On 19.02.2014 [18:32:59 +0100], Michal Hocko wrote:
> On Wed 19-02-14 09:16:28, Nishanth Aravamudan wrote:
> > On 19.02.2014 [18:03:03 +0100], Michal Hocko wrote:
> > > We had a report about strange OOM killer strikes on a PPC machine
> > > although there was a lot of swap free and a tons of anonymous memory
> > > which could be swapped out. In the end it turned out that the OOM was
> > > a side effect of zone reclaim which wasn't doesn't unmap and swapp out
> > > and so the system was pushed to the OOM. Although this sounds like a bug
> > > somewhere in the kswapd vs. zone reclaim vs. direct reclaim interaction
> > > numactl on the said hardware suggests that the zone reclaim should
> > > have been set in the first place:
> > > node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
> > > node 0 size: 0 MB
> > > node 0 free: 0 MB
> > > node 2 cpus:
> > > node 2 size: 7168 MB
> > > node 2 free: 6019 MB
> > > node distances:
> > > node 0 2
> > > 0: 10 40
> > > 2: 40 10
> > >
> > > So all the CPUs are associated with Node0 which doesn't have any memory
> > > while Node2 contains all the available memory. Node distances cause an
> > > automatic zone_reclaim_mode enabling.
> > >
> > > Zone reclaim is intended to keep the allocations local but this doesn't
> > > make any sense on the memory less nodes. So let's exlcude such nodes
> > > for init_zone_allows_reclaim which evaluates zone reclaim behavior and
> > > suitable reclaim_nodes.
> > >
> > > Signed-off-by: Michal Hocko <mhocko@xxxxxxx>
> > > ---
> > > I haven't got to testing this so I am sending this as an RFC for now.
> > > But does this look reasonable?
> > >
> > > mm/page_alloc.c | 5 +++--
> > > 1 file changed, 3 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > > index 3e953f07edb0..4a44bdc7a8cf 100644
> > > --- a/mm/page_alloc.c
> > > +++ b/mm/page_alloc.c
> > > @@ -1855,7 +1855,7 @@ static void __paginginit init_zone_allows_reclaim(int nid)
> > > {
> > > int i;
> > >
> > > - for_each_online_node(i)
> > > + for_each_node_state(i, N_HIGH_MEMORY)
> > > if (node_distance(nid, i) <= RECLAIM_DISTANCE)
> > > node_set(i, NODE_DATA(nid)->reclaim_nodes);
> > > else
> > > @@ -4901,7 +4901,8 @@ void __paginginit free_area_init_node(int nid, unsigned long *zones_size,
> > >
> > > pgdat->node_id = nid;
> > > pgdat->node_start_pfn = node_start_pfn;
> > > - init_zone_allows_reclaim(nid);
> > > + if (node_state(nid, N_HIGH_MEMORY))
> > > + init_zone_allows_reclaim(nid);
> >
> > I don't think this will work, because what sets N_HIGH_MEMORY (and
> > shouldn't it be N_MEMORY?)
>
> This should be the same thing AFAIU.
I don't think they are guaranteed to be? And, in any case, semantically,
we care if a node has MEMORY, not if it has HIGH_MEMORY?
> > is check_for_memory() (free_area_init_nodes() for N_MEMORY), which is
> > run *after* init_zone_allows_reclaim().
>
> early_calculate_totalpages sets the memory state before we get here.
Ah, I did not see this, thanks! But that does only
node_set_state(nid, N_MEMORY);
which to me means we should only rely on that being set in the
afore-mentioned loop.
> > Further, the for_each_node_state() loop doesn't make sense at this
> > point, becuase we are actually setting up the nids as we go. So node
> > 0, will only see node 0 in the N_HIGH_MEMORY mask (if any). Node 1,
> > will only see nodes 0 and 1, etc.
>
> I am not sure I understand what you are saying here. Why would se
> consider distance to a memoryless node in the first place.
> for_each_node_state just makes sure we are comparing only to a node
> which has some memory.
I apologize, I missed the call to early_calculate_totalpages(), so I was
simply saying that looping over the N_MEMORY/N_HIGH_MEMORY mask wouldn't
necessarily be right if we are setting up that mask on a node-by-node
basis. But early_calculate_totalpages ensures the N_MEMORY iteration
will be fine.
>
> > I'm working on testing a patch that reorders some of this in hopefully a
> > safe way.
>
> Although it might make some sense to reorganize the code (it's a mess if
> you ask me), but I do not think it is necessary.
Agreed.
Thanks,
Nish
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>