Re: [PATCH 14/14] mm, oom, compaction: prevent from should_compact_retry looping for ever for costly orders

[Michal Hocko]

On Thu 28-04-16 10:59:22, Vlastimil Babka wrote:
> On 04/20/2016 09:47 PM, Michal Hocko wrote:
> >From: Michal Hocko <mhocko@xxxxxxxx>
> >
> >"mm: consider compaction feedback also for costly allocation" has
> >removed the upper bound for the reclaim/compaction retries based on the
> >number of reclaimed pages for costly orders. While this is desirable
> >the patch did miss a mis interaction between reclaim, compaction and the
> >retry logic.
> 
> Hmm perhaps reversing the order of patches 13 and 14 would be a bit safer
> wrt future bisections then? Add compaction_zonelist_suitable() first with
> the reasoning, and then immediately use it in the other patch.

Hmm, I do not think the risk is high. This would require the allocate
GFP_REPEAT large orders to the last drop which is not usual. I found the
ordering more logical to argue about because this patch will be mostly
noop for costly orders without 13 and !costly allocations retry
endlessly anyway. So I would prefer this ordering even though there is
a window where an extreme load can lockup. I do not expect people
shooting their head during bisection.

[...]
> >
> >[vbabka@xxxxxxx: fix classzone_idx vs. high_zoneidx usage in
> >compaction_zonelist_suitable]
> >Signed-off-by: Michal Hocko <mhocko@xxxxxxxx>
> 
> Acked-by: Vlastimil Babka <vbabka@xxxxxxx>

Thanks!

-- 
Michal Hocko
SUSE Labs