: Ye Liu <liuye@xxxxxxxxxx>
> ---
>  mm/oom_kill.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> index 5f372f6e26fa..4b5c70aaece4 100644
> --- a/mm/oom_kill.c
> +++ b/mm/oom_kill.c
> @@ -268,9 +268,9 @@ static enum oom_constraint constrained_alloc(struct oom_control *oc)
>  	if (!oc->zonelist)
>  		return CONSTRAINT_NONE;
>  	/*
> -	 * Reach here only when __GFP_NOFAIL is used. So, we should avoid
> -	 * to kill current.We have to random task kill in this case.
> -	 * Hopefully, CONSTRAINT_THISNODE...but no way to handle it, now.
> +	 * __GFP_THISNODE means the allocation is restricted to a single node.
> +	 * There is no dedicated constraint type for this case, so return
> +	 * CONSTRAINT_NONE and let the normal OOM handling proceed.
>  	 */
>  	if (oc->gfp_mask & __GFP_THISNODE)
>  		return CONSTRAINT_NONE;
> -- 
> 2.25.1

-- 
Michal Hocko
SUSE Labs