Re: mmotm 2009-06-02-16-11 uploaded (staging)

[Andrew Morton <akpm@linux-foundation.org>]

On Tue, 2 Jun 2009 21:45:15 -0700 (PDT) David Rientjes <rientjes@google.com> wrote:

> On Tue, 2 Jun 2009, Randy Dunlap wrote:
> 
> > drivers/staging/android/lowmemorykiller.c:106: error: 'struct task_struct' has no member named 'oomkilladj'
> > 
> 
> Thanks Randy.
> 
> This is due to oom-move-oom_adj-value-from-task_struct-to-mm_struct.patch 
> which was merged in mmotm early this morning.
> 
> I had previously fixed this in an earlier version of the patch series, but 
> people didn't agree that an oom killer change should touch staging files 
> even though the Android lowmemorykiller is in both mmotm and git HEAD.  
> See http://marc.info/?l=linux-kernel&m=124200164711314.
> 
> I'm still a little curious about why it was decided like that since it 
> seems to be the simplest way to avoid these failures, which leave you with 
> a broken build, and an unnecessary burden on maintainers to magically fix 
> them up later (which I guess is now?).
> 
> So when I moved oomkilladj from struct task_struct to struct mm_struct and 
> renamed it to something more appropriate, I still think I should have been 
> able to change drivers/staging/android/lowmemorykiller.c for those 
> reasons and avoid all of this hassle.
> 
> Anyway, here's a patch that should work and will hopefully now be 
> accepted.  I know Greg is the maintainer for the staging tree and we're 
> talking about mmotm, but I don't know how else to do it anymore.

I think there's been a miscommunication somewhere.  Fixing the fallout
from a kernel-core change within the same patch is the correct thing to
do.

> diff --git a/drivers/staging/android/lowmemorykiller.c b/drivers/staging/android/lowmemorykiller.c
> --- a/drivers/staging/android/lowmemorykiller.c
> +++ b/drivers/staging/android/lowmemorykiller.c
> @@ -96,19 +96,21 @@ static int lowmem_shrink(int nr_to_scan, gfp_t gfp_mask)
>  
>  	read_lock(&tasklist_lock);
>  	for_each_process(p) {
> +		struct mm_struct *mm;
>  		int oom_adj;
>  
>  		task_lock(p);
> -		if (!p->mm) {
> +		mm = p->mm;
> +		if (!mm) {
>  			task_unlock(p);
>  			continue;
>  		}
> -		oom_adj = p->oomkilladj;
> +		oom_adj = mm->oom_adj;
>  		if (oom_adj < min_adj) {
>  			task_unlock(p);
>  			continue;
>  		}
> -		tasksize = get_mm_rss(p->mm);
> +		tasksize = get_mm_rss(mm);
>  		task_unlock(p);
>  		if (tasksize <= 0)
>  			continue;

Thanks.