Re: [PATCH] Revert "mm: vmscan: fix misused nr_reclaimed in shrink_mem_cgroup_zone()"

[Ying Han <yinghan@google.com>]

On Tue, Apr 10, 2012 at 8:00 AM, Hillf Danton <dhillf@gmail.com> wrote:
> On Tue, Apr 10, 2012 at 3:42 AM, Ying Han <yinghan@google.com> wrote:
>> This reverts commit c38446cc65e1f2b3eb8630c53943b94c4f65f670.
>>
>> Before the commit, the code makes senses to me but not after the commit. The
>> "nr_reclaimed" is the number of pages reclaimed by scanning through the memcg's
>> lru lists. The "nr_to_reclaim" is the target value for the whole function. For
>> example, we like to early break the reclaim if reclaimed 32 pages under direct
>> reclaim (not DEF_PRIORITY).
>>
>> After the reverted commit, the target "nr_to_reclaim" is decremented each time
>> by "nr_reclaimed" but we still use it to compare the "nr_reclaimed". It just
>> doesn't make sense to me...
>>
> I downloaded mm/vmscan.c from the next tree a couple minutes ago, and
> see
>                .nr_to_reclaim = SWAP_CLUSTER_MAX,
> and
>        nr_reclaimed = do_try_to_free_pages(zonelist, &sc, &shrink);
>
> in try_to_free_pages().
>
> I also see
>                total_scanned += sc->nr_scanned;
>                if (sc->nr_reclaimed >= sc->nr_to_reclaim)
>                        goto out;


> in do_try_to_free_pages(),
>
> then would you please say a few words about the sense
> of the check of nr_to_reclaim?

There are two places where we do early break out in direct reclaim path.

1. For each priority loop after calling shrink_zones(), we check
(sc->nr_reclaimed >= sc->nr_to_reclaim)

2. For each memcg reclaim (shrink_mem_cgroup_zone) under
shrink_zone(), we check (nr_reclaimed >= nr_to_reclaim)

The second one says "if 32 (nr_to_reclaim) pages being reclaimed from
this memcg under high priority, break". This check is necessary here
to prevent over pressure each memcg under shrink_zone().

Regarding the reverted patch, it tries to convert the "nr_reclaimed"
to "total_reclaimed" for outer loop (restart). First of all, it
changes the logic by doing less work each time
should_continue_reclaim() is true. Second, the fix is simply broken by
decrementing nr_to_reclaim each time.

--Ying

>
>
>> Signed-off-by: Ying Han <yinghan@google.com>
>> ---
>>  mm/vmscan.c |    7 +------
>>  1 files changed, 1 insertions(+), 6 deletions(-)
>>
>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>> index 33c332b..1a51868 100644
>> --- a/mm/vmscan.c
>> +++ b/mm/vmscan.c
>> @@ -2107,12 +2107,7 @@ restart:
>>                 * with multiple processes reclaiming pages, the total
>>                 * freeing target can get unreasonably large.
>>                 */
>> -               if (nr_reclaimed >= nr_to_reclaim)
>> -                       nr_to_reclaim = 0;
>> -               else
>> -                       nr_to_reclaim -= nr_reclaimed;
>> -
>> -               if (!nr_to_reclaim && priority < DEF_PRIORITY)
>> +               if (nr_reclaimed >= nr_to_reclaim && priority < DEF_PRIORITY)
>>                        break;
>>        }
>>        blk_finish_plug(&plug);
>> --
>> 1.7.7.3
>>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>