Re: [PATCH 3/3] compaction accouting fix

From: Minchan Kim
Date: Wed Aug 31 2011 - 10:57:02 EST


On Wed, Aug 31, 2011 at 01:37:10PM +0200, Johannes Weiner wrote:
> On Sun, Nov 13, 2011 at 01:37:43AM +0900, Minchan Kim wrote:
> > I saw the following accouting of compaction during test of the series.
> >
> > compact_blocks_moved 251
> > compact_pages_moved 44
> >
> > It's very awkward to me although it's possbile because it means we try to compact 251 blocks
> > but it just migrated 44 pages. As further investigation, I found isolate_migratepages doesn't
> > isolate any pages but it returns ISOLATE_SUCCESS and then, it just increases compact_blocks_moved
> > but doesn't increased compact_pages_moved.
> >
> > This patch makes accouting of compaction works only in case of success of isolation.
> >
> > CC: Mel Gorman <mgorman@xxxxxxx>
> > CC: Johannes Weiner <jweiner@xxxxxxxxxx>
> > CC: Rik van Riel <riel@xxxxxxxxxx>
> > Signed-off-by: Minchan Kim <minchan.kim@xxxxxxxxx>
>
> Acked-by: Johannes Weiner <jweiner@xxxxxxxxxx>

Thanks, Hannes.

>
> It's a teensy-bit awkward that isolate_migratepages() can return
> success without actually isolating any new pages, just because there
> are still some pages left from a previous run (cc->nr_migratepages is
> maintained across isolation calls).

If migrate_pages fails, we reset cc->nr_migratepages to zero in compact_zone.
Am I missing something?

>
> Maybe isolate_migratepages() should just return an error if compaction
> should really be aborted and 0 otherwise, and have compact_zone()
> always check for cc->nr_migratepages itself?
>
> if (isolate_migratepages(zone, cc) < 0) {
> ret = COMPACT_PARTIAL;
> goto out;
> }
>
> if (!cc->nr_migratepages)
> continue;
>
> ...
>
> Just a nit-pick, though. If you don't agree, just leave it as is.

--
Kind regards,
Minchan Kim
--
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/