Re: [PATCH 2/5] Revert "oom: give the dying task a higher priority"

[Peter Zijlstra]

On Mon, 2011-03-28 at 10:10 -0300, Luis Claudio R. Goncalves wrote:
> | There was meaningless code in there. I guess it was in there from CFS.
> | Thanks for the explanation, Peter.
> 
> Yes, it was CFS related:
> 
>         p = find_lock_task_mm(p);
>         ...
>         p->rt.time_slice = HZ; <<---- THIS

CFS has never used rt.time_slice, that's always been a pure SCHED_RR
thing.

> Peter, would that be effective to boost the priority of the dying task?

The thing you're currently doing, making it SCHED_FIFO ?

> I mean, in the context of SCHED_OTHER tasks would it really help the dying
> task to be scheduled sooner to release its resources? 

That very much depends on how all this stuff works, I guess if everybody
serializes on OOM and only the first will actually kill a task and all
the waiting tasks will try to allocate a page again before also doing
the OOM thing, and the pending tasks are woken after the OOM target task
has completed dying.. then I don't see much point in boosting things,
since everybody interested in memory will block and eventually only the
dying task will be left running.

Its been a very long while since I stared at the OOM code..

> If so, as we remove
> the code in commit 93b43fa5508 we should re-add that old code. 

It doesn't make any sense to fiddle with rt.time_slice afaict.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/