Re: [patch -mm v2] mm: introduce oom_adj_child

["KAMEZAWA Hiroyuki" <kamezawa.hiroyu@jp.fujitsu.com>]

David Rientjes wrote:

>> > It livelocks if a thread is chosen and passed to oom_kill_task() while
>> > another per-thread oom_adj value is OOM_DISABLE for a thread sharing
>> the
>> > same memory.
>> >
>> I say "why don't modify buggy selection logic?"
>>
>> Why we have to scan all threads ?
>> As fs/proc/readdir does, you can scan only "process group leader".
>>
>> per-thread scan itself is buggy because now we have per-process
>> effective-oom-adj.
>>
>
> Without my patches to change oom_adj from task_struct to mm_struct, you'd
> need to scan all tasks and not just the tgids because their oom_adj values
> can differ amongst threads in the same thread group.  So while it may now
> be possible to shorten the scan as a result of my approach, it isn't a
> solution itself to the problem.

Did I said "revert your patch in -rc" even once ?
livelock-avoidance itself is good work, thank you.
All my suggestion is based on your patch already in rc4.
Summarizing I think now .....
  - rename mm->oom_adj as mm->effective_oom_adj
  - re-add per-thread oom_adj
  - update mm->effective_oom_adj based on per-thread oom_adj
  - if necessary, plz add read-only /proc/pid/effective_oom_adj file.
    or show 2 values in /proc/pid/oom_adj
  - rewrite documentation about oom_score.
   " it's calclulated from  _process's_ memory usage and oom_adj of
    all threads which shares a memor  context".
   This behavior is not changed from old implemtation, anyway.
 - If necessary, rewrite oom_kill itself to scan only thread group
   leader. It's a way to go regardless of  vfork problem.



>
>> > How else do you propose the oom killer use oom_adj values on a
>> per-thread
>> > basis without considering other threads sharing the same memory?
>> As I wrote.
>>    per-process(signal struct) or per-thread oom_adj and add
>>    mm->effecitve_oom_adj
>>
>> task scanning isn't necessary to do per-thread scan and you can scan
>> only process-group-leader. What's bad ?
>> If oom_score is problem, plz fix it to show effective_oom_score.
>>
>
> When only using (and showing) mm->effective_oom_adj for a task, userspace
> will not be able to adjust /proc/pid/oom_score with /proc/pid/oom_adj
> as Documentation/filesystems/proc.txt says you can for a thread unless it
> exceeds effective_oom_adj.>

Is it different from old behavior ?
I think documentation is wrong. It should say "you should think of
multi-thread effect to oom_adj/oom_score".

Thanks,
-Kame

> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>