Re: [PATCH 05/11] mm, compaction: distinguish COMPACT_DEFERRED from COMPACT_SKIPPED

[Michal Hocko]

On Mon 11-04-16 13:02:20, Vlastimil Babka wrote:
> On 04/05/2016 01:25 PM, Michal Hocko wrote:
> >From: Michal Hocko <mhocko@xxxxxxxx>
> >
> >try_to_compact_pages can currently return COMPACT_SKIPPED even when the
> >compaction is defered for some zone just because zone DMA is skipped
> >in 99% of cases due to watermark checks. This makes COMPACT_DEFERRED
> >basically unusable for the page allocator as a feedback mechanism.
> >
> >Make sure we distinguish those two states properly and switch their
> >ordering in the enum. This would mean that the COMPACT_SKIPPED will be
> >returned only when all eligible zones are skipped.
> >
> >This shouldn't introduce any functional change.
> 
> Hmm, really? __alloc_pages_direct_compact() does distinguish
> COMPACT_DEFERRED, and sets *deferred compaction, so ultimately this is some
> change for THP allocations?

Hmm, you are right. In cases where we would return COMPACTION_SKIPED
even though there is a zone which would really like to tell us
COMPACT_DEFERRED then we would previously did __alloc_pages_direct_reclaim
and then bail out for THP which do not have __GFP_REPEAT while now we
would recognize DEFERRED and bail out without the direct reclaim. So
there is a functional change. Adnrew, could you drop the sentence about
no functional change and replace it by the following?

"
As a result COMPACT_DEFERRED handling for THP in __alloc_pages_slowpath
will be more precise and we would bail out rather than reclaim.
"

> Also there's no mention of COMPACT_INACTIVE in the changelog (which indeed
> isn't functional change, but might surprise somebody).
> 
> >Signed-off-by: Michal Hocko <mhocko@xxxxxxxx>
> 
> Patch itself is OK.
> 
> Acked-by: Vlastimil Babka <vbabka@xxxxxxx>

Thanks!

-- 
Michal Hocko
SUSE Labs