Re: [PATCH 0/5] Candidate fix for increased number of GFP_ATOMICfailures V2

[Mel Gorman]

On Mon, Oct 26, 2009 at 06:37:36PM +0100, Tobias Oetiker wrote:
> Hi Mel,
> 
> I have no done additional tests ... and can report the following
> 
> Thursday Mel Gorman wrote:
> 
> >   1/5 page allocator: Always wake kswapd when restarting an allocation attempt after direct reclaim failed
> >   2/5 page allocator: Do not allow interrupts to use ALLOC_HARDER
> >
> >
> > 	These patches correct problems introduced by me during the 2.6.31-rc1
> > 	merge window. The patches were not meant to introduce any functional
> > 	changes but two were missed.
> >
> > 	If your problem goes away with just these two patches applied,
> > 	please tell me.
> 
> 1+2 do not help
> 
> > Test 3: If you are getting allocation failures, try with the following patch
> >
> >   3/5 vmscan: Force kswapd to take notice faster when high-order watermarks are being hit
> >
> > 	This is a functional change that causes kswapd to notice sooner
> > 	when high-order watermarks have been hit. There have been a number
> > 	of changes in page reclaim since 2.6.30 that might have delayed
> > 	when kswapd kicks in for higher orders
> >
> > 	If your problem goes away with these three patches applied, please
> > 	tell me
> 
> 1+2+3 do not help either
> 
> > Test 4: If you are still getting failures, apply the following
> >   4/5 page allocator: Pre-emptively wake kswapd when high-order watermarks are hit
> >
> > 	This patch is very heavy handed and pre-emptively kicks kswapd when
> > 	watermarks are hit. It should only be necessary if there has been
> > 	significant changes in the timing and density of page allocations
> > 	from an unknown source. Tobias, this patch is largely aimed at you.
> > 	You reported that with patches 3+4 applied that your problems went
> > 	away. I need to know if patch 3 on its own is enough or if both
> > 	are required
> >
> > 	If your problem goes away with these four patches applied, please
> > 	tell me
> 
> 3 allone does not help
> 3+4 does ...
> 

This is a bit surprising.....

Tell me, do you have an Intel IO-MMU on your system by any chance?  It should
be mentioned in either dmesg or lspci -v (please send the full output of
both). If you do have one of these things, I notice they abuse PF_MEMALLOC
which would explain why this patch makes a difference to your testing.

Thanks

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab
--
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/