Re: [patch for-5.3 0/4] revert immediate fallback to remote hugepages

[David Rientjes]

On Tue, 29 Oct 2019, Andrew Morton wrote:

> On Tue, 29 Oct 2019 16:15:49 +0100 Michal Hocko <mhocko@xxxxxxxxxx> wrote:
> 
> > > 
> > > 1. local node only THP allocation with no reclaim, just compaction.
> > > 2. for madvised VMA's or when synchronous compaction is enabled always - THP
> > >    allocation from any node with effort determined by global defrag setting
> > >    and VMA madvise
> > > 3. fallback to base pages on any node
> > > 
> > > Signed-off-by: Vlastimil Babka <vbabka@xxxxxxx>
> > 
> > I've given this a try and here are the results of my previous testcase
> > (memory full of page cache).
> 
> Thanks, I'll queue this for some more testing.  At some point we should
> decide on a suitable set of Fixes: tags and a backporting strategy, if any?
> 

I'd strongly suggest that Andrea test this patch out on his workload on 
hosts where all nodes are low on memory because based on my understanding 
of his reported issue this would result in swap storms reemerging but 
worse this time because they wouldn't be constrained only locally.  (This 
patch causes us to no longer circumvent excessive reclaim when using 
MADV_HUGEPAGE.)