Re: [PATCH -mm] mm: have order>0 compaction start off where it left

[Rik van Riel <riel@xxxxxxxxxx>]

On 06/28/2012 01:16 PM, Jim Schutt wrote:
> On 06/27/2012 09:37 PM, Rik van Riel wrote:
> > Order> 0 compaction stops when enough free pages of the correct
> > page order have been coalesced. When doing subsequent higher order
> > allocations, it is possible for compaction to be invoked many times.
> >
> > However, the compaction code always starts out looking for things to
> > compact at the start of the zone, and for free pages to compact things
> > to at the end of the zone.
> >
> > This can cause quadratic behaviour, with isolate_freepages starting
> > at the end of the zone each time, even though previous invocations
> > of the compaction code already filled up all free memory on that end
> > of the zone.
> >
> > This can cause isolate_freepages to take enormous amounts of CPU
> > with certain workloads on larger memory systems.
> >
> > The obvious solution is to have isolate_freepages remember where
> > it left off last time, and continue at that point the next time
> > it gets invoked for an order> 0 compaction. This could cause
> > compaction to fail if cc->free_pfn and cc->migrate_pfn are close
> > together initially, in that case we restart from the end of the
> > zone and try once more.
> >
> > Forced full (order == -1) compactions are left alone.
> >
> > Reported-by: Jim Schutt<jaschut@xxxxxxxxxx>
> > Signed-off-by: Rik van Riel<riel@xxxxxxxxxx>
>
> Tested-by: Jim Schutt<jaschut@xxxxxxxxxx>
>
> Please let me know if you further refine this patch
> and would like me to test it with my workload.

Mel pointed out a serious problem with the way wrapping
cc->free_pfn back to the top of the zone is handled.

I will send you a new patch once I have a fix for that.

> So far I've run a total of ~20 TB of data over fifty minutes
> or so through 12 machines running this patch; no hint of
> trouble, great performance.
>
> Without this patch I would typically start having trouble
> after just a few minutes of this load.

Good to hear that!

Thank you for testing last night's version.

--
All rights reversed
--
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/